Is the Copyright Act Inconsistent with the Law of Employee Invention Assignment Contracts?

Is the Copyright Act Inconsistent with the Law of Employee  Invention Assignment Contracts?

Charles Tait Graves*

Download a PDF version of this article here

There is a latent conflict between the law of employee invention assignment contracts and the Copyright Act’s work for hire doctrine. Countless employees sign contracts specifying that, in most cases, the employer will own trade secrets and patentable inventions, as well as copyrightable works. When employees create in the workplace, these rules are largely uncontroversial. But when employees create something outside the workplace for a new venture, there can be a conflict between these two areas of intellectual property law. The work for hire doctrine is more favorable to employee-ownership than the law of invention assignment contracts. As a perhaps surprising result, where an employee’s outside-the-workplace creation might constitute both a trade secret and a copyrightable work, these two ownership tests can point in opposite directions. Further, when an employee prevails as to copyright ownership, there are good reasons why that result precludes an employer’s conflicting claim to trade secret ownership in the same work. This friction on the boundaries of two areas of intellectual property law has important policy ramifications for employees who create intellectual property on the side, while planning for their next job. **



This article considers the conflict between two commonplace but little-discussed areas of intellectual property law: the law of invention assignment contracts and the law of works for hire under the Copyright Act.

Both can encompass the same subject matter. Both offer tests to allocate ownership between an employer and an employee when the employee creates something. However, sometimes these tests are in tension, if not outright conflict. This conflict is particularly apparent in the context of employee mobility disputes; i.e. those that arise when someone creates intellectual property on the side, outside the workplace, before quitting and leaving for his or her next opportunity.

Employees planning a new venture face a tangle of potential tort, contract, and statutory claims brought by employers seeking to halt competition or to claim ownership of the departing employee’s ideas. Two such claims center on whether the former employer can claim a property right in ideas or work product that a departing employee created offsite, on his or her own time. This essay is the first to compare these related areas of law, tease out the potential conflicts in their tests for employee ownership, and identify the reasons one offers better protection than the other for departing employees.

The law of invention assignment contracts and the work for hire law under the Copyright Act provide the basis for two claims employers can bring to seek ownership of intellectual property an employee has created, whether at the workplace or outside the workplace. Both areas of law start with contract terms that purport to govern whether an employer or an employee owns intellectual property that an employee creates. One applies to trade secrets and patentable inventions. The other applies to copyrightable works of expression. The former, the law of invention assignment agreements, is largely a matter of state contract and state statutory law. The latter, the work for hire doctrine, is largely a matter of federal copyright law. Through these extra-contractual sources, both provide rules that allocate ownership of employee work product, including creative work done entirely apart from the employer’s business. Thus, while many may believe that ownership of employee-created work product is a straightforward matter of contract, that is not always the case when we focus on creative materials an employee has prepared outside the company, for purposes outside the company.

There is a marked contrast between these dueling areas of law. Invention assignment law is highly one-sided, favoring the employer, while work for hire case law is more balanced, and employees often prevail in establishing ownership of their copyrighted creations. This difference arises because the Copyright Act focuses on what an employee was hired to do and what he or she intended by creating the work at issue. The law of invention assignment contracts, by contrast, treats the employee’s role and mental state as irrelevant and instead asks whether the created work is related in some manner to the employer’s business – a broad definition the boundaries of which remain unclear.

Increasing this tension, the subject matter of these doctrines can overlap, which means that the law can dictate different ownership outcomes for the same employee work product. For example, whenever an employee creates a work of expression that is simultaneously a potential trade secret (such as software code), the law of invention assignment contracts and the law of works for hire are both in play. Ownership results may differ depending on the set of tests used to adjudicate a particular dispute. This article will analyze cases where the outcome might have differed if the court and litigants had analyzed a different body of law.

Where there is a conflict, copyright ownership appears to override a conflicting ownership claim to trade secret rights in the same work product. Specifically, where an employee prevails and a copyrightable work is not the employer’s work for hire, the employer likely cannot win by asserting trade secret rights. Because a copyright owner can disclose and publish the work at will, trade secret rights – which require secrecy – would be precluded from the outset. Thus, from a departing employee’s perspective, establishing copyright ownership may be paramount. That may well be better public policy given the potential for overreach in the law of invention assignment agreements.

To parse these areas of law which could govern ownership rights in the same subject matter, this article will explore both areas of law in depth, with a survey of virtually all of the available case law on ownership disputes between employers and employees. Providing this survey may help address issues that largely remain unexplored by courts and commentators, even those who are sympathetic to the obstacles departing employees can face.[1]

Section I explains that when employees create outside-the-workplace material for a future new venture, two categories of intellectual property law are relevant: the law of invention assignment contracts and work for hire law under the Copyright Act. This article will focus on employees acting in good faith to create for a potential new company, in order to set the stage for the latent clash between these doctrines.

Section II provides a historical overview of the development of invention assignment law. This runs from early common law rules and their important exceptions, to the modern era, where the combination of state invention assignment statutes and commonplace contract terms have substantially modified the common law. Setting up the conflict between the law of invention assignment contracts and work for hire law, this section also demonstrates how invention assignment case law significantly favors employers over employees.

Section III sets forth a contrasting body of law which also addresses employees’ creation of intellectual property, with sometimes different outcomes for disputes between employers and employee: work for hire law and its three-part scope of employment analysis. Gathering virtually every reported case, this section shows that work for hire law is more balanced than invention assignment law in weighing the interests of employers and employees.

With this divergence in mind, Section IV explains how both of these categories of intellectual property law can overlap in the same employee-created material. This is where the conflict between these doctrines arises – where applying one test instead of the other could lead to a different outcome in a dispute over whether the employer or the employee owns what the employee has created. This section also provides a tentative solution. Where the employee owns the copyright in a work under the work for hire doctrine, and therefore has a right to publish the work, the employer’s competing invention assignment claim should fail, because it is based on trade secrecy. As the Fourth Circuit held in a 1994 decision,[2] a decision which may provide the key to resolving the latent conflict between these areas of law, a competing trade secret claim cannot exist where a copyright-owner has the right to publish (and thus to disclose) the work at issue.

Finally, Section V examines the public policy implications of the clash between invention assignment law and work for hire law. In the context where an employee operating in good faith creates material outside the workplace that has only an attenuated connection to the employer’s business, the more balanced regime seen under the work for hire doctrine provides better protection for employees forming new ventures and thus better promotes innovation. Asserting the work for hire doctrine where copyrightable material is at stake may prevent overreaching by employers where their ownership claims are weakest.

Employees around the country are subject to contracts which grant their employers ownership in workplace creations – and sometimes also outside-the-workplace creations. Given the importance of these questions to mobile employees and those who wish to form new ventures, this essay looks to ask new questions about this corner of the law.

I. Setting the Stage: Pre-Departure Creative Activities

This essay focuses on a narrow period of time and the intellectual property laws that govern that moment. To set the stage for highlighting a conflict between two areas of intellectual property law, we focus on the point when an employee begins to think about changing jobs, perhaps starting his or her own new venture, or when an employee simply creates work on the side, for purposes not yet known. We consider the situation in which the employee does not misuse the employer’s trade secrets, the employee works on the proverbial nights and weekends outside the office, and the employee is not attempting to create material that is related to the employer’s business.[3]

Many people who would like to quit and start a new venture understandably do not want to resign until a concrete plan is more or less in place. Sometimes that means weeks or months of planning with other nascent co-founders or slowly seeking funding. If the employee creates no intellectual property before leaving, the only legal questions at stake are those of fiduciary duty, the duty of loyalty, and making preparations for the next job.[4]

Sometimes, however, pre-departure planning also means that one or more employees creates new ideas, and commits those ideas to writing or to software code. Assuming the ideas or expression are original, novel, or not generally known in the relevant industry, he or she has probably created some form of intellectual property. So the question arises: who owns that intellectual property – the current employer or the employee who created it?

Events such as these are hard to capture for analysis. They occur in secret, because the employee fears that discovery will result in immediate termination.[5] Further, where attorneys become involved in counseling departing employees, the attorney-client privilege means that practitioners are not free to discuss particular situations in a public setting. Even where the employer later discovers the pre-departure preparations and sends a cease-and-desist letter, many potential lawsuits are averted. Probably only a small fraction of such cases result in lawsuits, and even fewer court rulings are available in public databases.

To delve into the ownership disputes that may arise in these contexts, we will put aside disputes where departing employees misuse employer trade secrets, or create work product that overtly relates to (or even competes directly with) the business of the employer. Our focus will be situations where the employee intends to launch a different, non-overlapping business, and creates new intellectual property with that end in mind. That is, we will assume the departing employee’s good faith, and analyze who might own such work product when both the Copyright Act and the law of invention assignment contracts could encompass the work at issue. We begin with invention assignment law. [6]

II. Invention Assignment Contracts and Employee-Employer Conflicts

In order to understand how invention assignment law and work for hire law are sometimes in conflict, we will first examine the historical development of invention assignment law and then turn to the body of case law, which tends to significantly favor the employer over the employee.

Invention assignment agreements are commonplace in employment contracts. They reverse the historical common law presumption that an employee owns what he or she invents or develops, and they broaden a longstanding common law exception, which allows for employer ownership when the employee was hired to invent the type of intellectual property he or she then created.[7] Terms in invention assignment contracts address ownership of trade secrets and patentable inventions, but not copyrights.[8] Although these agreements allow for employee ownership in certain contexts, by and large they substantially favor employers.[9]

While it is hard to find a definitive source pinpointing when employers began using written invention assignment contracts to alter default common law ownership rules, the practice certainly dates back many decades. In one 1938 California case, for example, a court stated that “[a]n employee’s agreement in the contract of employment to assign patents to his employer is specifically enforceable as to patents, clearly within its terms, as strictly construed against the employer.”[10]

A. State Invention Assignment Statutes and Default Common Law Rules

In most states, invention assignment contracts are governed by common law, though there are not many reported cases in jurisdictions outside of California. However, eight states, beginning with Minnesota in 1977, have enacted similar statutes that seek to lay out boundaries for when employers can obtain ownership of employee creations, all of which favor the employer.[11]

California’s version is illustrative.[12] While awkwardly phrased, it provides six different ways in which the employer can own the employee’s creation: (1) use of the employer’s trade secrets; (2) use of the employer’s time; (3) use of the employer’s resources; (4) the creation is related to the employer’s current business; (5) the creation is related to the employer’s “demonstrably anticipated” research and development; and (6) the creation results from the employee’s work for the employer.[13] These possibilities are disjunctive, meaning that the employer need only demonstrate one of them to prevail.

State statutes regulating invention assignment agreements are not entirely identical. For example, California’s version requires that employers provide notice to the employee, and also allows employers to require that employees disclose all new creations – even if outside the scope of contractual ownership terms – for an ownership review “in confidence.”[14] Washington does not require that the employer treat employee disclosures in confidence, but gives the employee a right to disclose the creation to “the department of employment security.”[15] Both California and Washington put the burden of proof on the employee to show that the creation is outside the scope of employer ownership.[16]

A ninth state, Nevada, enacted a statute relating to invention assignment agreements later than the states discussed above, in 2001, with a different aim.[17] In what was apparently seen as a bid to improve the business environment for technology companies in the state, the statute renders an invention assignment contract unnecessary and broadens the scope of employer ownership compared to the common law. It states: “Except as otherwise provided by express written agreement, an employer is the sole owner of any patentable invention or trade secret developed by his or her employee during the course and scope of the employment that relates directly to work performed during the course and scope of the employment.”[18] As one commentator noted at the time, this sort of one-sided regime is unlikely to foster the birth of new technology ventures in Nevada.[19]

State law invention assignment agreements have been the subject of a fair amount of commentary, albeit mostly limited to the question of whether the default rules granting employers ownership of workplace inventions fairly compensate employee-inventors in a way that best promotes innovation equitably.[20] Some commentators have addressed other issues about the reach of invention assignment agreements which provide helpful background for the questions discussed in this article.[21]

Contracts between employers and employees dictating ownership of workplace creations arose because the default common law rule provided a narrower path to employer ownership. Specifically, case law dating to the nineteenth century provides that an employee owns his or her inventions, even when it is conceived during the term of employment,[22] subject to an important but narrow exception. Those who were “hired to invent” the same type of work product they then duly created owed an automatic assignment duty to the employer, vesting title in the company. These rules developed from early Supreme Court decisions.[23] In turn, the common law provided that if an employee who was not hired to invent nonetheless created an invention using the employer’s time or resources, the employee retained ownership, but the employer received a permanent, royalty-free license known as a “shop right” to practice the invention. Historical case law on the hired-to-invent exception and the shop right rule – that is, before the age of ubiquitous written invention assignment agreements and state statutes governing them – were often fact-intensive.[24] Even in more recent cases, these common law rules still apply in situations where, for whatever reason, there was no written contract.[25]

B. How Invention Assignment Contracts Favor the Employer

By and large, invention assignment contracts and related state statutes favor the employer over the employee, even though the text of some such statutes appears, at first glance, to be designed to protect employees. Generally speaking, an effective invention assignment to the employer occurs automatically, with no further steps required, and applies to inventions that the employee has not yet conceived.[26] In some jurisdictions, the contract can be effective even if the employer requires the employee to sign it after the employee has commenced employment.[27]

Expansive invention assignment statutes and agreements pose risks for employees who create outside-the-workplace developments, often with an eye towards forming a new company after resigning. Even if we assume good faith, with no use of employer trade secrets, time, or equipment, the employee faces many risks, and lines are not clearly drawn.

Part of the complication stems from changes in the workplace. When state invention assignment statutes were enacted in the 1970s and 1980s, perhaps it was easier to imagine a bright-line separation between the workplace and “nights and weekends” in an employee’s garage. Without the internet, employees typically did not work from home, much less on employee-owned phones and laptops. Moreover, it may then have been easier to determine with certainty what “related to” the employer’s business and what did not. Fewer companies were large multinational entities, with unrelated departments and operations of which an employee might be unaware. Though the scarce case law is hardly determinative on these questions, changes in modes of employment and the scope of companies’ operations may work to the detriment of employees who create new material on the side.

For similar reasons, employers may be able to argue that working hours – and thus what constitutes the employer’s time – are elastic, even 24/7, if employees typically answer email at home or on weekends. In one case that focused primarily on fiduciary duty but also involved an invention assignment contract, a court which issued a preliminary injunction was amenable to the notion that employer time meant more than ordinary working hours where employees often put in long days.[28] It remains to be seen how far this will go, however, as it seems that no court has yet analyzed whether a statute which appears to divide employer time and off-hours could be rendered meaningless if the employer is permitted to define its time as “24/7/365.”

A similar interpretive issue arises with the definition of the employer’s resources. While employees of technology companies in 1980 or 1990 used office workstations too unwieldy to take home, employees today commonly use company-issued laptops and mobile devices. Confusing the issues further, many companies permit dual use. For example, employees may use company-issued devices for personal use (or vice-versa), subject to codes of ethics or other workplace guidelines. If an employee creates an unrelated invention for a future venture on a weekend, but uses a dual use company-issued laptop to do so, can the employer claim ownership? If an employee creates a new idea for a future venture on a dual-use company-owned mobile phone, can the employer claim ownership? The answer again is unclear, as courts do not appear to have yet analyzed this question.[29]

The ambiguity of terms that appear in invention assignment statutes and contracts – such as terms that vest ownership in the employer when the employee’s creation relates to the employer’s business or research or when it results from the employee’s work – also leads to interpretive difficulties.[30] These terms notably alter the common law, under which an employee must be “hired to invent” in order for the employer to own the creation. Under these terms, even the receptionist would be subject to employer ownership if his or her outside-the-workplace creations could be said to relate in some fashion to the employer’s business.

More importantly, it remains unclear what phrases like “relates to” or “results from” mean in the everyday workplace. For example, what happens if an employee in good faith creates something offsite, intending to use it for a future business, but unbeknownst to the employee the company has been dabbling in something similar at a different office on the other side of the country? In other words, are the rules to be interpreted in a manner that takes the employee’s notice of the scope of the employer’s business into account? Does an employee’s good faith intent matter? Thus far, no reported case has addressed this question.

The closest may be a 2007 decision from the Northern District of California which ruled in favor of the employer where an employee argued that he should own his invention because his own department’s business did not encompass the work. The court, however, found that California’s invention assignment statute did not contain terms limiting it to the particular department in which the employee works.[31] That said, it appears that the employee was on notice that his or her invention related to the employer’s business,[32] and thus the question of whether the statute is to be interpreted to reflect the employee’s notice or state of mind is not at issue.

As explained below, the work for hire rules do explicitly consider the employee’s state of mind, setting up a potential conflict with the law of invention assignment contracts on these issues.

Further complicating the problems with the interpretation of invention assignment contracts, some companies define themselves in employment agreements to include their affiliates; different entities that are separately incorporated and separately managed. It is unclear whether an invention assignment statute or agreement could divest an employee of an outside-the-workplace invention in a situation where the employer contends that the invention relates only to the business of a differently-incorporated affiliate entity.

The expansive wording of invention assignment contracts and statutes is not the only difficulty facing employees who create offsite work product. Some courts have held that even where an employee notifies the employer about a development and the employer verbally disclaims any interest in it, the employer can later retract that apparent waiver. The reasoning is that since the invention assignment rules automatically transfer ownership at the moment of creation, if the employer owns the invention, there is nothing that can be verbally bargained over.[33] Because this sort of thinking is unlikely to occur to non-lawyers, this can pose an unfavorable trap for employees, to say the least.

Taking the concept of employer ownership to its extreme, some employers attempt to obtain ownership of developments created by former employees entirely after the employee has left the job. Known as “holdover” or “trailer” clauses, these terms typically assert ownership of inventions for a set period after the employee’s departure, or purport to establish a presumption of ownership. Some courts have rejected such clauses, at least where the former employee has not misappropriated the former employer’s trade secrets, or has narrowed their potential applicability.[34] But in 2012, the South Carolina Supreme Court found a broad holdover clause enforceable under the law of that state.[35] The court’s reasoning was similar to those of courts which enforce post-employment non-competition covenants. It also failed to analyze the case of former employees working in good faith to launch a new venture. The threat posed by holdover clauses to employees who resign and create new intellectual property for the first time after leaving is obvious.[36]

At the same time, courts have generally rejected employment contracts which purport to require employees to assign inventions created before starting the new job.[37] However, the employee’s incorporation of pre-existing work product into intellectual property created during the period of employment generally will not prevent assignment of that new work to the employer, assuming the conditions for such assignment exist.[38]

Finally, if an employee signs two invention assignment contracts, the first-in-time has priority. This means that if an employee has signed a valid invention assignment contract with a current employer, he or she cannot avoid that agreement by signing a new, second-in-time invention assignment contract in favor of a new start-up venture. Thus, creating an outside-the-workplace invention and assigning it under a new agreement to a new venture would not transfer ownership to the new venture if the prior first-in-time assignment agreement applies to vest the employer with ownership.[39]

C. The Case Law on Invention Assignment Contracts is One-Sided

Overall, these rules tend to favor the employer where an employee creates an outside-the-workplace invention. Where exactly an invention stops “relating to” or “resulting from” the company’s business or the employee’s work is not clear. That ambiguity empowers employers over employees. Adding further ambiguity about what constitutes use of the employer’s time or equipment, and the open question whether the employee’s personal notice of the employer’s activities matter, the statutes pose a minefield for offsite creations. Oral permissions may not suffice, undermining above-board efforts to be transparent. And in the most restrictive jurisdictions, former employers may even be able to obtain ownership of intellectual property conceived and developed entirely after the employee has quit and left.

Given the reach of state statutes and the manner in which these agreements override the common law “hired to invent” exception to broaden the scope of employer ownership, it is not surprising that the available case law strongly favors employers in disputes with employees, and in fact is nearly one-sided. Employees rarely prevail.[40]

Even a rare exception – a remand ruling in favor of a departing employee – rests on dubious conclusions. In the well-publicized Mattel v. MGA battle over ownership of doll designs where a departing employee had ideas and did at least some preliminary work before resigning, the Ninth Circuit found fault with the employer’s position because its invention assignment contract lacked the word “ideas” in its recitation of what the contract covered.[41] That ruling comes across as the product of incomplete analysis given that the contract seemingly included synonyms (such as “developments”) and because the court did not analyze the terms of California’s invention assignment statute. [42]

All told, and excluding cases about post-employment “holdover” clauses, there are many cases where the employer (whether a private company or a university) prevailed in an invention assignment contract dispute with an employee.[43] By contrast, only a few have found in favor of an employee.[44]

But as discussed below, a different regime governs when employees create copyrightable works – with different considerations, different outcomes, and potential direct conflicts with the law of invention assignment agreements.

III. The Work for Hire Doctrine and Employee-Employer Conflicts

A. Work for Hire Law and the Restatement of Agency

This article proposes that where an employee creates copyrightable material, the test for allocating ownership of that material differs from the invention assignment tests for allocating ownership of trade secrets and patentable inventions described above. We start with an analysis of modern work for hire law, developed following a major Supreme Court decision in 1989, Community for Creative Non-Violence v. Reid.[45]

The law of employee invention assignment contracts is pervasive, and encompasses most of the creative work product employees may generate on the job – whether trade secrets or inventions – that result in issued patents. However, it does not cover everything, and thus a separate, but seemingly similar body of law fills the gap. For employee creations that are expressive – and that meet the standards for copyright protection – the work for hire doctrine allocates ownership between employers and employees, just as the interplay of contract wording and (in some states) legislative enactments allocates ownership of potential trade secrets and patentable inventions.

The work for hire doctrine allocates ownership of copyrightable works in a manner that is more favorable to the employee, posing a direct conflict with the law of invention assignment contracts where an employee creates something that is potentially both copyrightable and a trade secret.

The work for hire doctrine is partly a matter of statute, partly a matter of contract law, and partly a matter of the common law of agency. Generally speaking, a copyrightable work created by an employee within the scope of his or her employment belongs, at creation, to the employer:

A “work made for hire” is (1) a work prepared by an employee within the scope of his or her employment; or (2) [specific categories of works not applicable to this essay].
In the case of a work made for hire, the employer or other person for whom the work was prepared is considered the author for purposes of this title, and, unless the parties have expressly agreed otherwise in a written instrument signed by them, owns all of the rights comprised in the copyright.

See 17 U.S.C. §§ 101, 201(b) (2012).[46] The work for hire doctrine does not extend to independent contractors, however, and thus a copyrightable work created by an independent contractor belongs to the contractor, absent a contract saying otherwise.[47]

In determining who is an employee and who is not, courts are to apply the federal common law of agency, rather than the employment law of particular states.[48] That question is solved by analyzing a large number of factors.[49] Disputes over whether a person is a contractor or an employee are common, and the case law on that question is extensive. However, where the individual at issue starts as a contractor and later becomes an employee, the question of whether a work for hire exists is further muddled.[50]

For our purposes, these disputes are peripheral. What matters instead is who owns the copyrightable material created by an employee when an employment relationship exists. As the Copyright Act states, when the author of a copyrightable work is clearly an employee of the party claiming ownership under the work for hire doctrine, the sole question is whether the author created the work within the scope of his or her employment:

A “work made for hire” is . . . a work prepared by an employee within the scope of his or her employment.[51]

Much like the law of invention assignment contracts, an employer’s agreement with an employee to assign works for hire is not the end of the story. A penumbra of case law defines what “scope of his or her employment” means, and in effect creates exceptions to the presumption of employer ownership.[52] Thus, while employment contracts ubiquitously recite that works for hire belong to the employer, the assignment is not monolithic.[53]

Every court to address the “scope of employment” issue since the Supreme Court addressed work for hire law in the 1989 Reid decision has applied a three-part test based on section 228 of the Restatement (Second) of Agency: “Conduct of a servant is within the scope of employment if, but only if: (a) it is of the kind he is employed to perform; (b) it occurs substantially within the authorized time and space limits; (c) it is actuated, at least in part, by a purpose to serve the master.”[54] This reference to the Restatement is typically perfunctory, as courts do not cite other sections from the same chapter.[55]

Supposedly, all three tests must be met for there to be a work for hire, meaning that the test is conjunctive, not disjunctive.[56] But at the same time, courts seem to place the least emphasis on the second factor and will find a work for hire if the other two factors are met and at least some of the creative work was done at the employer’s facility and during work hours – or in some cases none at all.[57] Further, like invention assignment law, courts may be open to an expansive interpretation of employer time where a company has a more 24/7 working culture.[58]

Perhaps because the work for hire scope of employment test presents a more balanced allocation of interests compared to the tests under a typical employee invention assignment contract, case law outcomes appear less one-sided. Putting aside inconclusive rulings, I have found around the same number of work for hire cases where the employer prevailed as to ownership of the work (at least as to injunctive relief)[59] as those where the employee prevailed.[60]

Scholarly writing about the work for hire doctrine has mostly centered on a question that precedes the topic of this article – whether the author is an independent contractor or an employee in the first place.[61] That focus is no surprise, as there is a prominent Supreme Court ruling (Reid) on that issue. Moreover, the rise of the so-called “gig economy,” where companies treat workers as independent contractors rather than employees, may also explain why many focus on the rights of contractors versus those who hire them. By contrast, commentary on the scope of employment test has been meager, with some exceptions noted below. The same is true in the leading treatise on copyright law.[62]

That said, several articles provide useful background for the issues analyzed here. A student note comes perhaps closest to the issues presented here, by contrasting copyright and patent rights in employee-created software.[63] Another commentator, in a detailed overview of the allocation of ownership interests in works created by employees, advocates for “an interpretive rule operating in favor of the employee” given that “[e]mployers are likely to attempt broad and vague definitions of the job description” in order to satisfy the first factor of the scope of employment test.[64] Finally, a definitive history of the work for hire concept and the transition to employer ownership of employee work product sheds light on the origins of the conundrum this essay addresses.[65]

B. The Scope of Employment Case Law is More Balanced

The manner in which courts have analyzed the three-part scope of employment test in work for hire disputes is noticeably different from the analysis seen in invention assignment disputes. As described in the cases summarized here, courts tend to focus most on the first and third factors of that test. And in contrast to invention assignment agreement law, employees prevail more often because the three factors are more employee-friendly.

Case Law on Factor One (Hired to Perform): So long as the employee’s job description/job duties are of the same general category as the copyrightable work, employers tend to win on the first factor – even if the work is only incidentally related to the employee’s job duties. The employer’s provision of guidance and supervision to the employee weighs in the employer’s favor as well.[66] At the same time (and even if the work might be incidentally related to the scope of employment) the employer’s disclaimer of the work is highly relevant, in contrast to the law of invention assignment contracts.[67] Conversely, if the employee is hired to do tasks in categories unrelated to the copyrightable work, the employee tends to prevail on the first scope of employment factor.[68]

Case Law on Factor Two (Employer’s Time and Space): Much of the case law on the second scope of employment factor focuses on whether or not the work was done at the employer’s facility, though the test as worded is broader than that inquiry. The focus is where the work was created, not where it was later used.[69] If job duties necessarily include working off-site, the fact that an employee creates the work at home may still favor the employer.[70] However, in many cases, the employee’s work off-site means that the employee wins this factor.[71]

That said, prevailing on the second factor may not favor the employee overall if the work was done during the time the employee was employed. Courts may not find the second factor determinative if the employer has prevailed on the first and third factors.[72]

Case Law on Factor Three (Employee’s Purpose): Sometimes there are cases where the employee did not seek to benefit the employer, beyond any doubt.[73] And sometimes the employee’s copyrightable work is developed in tandem with work done for the employer, so that it cannot reasonably be anything other than a work for hire.[74]

However, most cases are closer. Many involve a situation where an employee created something at home, used it at work but also for other purposes, and was motivated by both personal motives and a desire to help the employer. As noted above, the second factor can be discounted if the other factors are in the employer’s favor. That makes sense, as otherwise an employee could create something at home and automatically get ownership of it, no matter how closely it relates to his or her job role.

So where do courts draw the line when the employee may have had mixed motives? In general, the motivation to assist the employer need only be partial; it need not be the sole motivation. The employee must be at least “appreciably” motivated by a desire to serve the employer’s goals.[75] But what does this mean? Greater than a 50 percent motivation to serve those goals? Something else? The cases do not nail it down with precision. Both the employer[76] and the employee[77] can prevail on this factor.

Some courts, such as the Beasley and Roeslin cases cited below, suggest that looking at the employee’s subjective beliefs is one approach to analyzing the third scope of employment factor. One court made this approach the dispositive test. In Martin v. City of Indianapolis, the court, relying on the Restatement (Second) of Agency section 235, comment (a), stated that “[t]he state of mind of the employee is the material determination; the court may consider the employee’s actions or other manifestations only as evidence of the employee’s state of mind.”[78] Under that sort of analysis, the employee won easily, because he claimed that his sculpture was a personal project, that he did not get paid for it, spent a lot of time on it, worked on it in his free time, signed an agreement calling himself the “owner” of it, and was, apart from his work for the employer, “a serious, independent artist who uses stainless steel as a medium.” Focusing on the employer’s potential economic motivations, the court noted that the work was not created in order for the employer to get a commercial benefit, as it never sold the work to a customer.[79]

Another way to consider the question is to look to whether the employee was doing the work for a third party at the time – as doing the work for someone else is surely relevant to whether the work was done “appreciably” for the employer. In one case, the work in question (a bag design) was of the type the employee was hired to do, and the first factor favored the employer. However, the work was not a work for hire, because the employee created the design for a third party, assigned it to that third party, and was paid directly by the third party. Thus, the design was not “actuated by a purpose to serve [the employer’s] interest.”[80]

Overall, an employer can win on the first factor but still lose the third, and the focus – however a court words it – appears to be whether some significant percentage of the employee’s subjective motive was to help the employer, focusing on the time the work was created.[81] Because the law of invention assignment contracts does not expressly consider the employee’s intent when creating intellectual property, this consideration of the employee’s state of mind is quite different from the analyses seen under that body of law.

IV. Is Work for Hire Law Inconsistent with Invention Assignment Law?

This review of the work for hire doctrine and the law of invention assignment agreements shows why there is seemingly a direct conflict between them. Where an employee creates something that is both copyrightable and that could be an employer-assigned trade secret, the outcome of an ownership dispute could differ.

Unlike trade secrets and patent rights, trade secrets and copyrightable works are categories of intellectual property that can exist concurrently in the same work. Indeed, because the Copyright Act does not preempt state trade secret law, courts deem that confidentiality requirements sufficiently distinguish a state law trade secret cause of action from a copyright infringement cause of action. A plaintiff with ownership rights under both can theoretically pursue both causes of action against a defendant regarding the same work.[82]

Thus, it follows that when an employee creates new work, that work might constitute the employer’s trade secret – if it meets the conditions for automatic invention assignment under the employment agreement – and it also might constitute the employer’s work for hire – again, if the conditions for the scope of employment test are satisfied. In many cases, especially for routine situations where the employee creates work for the employer in the ordinary course of business, both tests will easily be satisfied, and nobody would imagine otherwise.[83]

However, on the margins, where employees create work outside the workplace and for outside purposes, the two tests can dictate different results as to the same work by the same employee. To begin with, the invention assignment analysis is disjunctive – the employer need only satisfy one of the conditions for ownership, such as demonstrating that its equipment was used. By contrast, the work for hire analysis is conjunctive – the employer must show all three factors, or at least such a strong showing on the first and third that the second is discounted.

Second, the invention assignment analysis is agnostic to an employee’s job role. If the receptionist writes software that relates to the employer’s business, and if the receptionist signed an invention assignment contract, the employer owns the work even if the receptionist programmed it during off-hours and away from the office. The invention assignment contract thus reaches beyond the common law “hired to invent” test for employee ownership, as discussed above. Again, by contrast, the work for hire scope of employment test looks to the employee’s job role as the first factor of the analysis, thereby providing a narrower path to employer ownership.

Perhaps most importantly, the invention assignment analysis is agnostic to the employee’s motive or intent. Even if an employee creates an off-site work in anticipation of a new job or creating a new company, that motive is not relevant if the work is found to relate to the employer’s business or relate to its demonstrably anticipated research and development.[84] But as discussed above, the third factor of the work for hire scope of employment test significantly considers the employee’s motive.

With these differences in mind, there are many decisions where the outcome might have differed depending on which of the two doctrines the court focused on (or perhaps more accurately, the doctrine the employer focused on when bringing the lawsuit).

One example where a conflict might have arisen had the court spent more time on the employer’s invention assignment claim was a 2004 ruling from the Southern District of Iowa, PFS Distribution Co. v. Raduechel. There, the employee was a truck driver for a poultry plant. A self-taught programmer, he created software at home for an order entry system. The employee licensed his software to two unrelated businesses and later created a version for use by his employer, brought it on site and installed it, and tailored it for the employer’s business.[85]

After the company was acquired and cut salaries, the employee and a co-worker planned a mass departure for a new venture, engaging in significant misconduct that led to a preliminary injunction.[86] However, the employer’s claim to ownership of the truck driver’s order entry software faltered because the court agreed that the employee owned the work under the work for hire doctrine. He was not hired to program, he created the work at home and even licensed it to others before bringing it to work, and his purpose in creating it was not to serve the employer.[87]

Raduechel is possibly the only case where a court considered a work for hire dispute alongside a claim for breach of an invention assignment contract. Had the court spent more time analyzing the latter, it might have recognized a potential conflict. After all, the employee’s software surely related to the employer’s business, and at least a version of it was modified for use at the employer’s facility and presumably on its equipment. Under the employer-friendly regime of invention assignment law, perhaps the court might have found that the employer owned at least a portion of the work. However, the court appears to have brushed off the employer’s invention assignment cause of action, having already decided the work for hire issue and quickly concluding that the result was the same under the employer’s contract claim.[88]

Another case illustrating the potential conflict is a 2010 ruling from the Central District of California, TAP Worldwide, LLC v. Becker. There, the employer sold “off-road parts and accessories.” The employee prevailed in a work for hire dispute over ownership of a software program he wrote to expedite the processing of export shipments.[89] He had created the work on his own time, and based on his own “market research,” without using the employer’s equipment.[90] The employer sought ownership and an order requiring the former employee to “restore access” to the software, but lost on all three work for hire scope of employment factors.[91] Would the result have been the same if the employer had sued for breach of an invention assignment agreement? In California, it would need only show that the work related to its business, and not that the employee was hired to create that type of work, or that the employee was motivated by a purpose to serve the master.[92]

Perhaps the case that comes closest to demonstrating the conflict between the work for hire doctrine and invention assignment law is a 1994 Fourth Circuit case where an employee’s ownership of the copyright in a work precluded the employer’s dueling claim to trade secret rights in the same information. In Avtec Systems, Inc. v. Peiffer, the employer provided simulations of orbital patterns to the federal government. The employee, whose job duties included computer simulations for satellite orbits, wrote a software program for that purpose, but did so on his own time. The company used the program as a marketing tool, and other employees suggested modifications. However, the employee later licensed the program to a third party.[93]

After a confusing bench trial ruling where the lower court found that the employee owned the copyright but the employer held a “shop right” and/or a “trade secret” in the same work, the Fourth Circuit remanded for clearer findings on the work for hire question. Crucially, the court rejected the notion that an employee’s copyright ownership could be undermined by an employer’s trade secret claim to the same information. Noting that copyright ownership confers a right of publication and distribution, while trade secret rights depend on non-publication and secrecy, the court held that “Avtec offers no authority, and we have found none, for the proposition that the alleged ‘owner’ of a trade secret . . . could maintain the secrecy of material that is subject under federal law to publication at the will of another.”[94]

Ultimately, after remand, the Fourth Circuit affirmed a ruling that the employee owned the copyright under the work for hire factors, and confirmed that an employee-owner of a copyrighted work cannot be subject to an employer’s trade secret misappropriation claim over that work.[95]

Although the employer brought trade secret and fiduciary duty causes of action against the employee rather than a breach of contract claim relating to invention assignment, that is a distinction without a difference for our purposes, as the intellectual property rights arising from an invention assignment contract would be trade secret rights. Thus, Avtec teaches that, should an employee prevail on copyright ownership under the work for hire analysis, an employer likely could not succeed in a breach of invention assignment claim over trade secret rights in the same work, as the conflicting trade secret claim would be void ab initio.[96]

Given this under-analyzed conflict, an employee’s potential copyright ownership of an outside-the-workplace creation may well be the dispositive question in clashes where an employee prepares work product for an outside opportunity while still employed. Litigants’ choices about what law to focus on during ownership disputes over copyrightable works may be more important than has been recognized thus far.

Conclusion – The Use-Value of the Scope of Employment Test

This essay is not about ownership of ordinary-course creative work performed in the workplace on behalf of an employer. It is instead about marginal cases where an employee, operating in good faith, avoids the employer’s trade secrets, time, and equipment to start preparing for a new job, forming a new company, or simply creating for personal interest. That is the context where employer ownership is most debatable, and that is the context where the clash between the work for hire doctrine and invention assignment agreement law is sharpest.

As we have seen, it is not difficult to imagine a situation where an employee creates something outside the workplace that is both copyrightable and something an employer might claim as its trade secret through contractual assignment. Whereas the scope of employment analysis might provide that the employee was not hired to create the sort of work in question, did not create it with the employer’s time or resources, and did not create it to serve the employer, invention assignment agreement law might find that the work nonetheless can be deemed “related” to the employer’s business in some manner.

Heightening the risk for employees, these rules are not intuitive. The work for hire doctrine is barely a ripple in the ocean of intellectual property law. Reported invention assignment disputes over outside-the-workplace creations that are not clearly related to the employer’s business are rare. Employees may not recognize that an employer may stretch invention assignment law to find ways to argue that the work relates to the employer’s business in some unexpected fashion. And as discussed above, if the employee were unaware of that purported connection between his or her at-home work and some project contemplated by a far-flung, international employer, it is not clear that the employee’s lack of notice would play a role in the analysis. In jurisdictions like South Carolina, invention assignment terms can reach beyond the grave, as it were, to allow past employers to claim ownership in work product created long after an employee’s departure.

This potential for overreach endangers employee mobility and subjects new enterprises to aggressive legal claims by former employers. Many employees may begin creating work product for a new venture before leaving their current job. Some will break the rules in clear-cut ways, but others try to do the right thing, yet still face legal traps for the unwary. Not everyone can afford to quit and then start development work. And not everyone can afford legal advice before quitting.

How should such conflicts be resolved? And how can these issues be resolved in a manner fair to an employee operating in good faith? As to copyrightable work, one answer is that the work for hire rules should control over the trade secret-centered rules of invention assignment agreement law. That is not merely because copyright law is federal law while invention assignment law is largely a matter of state contract and state statutory law.[97] After all, trade secret rights can now exist under federal law as well, and some aspects of invention assignment contracts are also matters of federal law.

Rather, the work for hire rules control because copyright ownership precludes a conflicting claim to trade secret ownership, whether via an invention assignment contract or otherwise. As the Avtec case explained, if an employee owns the copyright – and thus the right to disclose and publish the work – an employer’s conflicting trade secret claim is not possible because there can be no trade secret rights in something the adverse party has the freedom to disclose at will. To put it in the language of invention assignment law, an employee cannot agree to a future conveyance of a trade secret right that would be stillborn at the outset given overriding copyright ownership under the work for hire doctrine’s scope of employment test.

This outcome is the better result for outside-the-workplace creations where the employer’s ownership claim would require an attenuated argument that the employee’s at-home work “relates” to the employer’s business in some strained or unexpected manner – or where the employee had no notice that the employer was thinking of something similar in another office, or with a different engineering team. Where work product is copyrightable, the scope of employment test under the work for hire doctrine could become a tool to prevent overreaching, and thereby to protect departing employees.

To be sure, this is hardly the same as full reform of a confusing regime, as this prospect only applies in cases where the employee creates something copyrightable. People changing jobs face a confusing mix of potential claims, and this is only part of the difficulties they face.[98] Future essays in this series will examine others.

* Partner, Wilson Sonsini Goodrich & Rosati, San Francisco, and Adjunct Faculty, University of California, Hastings College of the Law.

** This essay is the first in a three-part series addressing under-analyzed areas of intellectual property and employee mobility law which impact creative employees when changing jobs. Departing employees can face a tangled body of contract, tort, and statutory claims brought by former employers. Academics and practitioners have provided little commentary about some of these overlooked areas of law. This relative inattention is surprising given the important policy concerns so often at stake in mobility disputes.

[1] One powerful, if rare, example of the sort of thoroughgoing critique that is all too often lacking is found in Orly Lobel, The New Cognitive Property: Human Capital Law and the Reach of Intellectual Property, 93 Tex. L. Rev. 789, 792-93, 798-99, 801-03, 813-15 (2015) (tying the overbreadth of employee invention assignment agreements – noting, among other things, post-employment holdover clauses – to broader concerns over employer overreach and resulting harms to mobility and growth of new enterprises).

[2] Avtec Sys. v. Peiffer, 21 F.3d 568 (4th Cir. 1994).

[3] In general, and while definitions vary slightly from state to state, a trade secret is non-public business information valuable to competitors that the owner has protected with reasonable security measures. E.g., Defend Trade Secrets Act of 2016 § 2(b)(1), Pub. L. No. 114-153, 18 U.S.C. § 1839(3), 130 Stat. 376, 380.

[4] Disputes over these so-called “preparations to compete” will be the subject of the next essay in this series.

[5] Even if preparations to compete are legal, an employer can terminate an employee found to be engaging in such activities. See Fowler v. Varian Assocs., Inc., 241 Cal. Rptr. 539, 544 (Ct. App. 1987) (granting summary judgment for employer where employee brought wrongful termination claim after being fired for making preparations to leave and compete).

[6] Readers may wonder why this essay refers to employers’ trade secret rights through invention assignment agreements, rather than patent rights. After all, and as the very phrase “invention assignment” suggests, patents are what scholars and practitioners likely think of first, if not exclusively, when the subject of invention assignment arises. But this misses an important timing point: when an employee conceives a patentable invention, it first exists as intellectual property in the form of a trade secret, not a patent. If the employer chooses to pursue a patent application, that election comes after the point of initial, automatic assignment. That is why this essay points to a clash between competing copyright and trade secret rights in the same work product.

[7] While ownership of an invention would otherwise vest in the inventor, the standard rule and practice is that an employee can contract ownership away through an invention assignment agreement. See Beech Aircraft Corp. v. EDO Corp., 990 F.2d 1237, 1248 (Fed. Cir. 1993); Access Cardiosystems, Inc. v. Fincke (In re Access Cardiosystems, Inc.), 340 B.R. 127, 147 (Bankr. D. Mass. 2006) (explaining rules). Thus, “an employer owns an employee’s invention if the employee is a party to an express contract to that effect.” Banks v. Unisys Corp., 228 F.3d 1357, 1359 (Fed. Cir. 2000); Grove v. Grove Valve & Regulator Co., 84 Cal. Rptr. 300, 303-04 (Ct. App. 1970) (early case finding that employer owned improvements to various inventions in enforcing terms of employee invention assignment agreement signed in 1959).

[8] Although such contracts assign ownership of rights including those governed by federal law – i.e., patents, but also trade secrets given the 2016 Defend Trade Secrets Act – their interpretation is largely, but not entirely a matter of state law. Until 2008, the interpretation was solely one of state law. See Affymetrix, Inc. v. Illumina, Inc., 446 F. Supp. 2d 292, 296 (D. Del. 2006) (applying California law; “The proper construction of assignment agreements is a matter of state contract law.”) (citing Minco, Inc. v. Combustion Eng’g, Inc., 95 F.3d 1109, 1117 (Fed. Cir. 1996)). This default rule was modified in 2008, when the Federal Circuit ruled that, at least as to the question of whether an assignment contract creates an automatic future conveyance of inventions not yet in existence – which is accomplished by the inclusion of the magic words “hereby assign” – is one of federal law. See DDB Techs., L.L.C. v. MLB Advanced Media, L.P., 517 F.3d 1284, 1289-90 (Fed. Cir. 2008). DDB’s reasoning was based on federal courts’ powers as to “the question of standing in patent cases.” Id. See also Intellectual Ventures I LLC v. Erie. Indem. Co., 850 F.3d 1315, 1320 n.1 (Fed. Cir. 2017) (stating rule). The decision was notably silent on what law would govern the future conveyance of state law trade secret rights in an employee creation.

[9] See discussion and cases infra Section B.

[10] Hercules Glue Co., Ltd. v. Littooy, 76 P.2d 700, 701 (Cal. Dist. Ct. App. 1938) (enforcing employee invention assignment contract in favor of employer: “By the terms of this contract, appellant agreed that he would devote his entire time, knowledge, skill, best efforts, and services to the work of respondent as it might direct; that all patents, processes, or formulas, pertaining to spreader, which he invented, developed or perfected during his original and present employment, should be the property of respondent, and that he would, on demand, execute any assignments, transfers or other instruments necessary to perfect respondent’s title thereto. He further agreed that he would not, during its life or subsequently, divulge to any person, or use for his own benefit, any secret processes or formulas invented, developed, or perfected by him or respondent between October 1, 1924, and the termination of the agreement.”). Other early cases involving written contacts include: Kober v. United States, 170 F.2d 590, 592, 594 (4th Cir. 1948) (finding that an employment agreement with a U.S. government employer providing for “the complete assignment” of inventions made by the employee was a “reasonable agreement entered into for a lawful and proper purpose” and holding that inventions made by an engineer subject to this agreement therefore belonged to the government-employer); Crown Cork & Seal Co., v. Fankhanel, 49 F. Supp. 611, 613, 615 (D. Md. 1943) (construing and finding enforceable an invention assignment agreement that required the employee to assign rights over to the employer for “[a]ll inventions and discoveries which I make while in the employ of said Company, along the lines of its general work, constituting improvements both in its then existing products and methods of manufacture, or otherwise, shall become its exclusive property,” and holding that the employment agreement was valid and enforceable).

[11] See Cal. Lab. Code §§ 2870-2872 (West 1979); Del. Code. Ann. tit. 19, § 805 (1984); 765 Ill. Comp. Stat. 1060 / 2 (1983); Kan. Stat. Ann. 44-130 (West 1986); Minn. Stat. Ann. § 181.78 (West 1977); N.C. Gen. Stat. §§ 66-57.1-.2 (1981); Utah Code. Ann. § 34-39-3 (West 1989); Wash. Rev. Code Ann. §§ 49.44.140, .150 (West 1979). Based on this author’s experience, companies often mimic the structure of these state statutes in employment agreements, even for employees residing in other states. As a result, these eight statutes have great influence beyond their home jurisdictions.

[12] See Cal. Lab. Code §§ 2870-72 (West 1979).

[13] Id. § 2870. The “results from” clause is unclear, but its interpretation has not yet been the subject of a published decision.

[14] Id. § 2871.

[15] Wash. Rev. Code Ann. § 49.44.150 (West 1979).

[16] See id.; Lab. § 2872.

[17] See Nev. Rev. Stat. Ann. § 600.500 (West 2001).

[18] Id.

[19] See Mary LaFrance, Nevada’s Employee Inventions Statute: Novel, Nonobvious, and Patently Wrong, 3 Nev. L.J. 88 (2002) (well-done dissection of the then-new state statute, noting its departure from the common law, contrast with other state statutes, and contrast with the work for hire rules; predicting the statute would not, as apparently intended, encourage the growth of technology companies in Nevada).

[20] There is a cottage industry of notes and law review articles positing that if employees were given greater compensation for, or some ownership stake in, inventions created for the employer in the workplace, innovation would increase. See Shlomit Yanisky Ravid, Rethinking Innovation and Productivity Within the Workplace Amidst Economic Uncertainty, 24 Fordham Intell. Prop. Media & Ent. L.J. 143 (2013) (arguing that employees would have a greater incentive to invent if they retained a greater interest in – and received financial compensation for – workplace inventions); Ann Bartow, Inventors of the World, Unite! A Call for Collective Action by Employee-Inventors, 37 Santa Clara L. Rev., 673, 677 (1997) (advocating that employees refuse to sign invention assignment agreements and instead try to own workplace inventions themselves); Evelyn D. Pisegna-Cook, Ownership Rights of Employee Inventions: The Role of Preinvention Assignment Agreements and State Statutes, 2 U. Balt. Intell. Prop. L.J. 163, 185 (1994) (general survey of common law and state statutes; concluding by advocating additional compensation for employee-inventors and greater adoption of statutes to regulate invention assignment contracts); Steven Cherensky, Note, A Penny for Their Thoughts: Employee-Inventors, Preinvention Assignment Agreements, Property, and Personhood, 81 Cal. L. Rev. 595 (1993) (in an argument premised on a Hegelian-inflected notion of personhood and property rights, advocating a greater property allocation to employees for workplace inventions); Henrik D. Parker, Note, Reform for Rights of Employed Inventors, 57 S. Cal. L. Rev. 603, 624-25 (1984) (arguing that compensation schemes for workplace inventions would increase innovation); William P. Hovell, Note, Patent Ownership: An Employer’s Rights to His Employee’s Invention, 58 Notre Dame L. Rev. 863, 887-88 (1983) (advocating that employee-patent inventors receive a “reverse shop right” “to use any patent assigned to his employer”). All are speculative, and some rest on seemingly needless concepts of Hegelian personhood and the like. As much as I tend to favor departing employees, this line of articles is frustrating. Many seem to imagine that workplace inventions are created by a sole inventor rather than by teams, none contemplate the quagmires that could result in bickering over who owns what within the office, and the philosophical reference points seem arbitrary.

I consider all of these fully answered by Robert P. Merges, The Law and Economics of Employee Inventions, 13 Harv. J.L. & Tech. 1, 14-18 (1999) (arguing that the current employer-ownership regime is best calibrated to avoid problems such as disruptive employee “holdups”: “The employees could hold up the firm for the full value of its investment.”). Merges notes the seemingly obvious problems that would arise in scenarios where employees share in ownership of inventions made for the employer’s benefit. This essay is by contrast focused on outside-the-workplace creations by employees planning their next venture which have no connection, or only some tenuous connection, to their present employer’s business.

[21] Articles which address topics other than whether assigning workplace inventions to the employer is equitable include: Parker A. Howell, Whose Invention is it Anyway? Employee Invention-Assignment Agreements and Their Limits, 8 Wash. J.L. Tech. & Arts 79 (2012) (practitioner-oriented overview of state statutes and a sampling of case law); Joshua L. Simmons, Inventions Made for Hire, 2 N.Y.U. J. Intell. Prop. & Ent. L. 1, 49 (2012) (arguing that the law of employee-patent assignments contains too many procedural stumbling blocks and advocating that patent law be amended to match the work for hire terms of the Copyright Act); Donald J. Ying, A Comparative Study of the Treatment of Employee Inventions, Pre-Invention Assignment Agreements, and Software Rights, 10 U. Pa. J. Bus. & Emp. L. 763 (2008) (survey of invention assignment laws in the United States and other countries); Michael R. Mattioli, The Impact of Open Source on Pre-Invention Assignment Contracts, 9 U. Pa. J. Lab. & Emp. L. 207, 228-34 (2006) (article on what might happen if an employee creates a patentable invention while contributing to an open source project).

[22] See generally Solomons v. United States, 137 U.S. 342, 346 (1890) (“[W]hatever invention [an inventor] may thus conceive and perfect is his individual property.”); Gayler v. Wilder, 51 U.S. 477, 478–93 (1851) (“But the discoverer of a new and useful improvement is vested by law with an inchoate right to its exclusive use, which he may perfect and make absolute by proceeding in the manner which the law requires.” The case involved an assignment by the inventor to a businessperson.).

[23] “[U]nless there is an agreement to the contrary, an employer does not have rights in an invention ‘which is the original conception of the employee alone.’” Bd. of Trs. of the Leland Stanford Junior Univ. v. Roche Molecular Sys., 563 U.S. 776, 786 (2011) (quoting United States v. Dubilier Condenser Corp., 289 U.S. 178, 189 (1933)). “In most circumstances, an inventor must expressly grant his rights in an invention to his employer if the employer is to obtain those rights” because “the respective rights and obligations of employer and employee, touching an invention conceived by the latter, spring from the contract of employment.” Id. at 786. (quoting Dubilier, 289 U.S. at 187). Thus, the Court has “rejected the idea that mere employment is sufficient to vest title to an employee’s invention in the employer.” Id. at 789. However, there is an exception for employment situations where the employee has “only produced that which he was employed to invent.” Dubilier, 289 U.S. at 187. This “hired to invent” exception exists because the employee’s invention “is the precise subject of the contract of employment. A term of the agreement necessarily is that what he paid to produce belongs to his paymaster.” Id. at 187. “On the other hand, if the employment be general, albeit it cover a field of labor and effort in the performance of which the employee conceived the invention for which he obtained a patent, the contract is not so broadly construed as to require an assignment of the patent.” Id. The Supreme Court first applied the hired to invent exception in a 1924 case, Standard Parts Co. v. Peck. In that case, the defendant was employed under a contract to “devote his time to the development of a process and machinery for the production of the front spring now used on the product of the Ford Motor Company.” Standard Parts Co. v. Peck, 264 U.S. 52, 59 (1924). After inventing the specific process and machinery contemplated by his employment contract, the defendant secured a patent on the aforementioned technology which he refused to convey to the plaintiff. Finding that the invention of the specific patent at issue was the “object and effect” of the defendant’s employment contract, and that in inventing the technology at issue, the defendant was “doing nothing more than he was engaged to do and paid for doing,” the Court held that the patent belonged to the plaintiff. Id. at 59-60.

For other early cases, see Houghton v. United States, 23 F.2d 386, 389 (4th Cir. 1928) (Surgeon General appointed government employees to a board whose purpose was to develop a safer fumigant, and thus the defendant “did merely that which he was being paid his salary to do,” so there could be “no doubt” that his invention belonged to the government.); Magnetic Mfg. Co. v. Dings Magnetic Separator Co., 16 F.2d 739, 741 (7th Cir. 1926) (on contested facts, finding that employee was hired to invent from his own conduct, where he had promptly assigned two prior inventions, and thus owed an assignment duty even as to an invention that related to only a peripheral part of his employment duties). An employer might overcome hired-to-invent issues where a manager came up with the invention, and the employee claiming ownership was merely the one who reduced it to practice. In one early California case, the court distinguished between an employer who had the “general idea of a machine” and an employee who implemented the idea – even though the employee was the signatory on the patent application – in granting ownership to the employer in an improvement for movie studio lighting. Famous Players-Lasky Corp. v. Ewing, 194 P. 65, 66 (Cal. Dist. Ct. App. 1920).

[24] See, e.g., Fish v. Air-O-Fan Prods. Corp., 285 F.2d 208, 210-11 (9th Cir. 1960) (finding implied in fact assignment contract where evidence showed employee was hired to invent); Consol. Vultee Aircraft Corp. v. Maurice A. Garbell, Inc., 204 F.2d 946, 949 (9th Cir. 1953) (although employee had conceived invention before joining employer, he had used employer’s resources to perfect it, and thus employer had shop right to manufacture, use, and sell “airplanes embodying” the invention); Aero Bolt & Screw Co. v. Iaia, 5 Cal. Rptr. 53, 58 (Cal. Ct. App. 1960) (employee retained ownership of invention with no shop right for employer where, among other things, trial court found that employee “on his own time, and at his own expense, designed and developed the [invention]; that he paid all of the development costs out of his own pocket”); Zahler v. Columbia Pictures Corp., 4 Cal. Rptr. 612, 617 (Ct. App. 1960) (where employee was hired to create background music, court applied a version of the hired-to-invent doctrine to grant employer performance rights; “Where an employee creates something as part of his duties under his employment, the thing created is the property of his employer.”); Banner Metals, Inc. v. Lockwood, 3 Cal. Rptr. 421, 428-33 (Ct. App. 1960) (where salesperson had only an oral employment contract, court found that employer did not own the employee’s invention through the hired-to-invent-doctrine and also did not have a shop right in the invention, because the employee had created it on his own time and, on contested facts, had not used the employer’s materials); Quaker State Oil Ref. Co. v. Talbot, 174 A. 99, 101-04 (Pa. 1934) (employee was hired to invent under oral contract, but employer did not obtain shop right to employer’s pre-existing work that was not created with use of employer’s resources).

[25] See Peregrine Semiconductor Corp. v. RF Micro Devices, Inc., No. 3:12-CV-0911-H (WMC), 2014 U.S. Dist. LEXIS 2668, at *13 (S.D. Cal. Jan. 8, 2014) (where former employer sued former employee who transferred patent rights to a third party, and where employer could not prove it had a signed employment agreement, hired to invent doctrine did not favor employer on motion for preliminary injunction where it did not appear employee was hired for specific inventions at issue); Gen. Elec. Co. v. Wilkins, No. CV F 10-0674 LJO JLT, 2012 U.S. Dist. LEXIS 124642, at *59-60 (E.D. Cal. Aug. 31, 2012) (finding trial issues of fact on a hired-to-invent dispute where employee had not signed the company’s invention assignment agreement, but was a patent co-inventor); McClain v. State, 269 S.W.3d 191, 198-99 (Tex. App. 2008) (prosecution failed to show that defendant was hired to invent, such that improvements he made to “set up sheets” belonged to him, with a shop right to his employer); Pedersen v. Akona, 429 F. Supp. 2d 1130, 1143 (D. Minn. 2006) (in absence of invention assignment contract, finding that employer owned patent because employee was hired to invent); Scott Sys., Inc. v. Scott, 996 P.2d 775, 779 (Col. App. 2000) (reversing summary judgment where there were triable issues of fact as to whether individual was hired to invent or not with respect to certain inventions); Teets v. Chromalloy Gas Turbine Corp., 83 F.3d 403, 408 (Fed. Cir. 1996) (using the terminology of an implied-in-fact contract, but engaging in same analysis as hired to invent cases, reversing trial court and finding implied employee assignment obligation for invention where employee worked at company’s direction and used its resources); Liggett Grp., Inc. v. Sunas, 437 S.E.2d 674, 678-80 (N.C. Ct. App. 1993) (reversing grant of summary judgment in favor of employer and finding triable issues of fact as to whether employee with no invention assignment contract was hired to invent in connection with tobacco-related patent); California E. Lab., Inc. v. Gould, 896 F.2d 400, 402-03 (9th Cir. 1990) (defining shop right concept and finding that right transferred to successor entity); Aetna-Standard Eng’g Co. v. Rowland, 493 A.2d 1375, 1381-82 (Pa. Super. Ct. 1985) (employee with no written contract was not hired to invent and thus owned no assignment duty to employer of invention, but employer had shop right license because its resources were used in development); Mechmetals Corp. v. Telex Comput. Prods., Inc., 709 F.2d 1287, 1292-94 (9th Cir. 1983) (finding no shop right where company CEO had worked with two others to create invention, and company paid for materials and machine time CEO used when working on invention; existence of that contract and CEO’s status militated against finding shop right); Vigitron, Inc. v. Ferguson, 419 A.2d 1115, 1117-18 (N.H. 1980) (employer owned invention because employee was hired to invent).

[26] An invention assignment agreement with the phrase “hereby assign” (or similar language) operates to automatically vest a covered invention with the employer at the moment it is created by the employee. No further action – that is, no formal act of assignment by the employee – is required for the automatic invention assignment to be effective. See Roche, 563 U.S. at 785-93 (noting historical rule that employee-inventors maintain ownership of their inventions absent an agreement with the employer; university did not own invention where agreement contained “will assign” language, separate agreement with private company had “hereby assign” language, and Bayh-Dole Act did not change the ordinary rules of employee invention assignment contracts); Advanced Video Techs. LLC v. HTC Corp., 879 F.3d 1314, 1317-18 (Fed. Cir 2018) (invention assignment agreement failed where it stated only that employee “will assign” rights, which is not a present conveyance of future conceptions); St. Clair Intellectual Prop. Consultants, Inc. v. Palm, Inc., No. 06-404-JJF-LPS, 2009 U.S. Dist. LEXIS 37512, *34-35 (D. Del. May 4, 2009) (finding that language in invention assignment agreements did not create a present conveyance of future rights); Freedom Wireless, Inc. v. Boston Comm. Grp., Inc., 220 F. Supp. 2d 16, 19 (D. Mass. 2002) (“In order for a pre-invention assignment contract to create a present assignment of an expectant interest in an invention that automatically vests by operation of law into an actual assignment upon conception, the contract must contain words of present conveyance and must require ‘no further act once an invention [comes] into being.’” (citation omitted) (contrasting contracts that use phrases like “does hereby grant” and “hereby does assign” with a contract that required the inventor to disclose the invention and perform acts necessary to establish ownership); Imatec, Ltd. v. Apple Comput., Inc., 81 F. Supp. 2d 471, 482 (S.D.N.Y. 2000) (“‘I agree to assign and hereby do assign’ . . . constituted a present assignment . . . of future inventions.”); Arachnid, Inc. v. Merit Indus., Inc., 939 F.2d 1574, 1580 (Fed. Cir. 1991) (consulting agreement with phrase “will be assigned” “does not rise to the level of a present assignment of an existing invention, effective to transfer all legal and equitable rights therein”); Filmtec Corp. v. Allied-Signal, Inc., 939 F.2d 1568, 1573 (Fed. Cir. 1991) (invention assignment contract that used the phrase “agrees to grant and does hereby grant” was not merely an agreement to grant future rights, but expressly granted “rights in any future invention”); Treu v. Garrett Corp., 70 Cal. Rptr. 284, 287 (Ct. App. 1968) (where inventor had an assignment clause with his employer, the inventor’s ownership “was a fleeting thing which he had bargained away even before it became a reality . . . the instant the improvement was invented it became the exclusive property of [employer] . . . [plaintiff] had and could have no interest in the invention.”).

[27] See, e.g., Preston v. Marathon Oil Co., 277 P.3d 81, 87-88 (Wyo. 2012) (invention assignment agreement valid when entered into after at-will employment began); Eaton Corp. v. Giere, 971 F.2d 136, 140 (8th Cir. 1992) (both earlier and later invention assignment contracts were valid); Mosser Indus., Inc. v. Hagar, 200 U.S.P.Q. (BNA) 608, 610, 616 (Pa. Ct. Com. Pl. Jan. 11, 1978) (employee signed second invention assignment contract for one dollar); Grove v. Grove Valve & Regulator Co., 84 Cal. Rptr. 300, 303 (Ct. App. 1970) (employee’s valid invention assignment contract was “backdated” by more than a year).

[28] See Iconix, Inc. v. Tokuda, 457 F. Supp. 2d 969, 989, 992 (N.D. Cal. 2006).

[29] By analogy, a federal court in Ohio declined to find that a group of departing employees violated their duties of loyalty by making preparations to compete before leaving where, among other things, the employer’s evidence that they had used its resources was limited to a de minimis claim that one of them spoke to an insurer about the planned new venture “on his business cell phone.” See Fitness Experience, Inc. v. TFC Fitness Equip, Inc., 355 F. Supp. 2d 877, 892-93 (N.D. Ohio 2004) (employees who planned new company and met with attorneys, insurers, and realtors granted summary judgment). The same result might apply for similar de minimis infractions in the invention assignment context.

[30] E.g., Cal. Lab. Code § 2870 (West 1979).

[31] See Cadence Design Sys., Inc. v. Bhandari, No. C 07-00823 MHP, 2007 U.S. Dist. LEXIS 83078, at *21-23 (N.D. Cal. Nov. 8, 2007) (explaining scope of invention assignment in detail and rejecting argument that scope of assignment extends only to specific department in which the employee works).

[32] An attorney had even warned the employee on this point. See id. at *21-22.

[33] While there is little case law on the question, courts have rejected employees’ assertions that company managers or other employees waived or otherwise acquiesced to their personal use of inventions created during employment on two grounds: (a) because an invention automatically vests in the employer at the moment of creation, before the time of any claimed waiver; and/or (b) because the employee did not completely or fully disclose every facet of the invention to the employer, the employer could not have waived an interest in something it did not know everything about. See, e.g., DDB Techs., L.L.C. v. MLB Advanced Media, L.P., 517 F.3d 1284, 1289-90 (Fed. Cir. 2008) (employee’s arguments of waiver and estoppel based on manager’s disinterest in idea were irrelevant because contract gave rise to automatic assignment at moment of creation); Iconix, 457 F. Supp. 2d at 986-87 (employee’s claim that manager and general counsel had expressed disinterest in the idea did not prevent invention assignment where idea was never fully disclosed). See also Eaton Corp., 971 F.2d at 137-38 (employee told employer just before his resignation “of his plans to develop and market his own competitive device” but invention vested in employer under invention assignment contract).

[34] See Applied Materials, Inc. v. Advanced Micro-Fabrication Equip. (Shanghai) Co., 630 F. Supp. 2d 1084, 1089-1091 (N.D. Cal. 2009) (voiding a clause under California law that created a purportedly rebuttable presumption that the employer owned inventions for one year after the employee left, even where no trade secrets were misappropriated); Ingersoll-Rand Co. v. Ciavatta, 542 A.2d 879, 890-95 (N.J. 1988) (applying a reasonableness analysis to a post-employment holdover invention assignment clause; finding that such covenants might be enforceable beyond “trade secrets and confidential information” to include some additional, poorly-defined category of “highly specialized, current information not generally known in the industry” that appears indistinguishable from contemporary definitions of what trade secret law covers); Fed. Screw Works v. Interface Sys., Inc., 569 F. Supp. 1562, 1564 (E.D. Mich. 1983) (agreement which required employees to assign all inventions for an indefinite time period relating to the employer’s field of activity or “contemplated field of activity” found unenforceable); Armorlite Lens Co. v. Campbell, 340 F. Supp. 273, 275 (S.D. Cal. 1972) (finding clause in employee invention assignment agreement which required assignment of new ideas and concepts for one year after employment ended was unenforceable unless such inventions used the trade secrets of the former employer); Winston Research Corp. v. Minn. Mining & Mfg. Co., 350 F.2d 134, 145-46 (9th Cir. 1965) (where employer’s contract required invention assignment of post-employment inventions that were “based upon” its confidential information, court affirmed rulings where such inventions were premised on former employer’s trade secrets, and thus owned by employer, or where no trade secrets were used); De Long Corp. v. Lucas, 176 F. Supp. 104, 127-28 (S.D.N.Y. 1959) (where settlement agreement between former employer and former employee only assigned rights to inventions developed during employment, and where evidence was that new inventions post-dated employment, former employee owed no assignment obligation), aff’d, 278 F.2d 804 (2d Cir. 1960); Guth v. Minn. Mining & Mfg. Co., 72 F.2d 385, 388 (7th Cir. 1934) (invention assignment agreement without time limitation was “contrary to public policy”); cf. Yield Dynamics, Inc. v. TEA Sys. Corp., 66 Cal. Rptr. 3d. 1, 26 (Ct. App. 2007) (where employer’s invention assignment contract created a one-year, post-employment holdover stating that there was a rebuttable presumption that inventions conceived within one year of leaving are owned by the employer, and that inventions during that time must be disclosed, court did not address legality of the clause and affirmed trial court on grounds that employee disclosed a post-employment development to the former employer, which did not request assignment).

[35] See Milliken & Co. v. Morin, 731 S.E.2d 288, 295 (S.C. 2012) (finding a one-year holdover clause enforceable even though it covered inventions created post-employment and without use of the employer’s trade secrets).

[36] For a well-written student note critiquing holdover clauses, see Marc B. Hershovitz, Note, Unhitching the Trailer Clause: The Rights of Inventive Employees and Their Employers, 3 J. Intell. Prop. L. 187, 206-12 (1995).

[37] See Feeney v. Transition Automation, Inc., No. 06-11677, 2008 WL 190766, at *17-19 (D. Mass. Jan. 9, 2008) (where individual disclosed invention before becoming employee and signing invention assignment contract, court interpreted language of contract to vest ownership with individual, not employer); Voith Hydro, Inc. v. Hydro W. Group, Inc., No. C–96–1170 SC, 1997 WL 154400, at *6 (N.D. Cal. Mar. 26, 1997) (although company had a valid employee invention assignment with individual, company did not own invention that employee had conceived and disclosed before becoming an employee of the company); Bailey v. Chattem, Inc., 684 F.2d 386, 391 (6th Cir. 1982) (employer did not own employee’s idea that “was complete prior to the time he began to work for” the employer, and thus prior to invention assignment agreement); Fox v. Kingsland, 81 F. Supp. 433, 437-38 (D.D.C. 1948) (where inventor was employed by Patent Office and also the Signal Corps, he retained ownership despite written agreement with the Signal Corps because invention pre-dated that employment).

[38] The incorporation of pre-existing material does not defeat an invention assignment of the new combination as a whole. See, e.g., ViChip Corp. v. Lee, 438 F. Supp. 2d 1087, 1095 (N.D. Cal. 2006) (granting summary judgment and rejecting employee’s argument that some of the technology in the invention was “preexisting”; employee failed to identify such material or explain how it could be “carved out from” the overall invention that was subject to his employment agreement); Mosser Indus., Inc. v. Hagar, 200 U.S.P.Q. (BNA) 608, 613-14 (Pa. Ct. Com. Pl. Jan. 11, 1978) (granting injunction for employer; even if employee thought up “individual elements” of his creation before he joined the employer, his “combination of those elements into a working prototype occurred during” his employment, and thus the invention assignment clause applied to that “combination”).

[39] See, e.g., Imatec, Ltd. v. Apple Comput., Inc., 81 F. Supp. 2d 471, 482 (S.D.N.Y. 2000) (prior employer had invention assignment contract which covered invention, and thus former employee had nothing to assign to third party and third party lacked standing to sue over patent at issue); Filmtec Corp. v. Allied-Signal, Inc., 939 F.2d 1568, 1573-74 (Fed. Cir. 1991) (stating same rule and remanding for factual findings as to whether company had actual or “inquiry” notice of a prior invention assignment to a former employer or a third party and thus may not have been bona fide purchaser); Thompson v. Automatic Fire Prot. Co., 211 F. 120, 121 (2d Cir. 1914) (where inventor already had assigned invention to plaintiff, and where defendants knew of that assignment, assignment to defendants was invalid: “It seems to us that a business man of reasonable care and prudence would, under these circumstances, before putting his money into an enterprise, have gone to [plaintiff] and asked him if he was making any claim to this invention of [inventor] and, if he said he was, would have asked him what was the nature of his claim, so that the inquirer might advise himself whether he could safely purchase.”).

[40] See Venture Corp. v. Barrett, No. 5:13-cv-03384-PSG, 2015 U.S. Dist. LEXIS 165809, at *13-14 (N.D. Cal. Dec. 9, 2015) (affirming jury finding that employee’s invention belonged to employer because it was created using the employer’s time or resources); Blackbird Techs., Inc. v. Joshi, No. 5:15-cv-04272-EJD, 2015 U.S. Dist. LEXIS 136505, at *24 n.6 (N.D. Cal. Oct. 6, 2015) (issuing preliminary injunction under duty of loyalty theory against employee who started competitive business while employed; terms of injunction barred use of inventions covered by invention assignment agreement); Preston v Marathon Oil Co., 684 F.3d 1276, 1359-60 (Fed. Cir. 2012) (invention assignment agreement valid and assigned rights in invention to employer); NovelAire Techs., LLC v. Harrison, 50 So. 3d 913, 919-21 (La. Ct. App. 2010) (affirming finding that employee violated invention assignment agreement by failing to disclose invention he created and took to new venture after departing); Iconix, Inc. v. Tokuda, 457 F. Supp. 2d 969, 989-92 (N.D. Cal. 2006) (issuing preliminary injunction over social media tool that related to company’s advertising strategy plan); Regents of the Univ. of N.M. v. Knight, 321 F.3d 1111, 1118 (Fed. Cir. 2003) (affirming finding that university patent policy vested ownership in invention in institution); Fenn v. Yale Univ., 283 F. Supp. 2d 615, 629-30 (D. Conn. 2003) (academic bound by university’s patent assignment policies, despite policy amendments since he agreed to it, and thus university owned patent he had licensed to a third party); Univ. of W. Va. v. Van Voorhies, 278 F.3d 1288, 1298 (Fed. Cir. 2002) (academic bound to assign patent under university’s patent policy); Waterjet Tech., Inc. v. Flow Int’l Corp., 996 P.2d 598, 601-02 (Wash. 2000) (where former employee refused to assign rights over patent to workplace invention, court found notice to employee satisfied Washington’s invention assignment statute such that invention assignment contract was enforceable to vest ownership in employer); Vt. Microsys., Inc. v. Autodesk, Inc., 88 F.3d 142, 150-51 (2d Cir. 1996) (affirming ruling that defendant, who started as intern and later became employee and signed an invention assignment agreement, had no interest in software created for the company); Eaton Corp. v. Giere, 971 F.2d 136, 140 (8th Cir. 1992) (granting summary judgment in favor of employer where employee had breached invention assignment agreement by creating a product within employer’s “actual or demonstrably anticipated research and development,” even though employee’s invention related to a market which employer had “not yet been able to crack”); Cubic Corp. v. Marty, 229 Cal. Rptr. 828, 835 (Ct. App. 1986) (affirming judgment for employer where assignment provision was for “inventions which relate to the ‘actual’ or ‘demonstrably anticipated’ business of the employer); Goldwasser v. Smith Corona Corp., 817 F. Supp. 263, 276 (D. Conn. 1993) (entering summary judgment for IBM, noting that employee’s invention or idea which “involve[d] the entry of text into a computer relates to IBM’s ‘actual or anticipated business’”); Gen. Signal Corp. v. Primary Flow Signal, Inc., No. 85-0471B, 1987 WL 147798, at *4 (D.R.I. July 27, 1987) (granting employer ownership of invention where employee was subject to invention assignment agreement, and claimed to have come up with new ideas just five days after leaving job; court found employee’s account “difficult to believe”); Syntex Ophthalmics, Inc. v. Tsuetaki, 701 F.2d 677, 682 (7th Cir. 1983) (affirming finding in favor of employer on invention assignment dispute given “the inherent improbability” of employee’s “story that he conceived of the [invention] four days after he left” his job); Andreaggi v. Relis, 408 A.2d 455, 460 (N.J. Super. Ct. Ch. Div. 1979) (invention related to employee’s work and thus fell within invention assignment agreement); Goodyear Tire & Rubber Co. v. Miller, 22 F.2d 353, 354 (9th Cir. 1927) (where defendant signed an employment contract in which he agreed expressly to assign title to any inventions that related to “methods, processes, or apparatus concerned with the production of any character of goods or materials sold or used” or “any character of goods or materials sold or used” by Goodyear, employer owned invention that fell within general range of what employee was hired to do).

[41] See Mattel, Inc. v. MGA Entm’t, Inc., 616 F.3d 904, 909-12 (9th Cir 2010) (vacating constructive trust for former employer because invention assignment agreement did not contain the word “ideas” and remanding for consideration of extrinsic evidence; court notably did not construe California Labor Code section 2870 except for a passing reference in a footnote). One student commentator found fault with the ruling, but erroneously conflated California’s invention assignment statute with the work for hire doctrine. See Connor Boyd, Note, The Bratz Trap: Ownership and Infringement at the Nexus of Copyright and Employment Law, 45 U.C. Davis L. Rev. 221, 237, 245 (2011) (asserting that section 2870 “refined” the work for hire doctrine, assuming that a state statute could alter the federal Copyright Act and that its coverage was not instead for trade secrets and patentable inventions).

[42] One court denied (in whole or in part) cross-motions for summary judgment where there were disputed facts regarding an employee’s development of source code on his own time, for an unrelated side business, but that the employer contended contained “similarities” with its own source code. See Enreach Tech., Inc. v. Embedded Internet Sols., Inc., 403 F. Supp. 2d 968, 974-75 (N.D. Cal. 2005). See also Rothschild v. Cree, Inc., 711 F. Supp. 2d 173, 182-83 (D. Mass. 2010) (finding fact issues on jurisdictional matters as to individual’s standing to sue for patent infringement where it was unclear whether individual had developed key patent ideas while under another company’s invention assignment agreement or afterwards).

[43] See cases cited supra notes 7, 34, 36. As with any attempt to count holdings from reported cases, caveats are in order. The LEXIS and Westlaw databases do not pick up every ruling in every case, especially those from state trial courts. Some rulings in trade secret cases may be under seal, so that details are not viewable to the public. Appellate rulings matter more than trial court rulings, but the degree of careful analysis varies from case to case. Reported cases may or may not reflect typical fact patterns. Many disputes settle without litigation, or settle before a court issues a ruling on issues of substantive law. As a result, apparent trends in case law can be somewhat arbitrary, as they depend on what lawsuits become easily available to researchers.

[44] Of the three, one remanded on a rather unique contract interpretation, one is unpublished, and one arose from a time period before California enacted its current invention assignment statute. The same outcome under today’s statute seems unlikely. See Mattel, 616 F.3d at 909-12; Applera Corp. v. Illumina, Inc., 375 Fed. App’x. 12, 17-18 (Fed. Cir. 2010) (in a case decided under an older version of California’s invention assignment statute, the court affirmed a finding that employee-attorney’s invention did not “result from” work for the plaintiff); NeoNetworks, Inc. v. Cree, A07-0729, 2008 Minn. App. Unpub. LEXIS 565, at *15 (Ct. App. May 20, 2008) (affirming dismissal of claims against former employees and costs award in their favor after failed ex-employers causes of action, including breach of contract over invention assignment, were rejected where defendants began project after company failed and did not use company resources).

[45] Cmty. for Creative Non-Violence v. Reid, 490 U.S. 730, 743 (1989)

[46] As the text of the statute indicates, no contract is required for a work for hire to exist. Nonetheless, as with the law of employee invention assignments, companies generally insert work for hire assignment terms into standard-form employment contracts, usually if not always next to the terms regarding invention assignment.

[47] See Reid, 490 U.S. at 743 (key case on the work for hire doctrine); JustMed, Inc. v. Byce, 600 F.3d 1118, 1128 (9th Cir. 2010) (affirming finding that software programmer was employee, giving company ownership of software as work for hire).

[48] Id. at 750-51 (“To determine whether a work is for hire under the Act, a court first should ascertain, using principles of the general common law of agency, whether the work was prepared by an employee or an independent contractor.”).

[49] Whether someone is an employee or an independent contractor for purposes of Copyright Act analysis involves the weighing of numerous factors. Reid lists twelve, none of which are “determinative”: (1) the hiring party’s right to control the manner and means by which the product is accomplished; (2) the skill required; (3) the source of instrumentalities and tools; (4) the location of the work; (5) the duration of the relationship between the parties; (6) whether the hiring party has the right to assign additional projects to the hired party; (7) the extent of the hired party’s discretion over when and how long to work; (8) the method of payment; (9) the hired party’s role in hiring and paying assistants; (10) whether the work is part of the regular business of the hiring party; (11) whether the hiring party is a business; (12) the provision of employee benefits and the tax treatment of the hired party. See id. at 751-52. Some courts have suggested that the tax/benefits classification of the author is the most important factor, or at least a disproportionately important factor. See Hi-Tech Video Prods., Inc. v. Capital Cities/ABC, Inc., 58 F.3d 1093, 1097 (6th Cir. 1995) (factor is a “strong indication of a worker’s employment status”); Kirk v. Harter, 188 F.3d 1005, 1008 (8th Cir. 1999) (finding factor important where hiring party reported author to IRS as an independent contractor); Aymes v. Bonelli, 980 F.2d 857, 862-63 (2d Cir. 1992) (failure to pay taxes and benefits “constitute virtual admissions of [author’s] status by [hiring party] himself;” citing numerous post-Reid cases and saying “every case since Reid that has applied the test has found the hired party to be an independent contractor where the hiring party failed to extend benefits or pay social security taxes.”) (citing cases); Numbers Licensing, LLC v. bVisual USA, Inc., 643 F. Supp, 2d 1245, 1251-52 (E.D. Wash. 2009) (where individual had his own consulting entity and company contracted with that entity, and where entity was responsible for individual’s “payroll obligations, tax obligations, and employee benefits,” those and other factors led to court’s conclusion that individual was independent contractor and software was not a work for hire). One court, however, rejected the “usually dispositive” tax/benefits factor where the author worked full-time, was paid weekly, believed he had a long-term relationship, was shown the hiring party’s source code, worked with the hiring party’s equipment and accessed the hiring party’s servers from home, telecommuted but also worked twice a week at the hiring party’s facility for at least some period, assisted with other assignments, received assistance in software programming from the hiring party, had access to the hiring party’s office and equipment, and met a third party as a representative of the hiring party. See Sasnett v. Convergent Media Sys., Inc., No. 95-12262-NG, 1997 WL 33142149, at *9 (D. Mass. Aug. 29, 1997) (on preliminary injunction request, finding that plaintiffs were likely to succeed on the merits).

Courts have identified other possible factors beyond those set forth in Reid. A possible additional factor could be whether the corporation existed at the time of creation. See Billy-Bob Teeth, Inc. v. Novelty, Inc., 329 F.3d 586, 591 (7th Cir. 2003) (doctrine did not apply where, among other things, “the corporation did not exist when the [works] were authored”; company incorporated the year after creation of works). Another possible additional factor is whether the author worked for others at the same time. See Kirk, 188 F.3d at 1008 (“[T]hroughout his six-year relationship with [hiring party], [author] continued to engage in computer consulting with other companies, a factor suggesting that he was an independent contractor.”); Aymes, 980 F.2d at 864 (finding it probative that author “did occasional work for others at the same time” despite working for two years for hiring party); Alcatel USA, Inc. v. Cisco Sys., Inc., 239 F. Supp. 2d 645, 654 (E.D. Tex. 2002) (noting that author “was simultaneously working as an independent contractor for two other companies”). Yet another possible factor is the hiring party’s own perceptions of whether the author was an employee or not. See Hi-Tech Video Prods., Inc., 58 F.3d at 1099 (fact that hiring party did not perceive assistants as employees weighed in favor of finding them to be independent contractors); see also Alcatel USA, 239 F. Supp. 2d at 654 (finding it significant that hiring party had called author an independent contractor in a prior lawsuit before time “when it finally became advantageous for [hiring party] to treat [author] as its employee in order for it to assert ownership of [author’s] creations”).

[50] See Massingill v. Stream, Ltd., No. 3:08-cv-0091-M, 2009 WL 3163549, at *5 (N.D. Tex. Oct. 1, 2009) (company’s motion for summary judgment on ownership of software program denied; fact issues remained where individual started work on software as contractor, later became employee, and whether the work “continued to be developed during his employment with [company] such that it would constitute a ‘work for hire’ under the Copyright Act.”).

[51] 17 U.S.C. § 101 (2012).

[52] As Nimmer puts it, “Under the current Act, even if a work is prepared by an employee, if it is not prepared ‘within the scope of his or her employment,’ it is not regarded as a ‘work made for hire.’ Therefore, an agreement between employer and employee whereby works prepared by the employee that are not prepared within the scope of employment are nonetheless deemed to be ‘works made for hire,’ will not in itself convert such works into the ‘for hire’ category.” 1 Melville B. Nimmer and David Nimmer, Nimmer on Copyright § 5.03[B][1][b][ii] (Matthew Bender, Rev. Ed. 2018).

[53] I have found no empirical study of employment contract terms assigning works made for hire. That said, many samples are publicly available. My own anecdotal experience, based on viewing thousands of employment agreements in Silicon Valley and elsewhere, drafted by a wide variety of law firms, suggests that most recite that works for hire in the scope of employment are assigned – that is, the typical agreement follows the text of the Copyright Act. A very typical sample, pulled from a website which collects such standard terms, reads: “Employee acknowledges that all original works of authorship which are made by him (solely or jointly with others) within the scope of his employment and which are protectable by copyright are works made for hire, pursuant to the United States Copyright Act (17 U.S.C. SS 101 et seq.), and are solely the property of the Company.” See Law Insider, Works Made for Hire Sample Clauses, (last visited Oct. 17, 2018).

Two other points about such contracts are worth noting. First, an employer and an employee can negotiate a different structure by contract – such as a royalty-based scheme – though it seems unlikely that a form contract could undercut rights granted by the Copyright Act through the scope of employment test any more than an invention assignment contract could get around state statutory limits on the scope of such agreements. See generally Comput. Data Sys., Inc. v. Kleinberg, 759 F. Supp. 10, 13-14 (D.D.C. 1990) (triable issue of fact on scope of employment because contract stated that if employee developed certain computer software and agreed to transfer it to employer, he would get a royalty to be negotiated at the time).

Second, as with any matter of employment contracts, questions of timing matter. Some courts have taken a position on the question whether, in general, a post-creation writing can give a hiring party work for hire rights. Because the statute’s text seems to indicate that ownership starts from the time of creation, the Seventh Circuit held in 1992 that there cannot be a work for hire based on a written agreement unless the writing predates the creation of the work. See Schiller & Schmidt, Inc. v. Nordisco Corp., 969 F.2d 410, 413 (7th Cir. 1992) (basing its rule on the need to have certainty). By contrast, the Second Circuit rejected this approach and held that it is possible to have a work for hire based on a post-creation written agreement, so long as that agreement evidences a prior agreement before the time of creation. See Playboy Enters., Inc. v. Dumas, 53 F.3d 549, 559 (2d Cir. 1995) (noting that the Nimmer treatise disagreed with Schiller and holding that “the writing requirement . . . can be met by a writing executed after the work is created, if the writing confirms a prior agreement, either explicit or implicit, made before the creation of the work”; remanding on that factual question). See also Compaq Comput. Corp. v. Ergonome Inc., 210 F. Supp. 2d 839, 843-45 (S.D. Tex. 2001) (following Playboy, holding that test requires showing that the parties “understood at the time the [works] were created that the works were made for hire,” and finding test met where written agreement made express reference to work done the year before; written agreement predated author’s subsequent assignment to another company by several years); Kasten v. Jerrytone, No. 02-421, 2004 WL 1857680, at *6 (E.D. La. Aug. 16, 2004) (noting Playboy rule, finding that agreement at issue did not mention previous works, only future works, but separately holding that contract was modified by acceptance of royalties for pre-agreement works).

[54] Restatement (Second) of Agency § 228 (Am. Law Inst. 1958).

[55] Although rarely cited by the courts, the Restatement (Second) of Agency has ten sections regarding “scope of employment.” For the most part, the focus of these sections is on whether the employer will be liable for tortious or illegal acts by an employee. As a result, many of the annotated cases address such matters as joy rides in the company car and assaults by employees, with little value for work for hire disputes. That said, section 229, “Kind of Conduct Within Scope of Employment,” details factors that might assist courts in work for hire disputes, such as “whether or not the act is one commonly done by such servants,” “the time, place and purpose of the act,” and “the similarity in quality of the act done to the act authorized.” Sections 235 and 236, in turn, address the third factor of the employee’s purpose, or intentions. Section 235 states “An act of a servant is not within the scope of employment if it is done with no intention to perform it as a part of or incident to a service on account of which he is employed.” The comment says “It is the state of the servant’s mind which is material.” Section 236 states “Conduct may be within the scope of employment, although done in part to serve the purposes of the servant or a third person.” That is consistent with the case law cited above. Id. §§ 228-237.

[56] See Genzmer v. Pub. Health Tr. of Miami-Dade Cty., 219 F. Supp. 2d 1275, 1280 (S.D. Fla. 2002) (stating rule: “Courts interpret this test as conjunctive. That is, the party attempting to establish that the work was made for hire must establish all three elements.”); Moonstruck Design, LLC v. Metz, No. 02 CIV. 4025(RWS), 2002 WL 1822927, at *3 (S.D.N.Y. 2002) (stating rule); Quinn v. City of Detroit, 988 F. Supp. 1044, 1052 (E.D. Mich. 1997) (same); Roeslin v. District of Columbia, 921 F. Supp. 793, 798 (D.D.C. 1995) (same); City of Newark v. Beasley, 883 F. Supp. 3, 8 (D.N.J. 1995) (same).

[57] See Avtec Sys. v. Peiffer, 21 F.3d 568, 571 (4th Cir. 1994) (noting that if first factor is met, courts will not favor employee just because work was done off-site); McKenna v. Lee, 318 F. Supp. 2d 296, 301 (E.D.N.C. 2002) (same; prisoner did work in his cell, but license plate design was what he was hired to do); Martin v. City of Indianapolis, 982 F. Supp. 625, 634 (S.D. Ind. 1997) (noting that you can’t avoid the work for hire doctrine by doing the work off-site), aff’d, 192 F.3d 608 (7th Cir. 1999); Marshall v. Miles Labs., Inc., 647 F. Supp. 1326, 1331 (N.D. Ind. 1986) (pre-Reid case; article held to be work for hire even though employee did much work away from the employer’s facility because work related to the things plaintiff was hired to do, and because it was based on research done by another employee for employer); Miller v. CP Chems., 808 F. Supp. 1238, 1243 (D.S.C. 1992) (employee worked off-site on software but “[o]n the other hand, the work was performed during the time period in which he was employed by [employer].”).

[58] See, e.g., U.S. Auto Parts Network, Inc. v. Parts Geek, LLC, 692 F.3d 1009, 1018-19 (9th Cir. 2012) (remanding where there was a triable issue of fact as to whether portion of software written when programmer was an employee was within his scope of employment, where among other things he “lived rent-free in a house that doubled” as a company office and did not “strictly” differentiate between work and personal hours, so that his work was “within authorized work hours and space limits.”).

[59] As of May 2018 when this research was completed, these cases are: Genzmer, 219 F. Supp. 2d at 1275; Molinelli- Freytes v. Univ. of P.R., No. 09-1655(DRD), 2012 U.S. Dist. LEXIS 143262 (D.P.R. Sept. 30, 2012); Fleurimond v. N.Y. Univ., 876 F. Supp. 2d 190 (E.D.N.Y. 2012); Le v. City of Wilmington, 736 F. Supp. 2d 842 (D. Del. 2010); Rouse v. Walter & Assocs., LLC, 513 F. Supp. 2d 1041 (S.D. Iowa 2007); McKenna, 318 F. Supp. 2d at 296; Miller, 808 F. Supp. at 1238; Vanderhurst v. Colorado Mountain Coll. Dist., 16 F. Supp. 2d 1297, 1307 (D. Col. 1998); Lewis v. Activision Blizzard, Inc., No. C 12-1096 CW, 2013 U.S. Dist. LEXIS 149784 at *1 (N.D. Cal. Oct. 17, 2013); and Sterpetti v. E-Brands Acquisition, LLC, No. 6:04-cv-1843-Orl-31DAB, 2006 U.S. Dist. LEXIS 21407 at *1 (M.D. Fla. Apr. 20, 2006).

[60] These are: Avtec, 21 F.3d at 568; Moonstruck, 2002 WL 1822927 at *1; TAP Worldwide, LLC v. Becker, No. CV 10-04903 DMG (JCx), 2010 WL 2757354, at *1 (C.D. Cal. July 12, 2010); PFS Dist. Co. v. Raduechel, 332 F. Supp. 2d 1236 (S.D. Iowa 2004), Roeslin, 921 F. Supp. at 793; Quinn, 988 F. Supp. at 1044; Food Lion, Inc. v. Capital Cities/ABC, Inc., 946 F. Supp. 420 (M.D.N.C. 1996); Beasley, 883 F. Supp. at 3; Cannon Grp., Inc. v. Better Bags, Inc., 250 F. Supp. 2d 893, 901 (S.D. Ohio 2003); and Martin, 982 F. Supp. at 625.

[61] See Ryan Vacca, Work Made for Hire – Analyzing the Multifactor Balancing Test, 42 Fla. St. U. L. Rev. 197 (2014) (thorough history and analysis of case law after Reid on employee versus contractor analysis); Llewellyn Joseph Gibbons, Love’s Labor’s Lost: Marry for Love, Copyright Work Made-for-Hire, and Alienate at Your Leisure, 101 Ky. L.J. 113, 180 (2013) (addressing work for hire in the community property context); Jon M. Garon & Elaine D. Ziff, The Work Made for Hire Doctrine Revisited: Startup and Technology Employee and the Use of Contracts in a Hiring Relationship, 12 Minn. J.L. Sci. & Tech. 489, 505-09 (2011) (addressing work for hire agreements and asserting in passing that, as to California’s invention assignment statute, “[t]he two exceptions in section 2870 reasonably approximate the ‘scope of employment’ prong of the work for hire doctrine, as applied to the state law of assignment of employee work product”); Assaf Jacob, Tort Made for Hire – Reconsidering the CCNV Case, 11 Yale J.L. & Tech. 96, 143-144 (2009) (touching on the scope of employment test, but primarily focusing on whether agency law provides the best fit for assessing employee/contractor status); Deborah Tussey, What Ifs and Other Alternative Intellectual Property and Cyberlaw Story: What If Employees Owned Their Copyrights?, 2008 Mich. St. L. Rev. 233, 242 (2008) (“In short, if Congress had not adopted the work for hire doctrine during the early years of corporatization of the copyright industries, it probably would have had to invent an alternative legal structure to handle the licensing of collaborative works in the later years of the copyright terms of those works.”); Vai Io Lo, Employee Inventions and Works for Hire in Japan: A Comparative Study Against the U.S., Chinese, and German Systems, 16 Temp. Int’l & Comp. L.J. 279, 305-06 (2002) (noting, in survey, that Japanese law requires “reasonable compensation if the employee assigns the right to obtain a patent, the patent right, or grant an exclusive license to the employer.”); Carolyn M. Salzmann, Comment, You Commissioned It, You Bought It, But Do You Own It? The Work for Hire: Why Is Something So Simple, So Complicated?, 31 U. Tol. L. Rev. 497 (2000) (focusing on a writing requirement for work for hire in order to clarify ownership); Colleen Creamer Fielkow, Note, Clashing Rights Under United States Copyright Law: Harmonizing an Employer’s Economic Right with the Artist-Employee’s Moral Rights in a Work Made for Hire, 7 DePaul-LCA J. Art & Ent. L. 218, 220 (1997) (reviewing potential conflicts between work for hire and employee-authors’ moral-rights); Shannon M. Nolley, Note, The Work for Hire Doctrine and the Second Circuit’s Decision in Carter v. Helmsley-Spear, 7 DePaul-LCA J. Art. & Ent. L. 103, 128 (1996) (examining the independent contractor vs. employee aspects of work for hire); Jennifer Sutherland Lubinski, Note, The Work for Hire Doctrine under Community for Creative Non-Violence v. Reid: An Artist’s Fair Weather Friend, 46 Cath. U. L. Rev. 119, 125 (1996) (arguing for revamp of Reid test to better benefit artists and incentivize creation of more works); Alan Hyde & Christopher W. Hager, Promoting the Copyright Act’s Creator-Favoring Presumption: “Works Made for Hire” Under Aymes v. Bonelli & Avtec Systems, Inc. v. Peiffer, 71 Denv. U. L. Rev. 693, 716-717 (1994) (primarily examining employee versus contractor case law and approving the concept that absent a written agreement, authors may often own their creations); Anne Marie Hill, Note, The “Work for Hire” Definition in the Copyright Act of 1976: Conflict Over Specially Ordered or Commissioned Works, 74 Cornell L. Rev. 559, 564-66 (1989) (early work assessing pre-Reid case law); Jon L. Roberts, Work Made for Hire: The Fiction, the Reality, and the Impact Upon Software Development, 1 Harv. J. L. & Tech. 97, 128 (1988) (addressing early employee versus contractor case law); Rochelle Cooper Dreyfus, The Creative Employee and the Copyright Act of 1976, 54 U. Chi. L. Rev. 590, 638-47 (1987) (examining work for hire law primarily in the context of academic versus university control over scholarly output).

[62] That treatise provides a general summary, with only a few case citations on “scope of employment” disputes, outside of a long section of disputes between universities and academics. See 1 Melville B. Nimmer and David Nimmer, Nimmer on Copyright §§ 5.03[B][1][b][i]-[ii] (Matthew Bender, Rev. Ed. 2018).

[63] David Loretto, Note, Employee Patents on Computer-Implemented Inventions: The Conundrum of Separate Ownership of Patent and Copyright, 23 Cardozo L. Rev. 705, 711 n.35, 727-28 (2002) (noting possibility of conflicting rights in software covered by copyright and trade secret law and primarily addressing potential conflicts between business-method patents (as patent law stood at that time), employer shop rights, and copyright law). See also Joshua L. Simmons, Inventions Made for Hire, 2 N.Y.U. Intell. Prop. & Ent. L. 1, 46 (2012) (comparing development of work for hire doctrine to incremental development of patent-related concepts like the hired-to-invent doctrine and the shop right doctrine, and arguing that patent law should cease requiring written inventor assignments and instead adopt an umbrella ownership doctrine like work for hire).

[64] Michael D. Birnhack, Who Owns Bratz? The Integration of Copyright and Employment Law, 20 Intell. Prop. Media & Ent. L.J. 95, 152-53, 157-59, 162 (2009) (seeking “the most efficient and fair rule of initial allocation of copyright in works created by authors in the workplace”).

[65] See Catherine L. Fisk, Authors at Work: The Origins of the Work-for-Hire Doctrine, 15 Yale J.L. & Human. 1 (2003) (covering the nineteenth century, the creation of the work for hire doctrine under the 1909 Copyright Act, and modern developments).

[66] For cases where the employer wins on factor one, see Lewis v. Activision Blizzard, Inc., No. C 12-1096 CW, 2013 U.S. Dist. LEXIS 149784, at *10 (N.D. Cal. Oct. 17, 2013) (where employer and former employee clashed over employee’s voice recording used in video games, first factor favored employer where employee handbook listed creation of game content as a job role); Molinelli-Freytes v. Univ. of P.R., No. :09-1655(DRD), 2012 U.S. Dist. LEXIS 143262, at *32-37 (D.P.R. Sept. 30, 2012) (granting summary judgment to employer over faculty member’s proposal for a graduate program; even though employee worked on vacations and holidays, that second factor mattered less where work was clearly of the type employee was hired to perform); Fleurimond v. N.Y. Univ., 876 F. Supp. 2d 190, 201-09 (E.D.N.Y. 2012) (granting summary judgment to employer, which met all three factors as to employee who created graphic design while employed to create such things, where she was expected to perform her job in part at home using her own equipment, and where “no reasonable juror could find that she was not motivated in large part to serve the interests of NYU.”); Rouse v. Walter & Assocs., LLC, 513 F. Supp. 2d 1041, 1056-60 (S.D. Iowa 2007) (applying three-part test; on tangled fact pattern where professors argued that software program created in the course of sponsored university research was not a work for hire, court rejected argument because, among other things, the software development was within the duties they were “hired to perform,” the software “was developed and tested on [university] computers” even if some work was done at home, and professors “were motivated at least in part” to further university research when creating the work); Sterpetti v. E-Brands Acquisition, LLC, No. 6:04-cv-1843-Orl-31DAB, 2006 U.S. Dist. LEXIS 21407, at *20-22 (M.D. Fla. Apr. 20, 2006) (where parties disputed ownership of pasta-making manual, employer prevailed on first factor because it asked the employee to create the manual and provided feedback, showing its creation was part of employee’s job duties); Genzmer v. Pub. Health Tr., 219 F. Supp. 2d 1275, 1281 (S.D. Fla. 2002) (employee hired as a doctor, but in fact his required research “included the drafting of computer programs” and job description was broad enough to cover “the development of the computer program at issue”; in addition, employee’s supervisor gave him “guidance and praise” for the software); Avtec Sys., Inc. v. Peiffer, No. 92-463-A, 1994 WL 791188, at *3-4 (E.D. Va. 1994) (employer won first factor in software case because employee’s job description including writing programs and implementing computer simulation, and because at his employment, employee had generated similar software, and because employer had been able to use the work to “win a contract”), aff’d, No. 94-2364, 1995 WL 541610, at *1 (4th Cir. Sept. 13, 1995); Miller v. CP Chems., Inc., 808 F. Supp. 1238, 1243 (D.S.C. 1992) (employee who wrote software was “not hired primarily for the development of computer programs,” but “the development of the computer programs was at least incidental to his job responsibilities because it was ‘within the ultimate objective of the principal and an act which it is not unlikely that a servant might do.’”) (quoting Restatement (Second) of Agency § 229 cmt. b (Aᴍ. Lᴀᴡ Inst. 1958)).

[67] See Roeslin v. District of Columbia, 921 F. Supp. 793, 798 (D.D.C. 1995) (“It is unfair for the District to now claim that an activity it discouraged – developing the system – was within the scope of plaintiff’s employment.”).

[68] For cases where the employee wins on factor one, see TAP Worldwide, LLC v. Becker, No. CV 10-04903 DMG (JCx), 2010 WL 2757354, at *3-4 (C.D. Cal. July 12, 2010) (export manager of parts manufacturer, which did not develop or sell software, created software to expedite export shipments, usable by any company for a monthly fee; finding in favor of employee when employer moved for preliminary injunction because “[c]reating software was not within the scope of his job and no one else at the company was tasked with anything similar,” employee identified a business need not specific to his employer “but in the business world generally,” and created it “on his own time, not during working hours”; employer did not even learn about software until it terminated the employee); Pavlica v. Behr, 397 F. Supp. 2d 519, 525 (S.D.N.Y. 2005) (finding triable issue of fact for trial where high school teacher created a program to train high school students to conduct scientific research where teacher did so without prompting or direction from school, and where his intent was a model useful for any high school); Moonstruck Design, LLC v. Metz, 02 Civ. 4025 (RWS), 2002 U.S. Dist. LEXIS 14583, at *12-13 (S.D.N.Y. Aug. 12, 2002) (denying motion for preliminary injunction where former employer failed to meet any of the three tests; jewelry designer was not an employee at the time of creation, was hired only in a sales capacity, did not create the design during working hours, and was motivated by desire to honor his wife and other cancer survivors); PFS Dist. Co. v. Raduechel, 332 F. Supp. 2d 1236, 1248 (S.D. Iowa 2004) (employee was truck driver, not a software developer, so this factor favored employee); Martin v. City of Indianapolis, 982 F. Supp. 625, 633-34 (S.D. Ind. 1997) (going a little against the grain of some other cases cited here, finding that employee won on first factor because only a little of his work for employer involved ornamental metal work, but also because he demanded his own fee paid directly to him for artistic work, and his work on sculpture in question differed significantly from his “normal course of work”), aff’d, 192 F.3d 608 (7th Cir. 1999); Roeslin, 921 F. Supp. at 798 (employee hired as labor economist, not software programmer and was discouraged from pursuing project); City of Newark v. Beasley, 883 F. Supp. 3, 8 (D.N.J. 1995) (employee was police officer, but work on classroom training materials to teach young people not to steal cars was unknown to employer, and employee learned nothing from employer that assisted in project, so factor favored employee); Quinn v. City of Detroit, 988 F. Supp. 1044, 1051 (E.D. Mich. 1997) (employee was an attorney, not a programmer, employer never requested the work).

[69] See Quinn, 988 F. Supp. at 1051 (software created at home but later installed at work).

[70] See Fleurimond, 876 F. Supp. 2d at 201-09 (granting summary judgment to employer, which met all three factors as to employee who created graphic design while employed to create such things, where she was expected to perform her job in part at home using her own equipment, and where “no reasonable juror could find that she was not motivated in large part to serve the interests of NYU.”); Rouse, 513 F. Supp. 2d at 1056-57 (applying three-part test; on tangled fact pattern where professors argued that software program created in the course of sponsored university research was not a work for hire, court rejected argument because, among other things, the software development was within the duties they were “hired to perform,” the software “was developed and tested on [university] computers” even if some work was done at home, and professors “were motivated at least in part” to further university research when creating the work); Sterpetti, 2006 U.S. Dist. LEXIS 21407, at *27-28 (employee created pasta-making manual offsite, but at employer’s direction, so employer prevailed on second factor); Genzmer, 219 F. Supp. 2d at 1282 (employee was involved in research and thus would not have been at clinic seeing patients; “it follows that he would not have developed the software” at the employer’s facility).

[71] See Moonstruck Design, 2002 U.S. Dist. LEXIS 14583, at *12-13 (S.D.N.Y. Aug. 12, 2002) (denying motion for preliminary injunction where former employer failed to meet any of the three tests; jewelry designer was not an employee at the time of creation, was hired only in a sales capacity in any event, did not create the design during working hours, and was motivated by desire to honor his wife and other cancer survivors); Raduechel, 332 F. Supp. 2d at 1248 (software work done at home with employee’s own equipment and software licensed to other companies; employee did later modify software for use at employer but factor favored employee); Martin, 982 F. Supp. at 634 (off-site and on own time); Roeslin, 921 F. Supp. at 798 (employee created software in 3000 hours of work at home, fact that he later used it at work immaterial); Beasley, 883 F. Supp. at 8 (police officer created anti-car theft training materials solely off-site, and thus factor favored employee); Avtec, No. 94-2364, 1994 WL 791188, at *4-5 (employee created software at home and with own equipment, even though he later used it at work).

[72] See, e.g., Miller v. CP Chems., Inc., 808 F. Supp. 1238, 1243 (D.S.C. 1992) (“On the other hand, the work was performed during the time period in which he was employed by [employer].”).

[73] See Food Lion, Inc. v. Capital Cities/ABC, Inc., 946 F. Supp. 420, 421-22 (M.D.N.C. 1996) (journalists working as undercover employees made secret tapes of employer’s working conditions, employer tried to claim the tapes as works for hire to suppress them; the court rejecting this stratagem as not meeting any “reasonable interpretation of scope of employment”).

[74] See Lewis v. Activision Blizzard, Inc., No. C 12-1096 CW, 2013 U.S. Dist. LEXIS 149784, at *13-14 (N.D. Cal. Oct. 17, 2013) (employer easily prevailed where employee created sound recording at employer’s request, for employer’s purposes, and under employer’s supervision); Quinn, 988 F. Supp. at 1052; Miller, 808 F. Supp. at 1243-44 (employee’s software was “created to simplify [employee’s] job and to eliminate errors,” employer asked employee to develop similar software for other company products, and all software related specifically to a product made by the employer).

[75] See Avtec Sys., Inc. v. Peiffer, 21 F.3d 568, 572 (4th Cir. 1994) (as to the third scope of employment factor, the employer must show that the employee “was at least ‘appreciably’ motivated by a desire to further its corporate goals[.]”) (citing Restatement (Second) of Agency § 226 cmt. b (Am. Law Inst. 1958)) (“The fact that the predominant motive of the servant is to benefit himself or a third person does not prevent the act from being within the scope of employment. If the purpose of serving the master’s business actuates the servant to any appreciable extent, the master is subject to liability[.]”)).

[76] For cases where the employer wins on factor three, see Fleurimond v. N.Y. Univ., 876 F. Supp. 2d 190, 201-09 (E.D.N.Y. 2012) (granting summary judgment to employer, which met all three factors as to employee who created graphic design while employed to create such things, where she was expected to perform her job in part at home using her own equipment, and where “no reasonable juror could find that she was not motivated in large part to serve the interests of NYU.”); Le v. City of Wilmington, 736 F. Supp. 2d 842, 845-51 (D. Del. 2010) (on summary judgment by employer, municipal employee hired as Information Analyst who created an “Instant Ticketing” program for a city department of licensing and inspections to use in lieu of paper tickets did not own the software, even though he wrote the software “essentially exclusively on his own time, on his own computer at home,” because his purpose was “to facilitate the City’s business of issuing tickets,” and did so at the direction of a supervisor, received input from other employees, and used code previously used for city-owned software); Rouse v. Walter & Assocs., LLC, 513 F. Supp. 2d 1041, 1056-57 (S.D. Iowa 2007) (applying three-part test; on tangled fact pattern where professors argued that software program created in the course of sponsored university research was not a work for hire, court rejected argument because, among other things, the software development was within the duties they were “hired to perform,” the software “was developed and tested on [university] computers” even if some work was done at home, and professors “were motivated at least in part” to further university research when creating the work); Sterpetti v. E-Brands Acquisition, LLC, No. 6:04-cv-1843-Orl-31DAB, 2006 U.S. Dist. LEXIS 21407, at *18-19 (M.D. Fla. Apr. 20, 2006) (employee’s ideas about “gaining personal benefit” from creation of pasta-making manual insufficient where he clearly created manual at employer’s behest, showing at least a partial motivation to server the employer’s interests); Genzmer v. Pub. Health Tr., 219 F. Supp. 2d 1275, 1283 (S.D. Fla. 2002) (employee claimed vague personal reasons, but evidence showed that he “tailored” software to “fit [employer’s] needs”).

[77] For cases where the employee wins on factor three, see Moonstruck Design, LLC v. Metz, 02 Civ. 4025 (RWS), 2002 U.S. Dist. LEXIS 14583, at *12-13 (S.D.N.Y. Aug. 12, 2002) (denying motion for preliminary injunction where former employer failed to meet any of the three tests; jewelry designer was not an employee at the time of creation, was hired only in a sales capacity in any event, did not create the design during working hours, and was motivated by desire to honor his wife and other cancer survivors); Avtec, 21 F.3d at 572 (“appreciably” standard); Avtec Sys., Inc. v. Peiffer, No. 92-463-A, 1994 WL 791188, at *6 (E.D. Va. 1994) (applying test, programmer created work off-site and later used it at work, but his conduct was consistent with a belief that he was the owner, and employer’s bonus payment could have been made “as a reward for [employee’s] willingness to utilize his own property for the benefit of his employer”); PFS Dist. Co. v. Raduechel, 332 F. Supp. 2d 1236, 1248-49 (S.D. Iowa 2004) (where employee had previously licensed software to other companies and then modified it to work with employer’s system, not enough to favor employer on third factor); Roeslin v. District of Columbia, 921 F. Supp. 793, 798 (D.D.C. 1995) (court found evidence that employee was primarily creating software for self-motivation and to create job opportunities for himself, so fact that software also benefited his employee immaterial); City of Newark v. Beasley, 883 F. Supp. 3, 9 (D.N.J. 1995) (police officer intended that many cities, not just his own, might purchase his training materials; even though he believed his employer might be one customer for his work, that was not enough).

[78] See Martin v. City of Indianapolis, 982 F. Supp. 625, 634-35 (S.D. Ind. 1997), aff’d, 192 F.3d 608 (7th Cir. 1999).

[79] See id.

[80] See Cannon Grp., Inc. v. Better Bags, Inc., 250 F. Supp. 2d 893, 901 (S.D. Ohio 2003). This case is atypical, however, as the employer was not a party. The party holding the assignment was the cross-plaintiff in an infringement case, and the cross-defendant was trying to undermine the cross-plaintiff’s copyright registration. There was nothing in the case about whether the employee had been permitted to do the work for the third party while employed, and the court appeared to be moving hastily to shut down the cross-defendant’s strained attempt to defeat the cross-plaintiff’s copyright. The employer was not present to argue its own case for a work for hire. In other words, courts may be less exacting when an accused third party infringer tries to invalidate the copyright on some technicality.

[81] Some courts, of course, find the evidence in conflict, and hold that a trial is necessary. E.g., Koenig v. Dowdy, No. 5:15-CV-00347-RN, 2017 U.S. Dist. LEXIS 163850, at *19 (E.D.N.C. Sept. 28, 2017) (denying motion for summary judgment as to copyright ownership where, among other things, third work for hire factor was subject to disputed evidence over an architectural house plan created by an employee/architect); Bell v. Maloney, No. 1:16-cv-01193-RLY-DML, 2017 U.S. Dist. LEXIS 111867, at *4-5 (S.D. Ind. July 19, 2017) (triable issues of fact on all three factors where attorney took photo that was used on law firm’s website and parties disputed ownership).

[82] Federal courts agree that trade secret causes of action are not copyright-preempted. E.g., GlobeRanger Corp. v. Software AG USA, Inc., 836 F.3d 477 (5th Cir. 2016) (collecting cases from other circuits and joining them). Notably, with the May 2016 enactment of the federal Defend Trade Secret Act, no preemption analysis is necessary as to a cause of action for trade secret misappropriation brought under federal law. See generally Defend Trade Secrets Act of 2016, Pub. L. No. 114-153, 18 U.S.C. §§ 1831-1839, 130 Stat. 376.

[83] For example, in the well-known California case, Cubic Corp. v. Marty, 229 Cal. Rptr. 828, (Ct. App. 1986), the employee’s work product in dispute included a “manuscript describing his invention,” which surely would have been a work for hire under the facts, where the employer easily prevailed under its invention assignment contract. Id. at 830.

[84] E.g., Cal. Lab. Code §§ 2870-72 (West 1979). As noted, however, it remains unclear whether an employer could meet these tests where the employee had no notice of related work by a large, far-flung employer.

[85] PFS Dist. Co. v. Raduechel, 332 F. Supp. 2d 1236, 1240 (S.D. Iowa 2004).

[86] See id. at 1240-43, 1246-47, 1251-53.

[87] See id. at 1247-49.

[88] See id. at 1249. The court spent all of five sentences on the invention assignment claim, and found that “[d]ue to the insufficiency of the evidence to establish the database software as a ‘work for hire,’ or otherwise rightfully owned by plaintiffs, the Court declines to award injunctive relief that ordinarily might be warranted by this claim.”

[89] TAP Worldwide, LLC v. Becker, No. CV 10-04903 DMG (JCx), 2010 WL 2757354, at *1 (C.D. Cal. July 12, 2010).

[90] See id. at *1, 3.

[91] See id. at *4-5.

[92] See Cal. Lab. Code § 2870 (West 1979).

[93] This factual summary is taken from the Fourth Circuit’s first appellate ruling in the case. Avtec Sys., Inc. v. Peiffer, 21 F.3d 568, 569-70 (4th Cir. 1993). See also Avtec Sys., Inc. v. Peiffer, 805 F. Supp. 1312, 1315-17 (E.D. Va. 1992).

[94] Avtec, 21 F.3d at 575. This logic – that copyright ownership precludes a conflicting trade secret ownership claim in the same information – formed the basis for a 2002 ruling in the Eastern District of Texas, where a contractor’s ownership in a software program overrode the hiring company’s claim to trade secrets in the same program. See Alcatel USA, Inc. v. Cisco Sys., Inc., 239 F. Supp. 2d 645, 659 (E.D. Tex. 2002). That case did not involve the work for hire doctrine because the author was an independent contractor, but its holding is consistent with the theme of this essay, that copyright ownership in some instances can conflict with dueling claims under other categories of intellectual property law.

[95] See Avtec Sys., Inc. v. Peiffer, No. 94-463-A, 1994 U.S. Dist. LEXIS 16946, at *11-19, 25-26 (E.D. Va. 1994) (although the employer prevailed on the first scope of employment factor, the employee won on the second and third factors; “Peiffer’s creation of the Orbit Program was too little actuated by a purpose to serve the master.”; rejecting employer’s trade secret claim and awarding no damages on breach of fiduciary duty claim), aff’d, No. 94-2364, 1995 WL 541610, at *1 (4th Cir. Sept. 13, 1995).

[96] Cases in this area appear to be exceedingly rare. In one odd Texas case, a court found that an employer’s state-law claims against a former employee (including trade secret misappropriation, conversion, and unjust enrichment) were preempted by the Copyright Act – a conclusion that would not be reached today at least as to a trade secret claim. Thus, a potential clash between the work for hire doctrine and trade secret ownership through an invention assignment agreement was averted. See Butler v. Continental Airlines, Inc., 31 S.W.3d 642, 651-52 (Tex. App. 2000).

[97] See DDB Techs., L.L.C. v. MLB Advanced Media, L.P., 517 F.3d 1284, 1289-90 (Fed. Cir. 2008) for the exception to this point.

[98] For example, the law of fiduciary duty may require that certain employees provide an “opportunity” to employers. The next essay in this series will focus on that “preparations to compete” context.

Analyzing Aalmuhammed v. Lee in the Context of Entertainment Industry Employment

Analyzing <i>Aalmuhammed v. Lee </i>in the Context of Entertainment Industry Employment

By: Jennifer Yamin*

Download a PDF version of this article here

In Aalmuhammed v. Lee, the Ninth Circuit established a test for determining whether an individual contributor to a work may qualify as a joint author. The test identified three main factors: 1) the author must superintend the work by exercising control; 2) the putative co-authors must make objective manifestations of a shared intent to be co-authors; and 3) the audience appeal of the work must turn on both contributions and the share of each in its success cannot be appraised. Applying these factors, the court concluded that authorship rights could not be granted to a film consultant hired to assist in the creation of the film Malcolm X despite his sizable contributions to the final product.

By analyzing the unique interplay between intellectual property rights and entertainment industry employment law, this Note explores the harmful effects of the Aalmuhammed test on employment and unions across all types of entertainment works. The Note argues that the Ninth Circuit’s test hinders, rather than furthers Congress’s explicit constitutional duty to promote the growth of the arts. In doing so, the test establishes a dangerous precedent that is incompatible with the modern operation of the entertainment industry and paradoxically is detrimental to the very people it intends to protect: creators. The Note concludes that the Aalmuhammed test should no longer serve as the standard courts rely on to determine authorship rights and offers various proposals for reform.



It may take two to tango, but it takes far more than two to make a film, television show, or music video. It is show business, after all. The rolling credits at the end of Denzel Washington’s motion picture Fences or an episode of the hit television show Game of Thrones exemplify the vast number and diverse array of participants involved in the creative process. From makeup artists to background dancers to even the boom mic operator, any set will undoubtedly be filled with various creative contributors. The credits may fail to accurately portray, however, the frequency at which individuals in the entertainment industry step outside of their designated roles. A background dancer may suggest new choreography or a film consultant can rewrite entire scenes, partaking in the collaborative process in non-stereotypical ways and influencing the direction or even the outcome of the work. This fluidity has freed the entertainment industry from having to stick to the script, allowing the creative process to thrive in an unbound, collaborative environment.

While such flexibility has fueled Hollywood for decades, it has also led many entertainment industry participants to litigate claims over the authorship rights to various creations. Recent litigation has explored this exact issue: who is an author of the work for the purposes of copyright law? In this capacity, the law of intellectual property plays a formidable role in the entertainment industry because of its ability to either grant or deny authorship status to creators. Aalmuhammed v. Lee[1] exemplifies the ambiguity surrounding which contributors can walk away with authorship rights. In this case, a film consultant that presented evidence that he independently wrote at least two entire scenes, translated Arabic into English for subtitles, and participated in editing tried to gain copyright to the motion picture, but was unable to do so because he was not found to be a joint author to the work.[2] In coming to this decision, the Ninth Circuit articulated a three-factor test to determine whether an individual qualifies as a joint author to a work.[3] However, as this note will show, this test has proven to be inconsistently applied throughout the courts, leading both to confusion and a lack of direction when providing guidelines for determining authorship.

In this Note, I argue that the joint authorship test established by the Ninth Circuit is problematic, particularly with respect to the entertainment industry. Part I addresses the current copyright law landscape regarding authorship following Aalmuhammed. Part II showcases the recent inconsistent and ambiguous applications of the Aalmuhammed test, specifically of the audience appeal factor, as applied to different types of entertainment content, including screenplays, songs, and music videos. Part III introduces the interplay between intellectual property rights and entertainment industry employment law. Part IV discusses how the Aalmuhammed test is inconsistent with the modern operation of the entertainment industry. It argues that the application of the test yields negative effects on entertainment industry employment and unions. Part V offers resolutions to this problem by discussing various proposals for reform. This Note concludes that the Ninth Circuit’s Aalmuhammed test is not only incompatible with the functions of the entertainment industry, but also detrimental to employment in the industry, further demonstrating the need for reform.

I. Copyright Landscape

A. 1976 Copyright Act

The United States Constitution promulgates the Copyright Clause under Article I, Section 8, Clause 8. The Copyright Clause specifies that one of the powers of Congress is “ to promote the Progress of Science and useful Arts, by securing for limited Times to Authors and Inventors the exclusive Right to their respective Writings and Discoveries.”[4] Notably, included within this delegation of power is the specification that Congress is empowered to provide copyright protection to “ authors.”[5]

The prevailing statute for copyright law in the United States is the 1976 Copyright Act, (“ The Act”).[6] Oddly, the words “ author” and “ authorship” have yet to be defined, despite their presence in the Copyright Clause and importance to the statutory scheme. Joint works, however, are defined as “ a work prepared by two or more authors with the intention that their contributions be merged into inseparable or interdependent parts of a unitary whole.”[7] Because “ authorship” is left undefined, the legislative history surrounding the definition and inclusion of “ joint work” is often used to shed light on the congressional intent behind the term. The House Report describes that the “ touchstone” of the joint work question “ is the intention, at the time the writing is done, that the parts be absorbed or combined into an integrated unit, although the parts themselves may be either ‘inseparable’ (as the case of a novel or painting) or ‘interdependent’ (as in the case of a motion picture, opera, or the words and music of a song).”[8]

The Act lists the different types of works of which authors may claim authorship. Works of authorship include the following categories: literary works, musical works (including any accompanying words), dramatic works (including any accompanying music), pantomimes and choreographic works, pictorial, graphic and sculptural works, motion pictures and other audiovisual works, sound recordings, and architectural works.[9] Under section 101, each of these categories is individually defined.[10] Motion pictures, for example, are defined as “ audiovisual works consisting of a series of related images which, when shown in succession, impart an impression of motion, together with accompanying sounds, if any.”[11]

B. Joint Authorship in the Ninth Circuit: Aalmuhammed v. Lee

Beginning in 1991, Warner Brothers along with Spike Lee and his production companies, entered into a contract to make the film Malcom X, based on the book The Autobiography of Malcom X.[12] Lee served as the director, co-writer, and co-producer of the film. The film’s leading actor, Denzel Washington, played Malcom X. Washington sought out Jefri Aalmuhammed to assist with the role specifically because Aalmuhammed had previously made a film about Malcom X, and was known to be well-informed and knowledgeable about the leader’s life.[13]

Although Aalmuhammed did not have a contract with Warner Brothers or the director, he still wanted to be credited as the co-writer of the film. In his complaint, he cited evidence showing that his involvement with the film was substantive.[14] Aalmuhammed’s evidence revealed he had reviewed the shooting script, suggested extensive script revisions, directed Washington and other actors, “ created at least two entire scenes with new characters, translated Arabic into English subtitles, supplied his own voice for voice-overs, selected the proper prayers and religious practices for the characters, and edited parts of the movie during post production.”[15]

Aalmuhammed subsequently sought a copyright with the U.S. Copyright Office. The Office issued him a “ Certificate of Registration,” but noted there were conflicting previous registrations of the film. On November 17, 1995, he filed suit against Spike Lee, Lee’s production companies, and Warner Brothers alleging a variety of claims: breach of implied contract, quantum meruit, unjust enrichment, and federal and state unfair competition claims. Lee sought declaratory relief and an accounting under the Copyright Act.[16] Most of these claims were dismissed under a motion to dismiss or on summary judgment.[17] In February 2000, the Ninth Circuit heard argument in the case. The court heard the copyright claim, the quantum meruit claim, and the unfair competition claim.[18]

The court first addressed Aalmuhammed’s joint authorship claim. To determine whether the motion picture qualified as a joint work, the court looked to the definition and language of “ joint work” in the Act. It concluded that for a work to qualify as a joint work it must: 1) be a copyrightable work; 2) by two or more “ authors;” 3) the authors must intend their contributions to be merged into inseparable or interdependent parts of a unitary whole; and 4) each author is required “ to make an independently copyrightable contribution” to the disputed work.[19]

The court found that Aalmuhammed had established a genuine issue of fact for three of the four elements of a “ joint work.” Specifically, the court noted that Malcolm X was a copyrightable work and intended to be a unitary whole. Aalmuhammed’s evidence that he “ rewrote several specific passages of dialogue that appeared in the film” and “ wrote scenes relating to Malcolm X’s Hajj pilgrimage that were enacted in the movie” was credited by the court as a copyrightable contribution.[20] Further, all parties involved had the intent that Aalmuhammed’s contributions would be merged into interdependent parts of the movie as a unitary whole.[21] Despite the fact that Aalmuhammed readily met these standards, the court ultimately held he did not qualify as an “ author” for the purposes of the Act, and thus found that Aalmuhammed was not one of at least two authors required to establish a “ joint work.”[22]

The court devoted a significant portion of its decision to the joint authorship question. After noting that “ authorship is not the same thing as making a valuable and copyrightable contribution,” the court analyzed the traditional uses and applications of the word “ author,” concluding that the word “ author,” as used in this context, refers to the “ activity of one person sitting at a desk with a pen and writing something for publication.”[23] The idea and definition of “ author,” the Ninth Circuit notes, is “ relatively easy to apply to a novel” and “ to two people who work together in a fairly traditional pen-and-ink way.”[24] The relatively easy application ends, however, when “ the number of contributors grows and the work itself becomes less the product of one or two individuals who create it without much help.”[25] Reiterating the traditional bases of the word, the court added, “ the word is traditionally used to mean the originator or the person who causes something to come into being, or even the first cause, as when Chaucer refers to the ‘Author of Nature.’”[26] Lastly, the court cited the Gilbert and Sullivan song, “ I Am the Very Model of a Modern Major General,” to show that because Gilbert’s words and Sullivan’s tune are inseparable, the audience can know that both were the creative originators or authors.[27] The court then listed several theories that could help establish the authorship of a film. For example, they noted the author may be “ the producer who raises the money.”[28] Or, under Eisenstein’s theory and the “ auteur” theory, the author may be the editor or the director, respectively.[29] Lastly, they stated that “ traditionally, by analogy to books, the author was regarded as the person who writes the screenplay, but often a movie reflects the work of many screenwriters.”[30]

Turning to case law to shed further light on the discussion of who an author may be, the court cited the Supreme Court’s decision in Burrow-Giles Lithographic Co. v. Sarony.[31] In Burrow-Giles, the Court expanded the concept of authorship to include a photographer who exercised a sufficient degree of control over the subject of his photograph. The Court’s support for this decision included various English authorities and the Founding Fathers of the Constitution.[32] Ultimately, the Ninth Circuit concluded that an “ author” is “ the person to whom the work owes its origin and who superintended the whole work, the ‘master mind.’”[33] Applying this definition to a film, the Aalmuhammed court finds that an author is limited “ to someone at the top of the screen credits, sometimes the producer, sometimes the director, possibly the star, or the screenwriter – someone who has artistic control.”[34] And, lastly, the Aalmuhammed court relied on Thomson v. Larson,[35] a Second Circuit decision in which it was affirmed that Larson, a playwright, did not intend joint-authorship with Thomson, the dramaturg for the Broadway musical Rent. The absence of both decision-making authority and billing as a co-author led the Second Circuit to find that the work was not joint.[36]

Extracting from Burrow-Giles, Thomson, and the Gilbert and Sullivan example, the Ninth Circuit compiled a list of three factors that would constitute the criteria to establish who should be considered an author for the purposes of a joint work (assuming there is no contract stating otherwise): 1) the author must superintend the work by exercising control; 2) the putative co-authors must make objective manifestations of a shared intent to be co-authors; and 3) the audience appeal of the work must turn on both contributions and the share of each in its success cannot be appraised.[37] With respect to the first factor, the court specified that “ this will likely be a person ‘who has actually formed the picture by putting the persons in position, and arrang[ed] the place where the people are to be – the man who is the effective cause of that,’” or “ the inventive or master mind” who “ creates, or gives effect to the idea.”[38] The second factor requires the putative co-authors make objective manifestations of a shared intent to be co-authors. For example, listing both “ Gilbert and Sullivan” as the authors of the song would satisfy this requirement. Objective manifestation—as opposed to subjective manifestation—is required because a fraudulent outcome might result if co-authors were to conceal their true intentions regarding a work’s authorship.[39] For example, an author could communicate to the other “ author” that she intended to be co-authors to the work, only to register for a copyright in her own individual name. The court expresses concern over mutual intent under a subjective intent standard as becoming “ an instrument of fraud.”[40] And under the third factor, the audience appeal of the work must turn on both contributions and “ the share of each in its success cannot be appraised.”[41] Notably, the court specified that control will be the most important factor in many cases. They also qualified that the factors “ cannot be reduced to a rigid formula, because the creative relationships to which they apply vary too much.”[42]

In applying these factors, the court first found that Aalmuhammed did not superintend the work, and instead, Warner Brothers and Lee controlled it. Comparing Aalmuhammed to the dramaturg in Thomson, they found that Aalmuhammed may have made useful recommendations, but the film may not have benefited from them unless Lee chose to accept them.[43] Secondly, they concluded that there were no objective manifestations of intent to be co-authors among any of the parties involved.[44] The court did not address the third factor.

The court grounded its decision to withhold authorship from Aalmuhammed in policy concerns. According to the court, granting authorship to the plaintiff in this scenario would frustrate the constitutional goal to “ promote the progress of Science and useful Arts.” The court feared that the threat of losing sole ownership of the work itself may incentivize authors to insulate themselves throughout the creative process.[45] As a result, the overall decrease in collaboration could thereby impoverish the arts in terms of both quantity and quality.

Moreover, adopting Aalmuhammed’s broader definition of joint authorship would allow many other contributors to deprive the sole authors of proper title. Specifically, “ research assistants, editors, and former spouses, lovers and friends would endanger authors who talked with people about what they were doing, if creative copyrightable contribution were all that authorship required.”[46] From the court’s point of view, the fear of being stripped of absolute title thus becomes even more prevalent, and consequently, the urge to resist collaboration as well.

II. Recent Inconsistent and Ambiguous Applications of the Audience Appeal Factor in Entertainment

The judicial system is no stranger to the issue of joint authorship in the entertainment industry. In applying the Aalmuhammed test, however, lower courts have reached inconsistent results when considering various kinds of entertainment works, such as screenplays,[47] songs,[48] and music videos.[49] In dealing with frequent litigation surrounding the joint authorship issue, courts have turned to the Ninth Circuit’s Aalmuhammed decision for guidance. The third factor of the Aalmuhammed joint work analysis, the audience appeal factor, poses particular issues with respect to entertainment works in that it is applied inconsistently and ambiguously to screenplays, songs, and music videos.[50]

A. Screenplays

In Richlin v. MGM, for example, the same court that crafted the audience appeal factor admitted that “ it is nearly impossible to determine how much of [a] Motion Picture’s audience appeal and success can be attributed to the [Screenplay].”[51] The Ninth Circuit suggested that a film’s success might be attributed to a number of factors other than the screenplay itself, such as the actor’s “ legendary comedic performance,” the composer’s “ memorable score,” or the director’s “ award-winning direction.”[52] Despite acknowledging that there were various creative contributions, none of which could have independently and solely drawn in the audience, the court still found that the audience appeal factor favored the screenwriter plaintiffs seeking copyright of the film, because without their original screenplay, the motion picture would not exist. Ultimately, the analysis surrounding the audience appeal factor proved moot, as the court found that the first two factors weighed heavily in favor of the defendant, Metro-Goldwyn-Mayer Pictures, Inc., thereby outweighing the third factor entirely.[53]

The treatment of the third factor in Richlin is problematic in three ways. First, many motion pictures stem from or are a derivative of a screenplay or book. Without the original writing, the motion picture, in many instances, would cease to exist. The conclusion that the audience appeal and success of a work can be attributed to the screenplay simply because the motion picture was derived from it represents an analytical gap in that it forecloses the recognition of any other valuable creative contributions. By limiting its reasoning to this logic, the Ninth Circuit set a standard in which the audience appeal factor of the Aalmuhammed joint work analysis may favor the party who contributed to the original work and seeks rights as a joint author. This approach circumvents the third factor analysis entirely, where the court would ordinarily need to compartmentalize the various creative contributions, such as those of the writers, actors, composers, and directors, to determine which are responsible for the majority of the audience appeal.

Second, had the court performed a proper audience appeal analysis, they would have been met with a burdensome task that could result in divergent results given the subjective interpretations of motion pictures and what factors contribute to their success. The Ninth Circuit, acknowledging that this analysis would be “ nearly impossible,” instead decided that concluding one could not exist without the other is sufficient.[54]

Third, Richlin represents a case in which the first two factors weighed in favor of the defendants’ claim that the parties were not joint authors. The court left no guidance for a case in which the first two factors weigh in favor of the plaintiff seeking to establish joint authorship. In such a situation, it remains unclear if the third factor would be determinative or continue to carry little or no weight at all. The court also failed to explain whether a more thorough analysis, like the one the Ninth Circuit avoided, should take place. And although the Ninth Circuit qualified in Aalmuhammed that the factors could not be boiled down to an exact formula,[55] the court’s reasoning in Richlin leaves a great deal of ambiguity for litigants and courts to confront in future disputes.

B. Songs

There have been inconsistent ways of analyzing the audience appeal factor in the context of songs as well. For example, in Eli Brown, III v. Michael C. Flowers,[56] the plaintiff and defendant had formed a partnership called Hectic Records. Flowers and Brown recorded hundreds of demo R&B songs together, where Brown, in his role as sound engineer and producer, “ added riffs and beats,” thus “ establishing tempo, ambiance, echo, reverberation, treble, bass, frequency, gain, bandwidth, distortion, and equalizing.”[57] Flowers, a songwriter, then traveled to New York and New Jersey to market the songs he and Brown recorded, including “ I Wish,” a track that Brown had engineered and produced. The song was picked up by singer Carl Thomas, who re-recorded the song and released it on an album that eventually went platinum. The partnership later dissolved, and, according to Brown, Flowers “ applied for and received copyright registrations for these recordings” and “ subsequently licensed these recordings to other artists, who, in turn, made re-recordings.”[58] Brown then filed suit against Flowers under the Copyright Act of 1976, alleging that the recordings were joint works and he thus had copyright in them as well.[59] The Fourth Circuit ultimately affirmed the district court’s dismissal of Brown’s federal copyright claim, finding he did not meet the requisite author requirement.[60]

In an unpublished concurrence and dissent in part, Judge Gregory wrote a separate opinion in which he applied the Aalmuhammed three-factor test and reached a result that differed from the majority. He found that “ because Brown allege[d] that he served as the engineer and producer of the master recordings, [he] conclude[d] that [Brown] had significant decision making authority over the substance and form of the master recordings.”[61] Second, he found that Brown had sufficiently alleged that the parties made objective manifestations of their intent to be co-authors because of Brown’s allegations “ that the parties collaborated over a period of four years in recording demo R&B songs under their Hectic Records partnership label and in attempting to distribute those recordings to third parties under that label.”[62] Lastly, the third factor was met “ because the audience appeal of R&B songs is attributable in large measure to their underlying riffs and beats, which, invariably, are incorporated into those songs by recording engineers and producers.”[63]

Although he found joint authorship for Brown, Judge Gregory extracted from the ambiguity surrounding the Aalmuhammed audience appeal factor in ways that other courts may not. For example, despite attempting to apply an objective approach to the analysis, Judge Gregory generalized and promulgated stereotypes about the specific type of work at issue. In doing so, his analysis ran counter to the purpose of the audience appeal factor. His conclusion that the appeal of R&B songs is attributable in large measure to the songs’ “ underlying riffs and beats,” neglects the subjective analysis owed to creative works of music. For example, the song at issue could depart from the expectation or standard that R&B songs are successful because of those specific features. Judge Gregory does not constitute nor represent the “ audience” and their views of the song. The vocalist on the song, for example, could be the sole reason the audience is drawn to the piece. Thus, while Judge Gregory did recognize Brown’s right, he did so in a way that did not provide for sufficient flexibility in the analysis that such creative works of entertainment are owed.

In Ford v. Ray,[64] a claim for joint authorship was dismissed based on the Aalmuhammed test. In this case, Ford, the plaintiff, collaborated with Ray, the defendant, on an album where Ford “ independently created a handful of drum tracks, called ‘beats,’ that he thought would make a good foundation for hip hop songs” as well as “ provided ‘scratching’ for the chorus of the song and a solo.”[65] Ray’s role was to choose which of the beats would be used for Sir-Mix-a-Lot’s Grammy award-winning song, “ Baby Got Back.” Ray, without Ford’s knowledge, filed copyright registrations for some of the songs they worked on together, identifying himself as the sole author.[66] Ford subsequently filed suit alleging the works were “ joint works” and that he was the co-author.

In employing the Aalmuhammed three-factor test, the United States District Court for the Western District of Washington found that the first two factors – that the author must superintend the work by exercising control and the putative co-authors must make objective manifestations of a shared intent to be co-authors – weighed in favor of Ray, and thus concluded that Ford was not a co-author.[67] In acknowledging the audience appeal factor, the court stated that Ford “ fail[ed] to allege facts from which one could conclude that his contributions account for the appeal of ‘Baby Got Back.’”[68] The court sought more information such as “ how much of the music was [Ford’s] creation and how much was edited, programmed, and altered by [Ray].”[69] They found the fact that Ray used one of Ford’s “ drum tracks ‘as the basis for the song’ and incorporated [Ford’s] scratching” insufficient.[70] Relying on the complaint, the court concluded that Ford had made no contribution to the lyrics nor did he have any control over the music after handing it over to Ray.[71] In coming to this conclusion, the court acknowledged that the audience appeal factor remains “ something of an unknown” to the analysis and granted the dismissal based solely on the first two factors.[72]

Brown and Ford are thus paradigmatic of how the Aalmuhammed test, particularly the audience appeal factor, has been interpreted, analyzed, and implemented inconsistently when applied to songs.

C. Music Videos

Litigation surrounding music video joint authorship issues further exemplifies the ambiguous application of the audience appeal factor with respect to various kinds of entertainment works. In Morrill v. The Smashing Pumpkins,[73] Morrill was a member of the band The Marked. The band produced a music video titled “ Video Marked” which had been shown at various clubs where The Marked performed in order to promote their group.[74] Defendant Corgan, one of the members of The Marked and the sole permanent member of The Smashing Pumpkins, was a member of The Marked at the time the music video was made, but later left the group. Years later, defendants Corgan, The Smashing Pumpkins, and Virgin Records America released a video titled “ Vieuphoria” which featured short clips of images from “ Video Marked.”[75] Morrill brought suit against the defendants, alleging he was the sole owner of the copyright for the music video. The United States District Court for the Central District of California found that the first two factors favored a finding of a joint work.[76] With respect to the audience appeal factor, the court concluded that the audience appeal can be attributed to both parties because the appeal of “ Video Marked” “ was most likely based on the audience’s ability to view images of a younger Corgan. This is suggested by the packaging for ‘Vieuphoria,’ which advertises super secret, super special extra stuff shot by the band.”[77] The court focused on Corgan’s appearance as fulfilling the audience appeal requirement which ostensibly “ rests both on the video’s visual aspects and on the composition and performance of the music.”[78]

In Eagle Rock Entertainment, Inc. v. Coming Home Productions, Inc.,[79] the Central District of California faced an ownership dispute over “ Guns, God and Government,” a live concert video album filmed during shock rock musician Marilyn Manson’s controversial 1999-2000 world tour.[80] Both parties claimed to have produced the video by contributing to the editing and production of it. The court found the parties to be joint authors by analyzing the audience appeal factor and relying on Morrill. Unlike in Morrill, however, the court did not specifically address which aspects contributed to the visual appeal or the composition and performance of the music. Thus, the court seemed to employ the same standard to measure audience appeal differently in that it specifically sought out and emphasized an extra element in one music video, but did not require or address it in the other.

III. The Interplay Between Intellectual Property Rights and Entertainment Industry Employment Law

The Aalmuhammed test proves significant even outside of the courtroom, where its application has a direct impact on the livelihoods of those who participate in the entertainment industry. Recent examples in pop culture reflect the close link between intellectual property rights and employment. Take Steven Jan Vander Ark, for example, the Harry Potter fanatic who sought to publish an encyclopedia[81] featuring the many characters, settings, and overall magic of J.K. Rowling’s popular book series.[82] Rowling and Warner Brothers Entertainment sued Vander Ark seeking to enjoin the publication of the book alleging it infringed their copyrights given the substantially similar, if not equivalent, nature of the content.[83] While Rowling walked away victorious, Vander Ark was left sobbing on the stand stating that Rowling’s public denunciation of his book “ had ostracized him from the ‘Harry Potter community.’”[84] Why was Rowling, a prominent, successful author of a world-famous series, so keen on stopping one author—or more properly labeled, a fan—from producing a Harry Potter encyclopedia? And why was Vander Ark left so concerned about his reputation among other Harry Potter fans?

Intellectual property rights hold so much force in the realm of entertainment employment that even renowned authors like J.K. Rowling are adamant about protecting them. As of November 2018, the U.S. media and entertainment industry is a $735 billion market, representing a third of the global industry.[85] In 2017, the arts, entertainment, and recreation industry employed 2,370,160 people.[86] Duncan Crabtree-Ireland notes that “ [a]lmost all workers in the entertainment industry . . . are dependent on intellectual property law (and particularly copyright) for their livelihoods.”[87] Overall, the interplay between intellectual property and employment is a unique one in entertainment, where intellectual property rights serve as a policing mechanism to ensure that parties are rewarded for original works of authorship by establishing standards and thresholds for artists to qualify for the benefits of employment.

Despite the significant effect of intellectual property rights on employment opportunities, practices within the entertainment industry often make it difficult to determine the appropriate allocation of rights. First, the industry is inherently filled with intangible content that without proper regulation could be subject to taking.[88] Furthermore, technological developments have augmented the means through which audiences consume content, and facilitated the dissemination of entertainment content throughout the consumer marketplace. Cynthia Baron explains this phenomenon, noting how the “ increasing convergence of film and new media ‘has simultaneously increased the availability of film and turned it into part of a data stream where images become information that is simply passing through.’”[89] In this capacity, “ film has also become part of the flow,” blurring and stretching the origins of the content farther away from the original artist or author. [90] As a result, it becomes a challenge to pinpoint the original source or author of a work, leaving the claimant to intellectual property rights with a daunting task. Justice Scalia voiced similar concerns in Dastar Corporation v. Twentieth Century Fox Film Corporation,[91] stating “ without a copyrighted work as the basepoint, the word ‘origin’ has no discernable limits” and “ figuring out who is in the line of ‘origin’” would prove difficult.[92] The high transaction costs for identifying and crediting sources may serve to decrease, or even paralyze, creative cultural production.

Second, the length and pace of projects in entertainment production make it difficult to monitor the contributions and rights of those who work on entertainment projects. Employment in Hollywood’s film industry, in particular, is project-based so that employers can “ quickly assemble personnel with highly specialized skills for a short period of time. Producers have no incentive to offer long-term contracts because informationally complex jobs are difficult to monitor.”[93] In hiring for short-term projects:

[T]he entertainment industries exist on ideas turned into deals. When an idea is ‘hot,’ immediate action is desired. Parties rush to agree, and, in the process, desire at times outraces common sense. The ‘deal,’ as it turns out, is strictly verbal, or there are scattered memos but no single, final, formal written agreement. If the production proceeds as envisioned . . . [t]here is no problem because the idea becomes a deal that produces a success, and everyone is happy.[94]

As a result of the quick pace of deal-making, oral contracts have replaced written agreements as the primary means of memorializing deals. The industry reliance on oral contracts—often referred to as the “ handshake deal”[95]—reflects the priority of completing projects on time. This priority, in turn, comes at the expense of fleshing out the full details of an employment agreement, which can suppress workers’ efforts to obtain credit for their contribution. If two parties orally and mutually agree to certain terms of a relationship or a course of conduct, the agreement becomes enforceable regardless of whether or not a written agreement was signed.[96] Main Line Pictures, Inc. v. Basinger,[97] one of the pivotal cases involving oral contracts in Hollywood, epitomizes the ubiquity and enforceability of oral contracts in the film industry.

In 1990, Main Line Pictures asked actress Kim Basinger to star in the movie “ Boxing Helena.” On several occasions, Basinger orally agreed to star in the film. A year and a half later, Main Line learned through rumors that Basinger no longer intended to act in the film as she had agreed to do. Main Line subsequently filed a complaint alleging breach of an oral and written contract.[98] The jury concluded that Basinger had breached her oral contract with Main Line and awarded them damages.[99] On appeal, the California Court of Appeals took note of and gave credence to the practice of oral contracts in entertainment, explaining “ because timing is critical, film industry contracts are frequently oral agreements based on unsigned ‘deal memos.’”[100] Even though this case was later reversed on a technicality and settled, it reiterates the control and influence of oral contracts in the industry.

Another common practice unique to the film industry is what Mihaela Mihailova has called “ invisible labor,” where many contributors or collaborators to a film often go unrecognized.[101] She notes that “ Hollywood labor history is marred by a host of such obfuscations,”[102] pointing to unrecognized contributions in Darren Aronofsky’s film Black Swan as exemplifying such an omission of credit. In Black Swan, actress Natalie Portman was largely recognized as the star of the film, displaying professional ballet movements on screen. In reality, however, ballet dancer Sarah Lane was the one who performed most of the challenging dance scenes. With the help of post-production technology, Lane’s face was simply replaced with Portman’s. Neither Lane nor the animation team were credited in the marketing or promotional materials, and instead Portman walked away with the accolades, winning an Academy Award.[103] Mihailova sums up this practice as a “ policy of deliberately suppressing any public acknowledgment of the invisible labor that facilitates the star’s performance.”[104]

Entertainment industry employees are heavily reliant on the recognition of intellectual property rights for employment opportunities. Industry practices, however, make it difficult to determine who can claim those rights. As will be discussed further, authorship can directly affect the salaries, royalties, and overall benefits associated with recognition. In addition to the numerous barriers these participants have to overcome, such as the implications of intangible property, the dominance of oral contracts, or the “ invisible labor” theory, the Aalmuhammed test only exacerbates the difficulties creators in entertainment must face.

IV. Why the Aalmuhammed Test is Problematic for the Entertainment Industry

A. The Aalmuhammed test is incompatible with the operation of the modern Entertainment Industry

The creative process is anything but streamlined and Hollywood’s productions are no exception. While certain structures and processes exist as guidelines, deviation occurs far more often than in other industries. Many factors inherent to the filmmaking process lead to situations in which contributors take on roles outside of their designated responsibilities, such that even in the case of a formal written contract, departures from official roles are frequently made. Thus, while the credits to a film may describe the job titles that participants in the creative process held, they should by no means convey the message that the producer, for example, stuck solely to the responsibilities with which a producer is conventionally associated. The overall creative process behind filmmaking involves many people who played a role, small or large, in the creation of the film.

The three-factor test articulated in Aalmuhammed to determine joint authorship, however, does not adhere to the highly collaborative nature of the filmmaking process, nor to the widespread custom of creative collaboration in entertainment in general. One of the underlying reasons the test is incompatible with the contemporary practices of entertainment is that it is extracted from antiquated sources and thus has proven incongruous in the modern industry. The Ninth Circuit heavily rooted its analysis of “ author” in tradition, citing cases from 1884, Eisenstein, Chaucer, and the Founding Fathers.[105] Instead, the Court should have drawn from these sources with greater generality to maintain an appropriate balance between tradition and innovation. For example, the court could have taken the idea of the writer as the “ person behind the pen”[106] and applied it more liberally to the context at issue. While Aalmuhammed may not have been the physical person holding the pen and writing the screenplay, he did write entire scenes as well as much of the dialogue that was incorporated into the final product.

Scholars also support the argument that the Aalmuhammed test diverges from the Act’s original intent and, as a result, is inconsistent with the needs and goals of the entertainment industry. Professors Shyamrkishna Balganesh, Justin Hughes, Peter Menell, and David Nimmer argue that “ interpreting the statute so narrowly ignores the careful study and negotiation that went into crafting the Section 101 definitions.”[107] The professors assert that the definition of “ joint work” included in Section 101 “ recognizes a wide range of collaborative working arrangements by requiring only that the authors collaborated with each other, or if each of the authors prepared his or her contribution with the knowledge and intention that it would be merged with the contributions of other authors.”[108] Further, the use of the mastermind standard could be interpreted as an evasive strategy in that it could be used for the sake of “ administrative convenience or to avoid unjustified windfalls, not out of fidelity to legislative intent.”[109] They further note:

Such an interpretation misses the broad and open-ended recognition of collaborative creativity that Congress intended. Undue emphasis on singular control (“ the . . . mastermind”) is antithetical to the very nature of joint authorship, which is an intrinsically collaborative exercise. The nature of a collaborative enterprise is such that at times different authors will exercise more control than the others over the work. To require a contributor to exercise equal “ inventive” control in order to be a joint author is therefore unrealistic. Second, under the control standard, it is impossible for contributing authors to know in advance whether they are exercising sufficient control over the unitary work while making their individual contributions. The joint authorship doctrine thereby becomes unpredictable, defeating the “ paramount goal” of the 1976 Act.[110]

The Ninth Circuit also mistakenly assumed that the factors apply neutrally to all subject matter. The reliance on the Gilbert and Sullivan example assumes that a songwriter and film consultant have made equal contributions throughout the creation of the work and to the final product in identical ways. The creative process behind the song that Gilbert and Sullivan wrote, however, is extremely simplified and not representative of the creation of a motion picture.[111] Similarly, the court relied on Thomson and assumed that the process of creating a Broadway musical mimics that of a motion picture.[112] Another assumption undergirding the Ninth Circuit’s opinion is that the creative process for one motion picture is the same for all other motion pictures.[113] The function and operation of creative practices within the industry are far from uniform, and to assume so was a dangerous judicial mistake.

The test is also incompatible with the “ above the line” versus “ below the line” distinction in entertainment, which plays a primary role in monitoring the types of relationships formed during the creative process.[114] Those in Hollywood who are “ above the line” are recognized as the creative talent or financial supporters of a project.[115] The people “ below the line,” on the other hand, hold the technical and behind-the-scenes positions.[116] In particular, the audience appeal factor of the test clashes with the first factor because the audience appeal of a film, more often than not, is owed to the “ above the line” personnel such as the A-list stars like Meryl Streep or Emma Stone, who “ would not qualify under the court’s ‘inventive or master mind’ test.”[117] Instead, the director of the film is usually credited and recognized as the mastermind of the film. If these famous participants are unable to secure copyright interests in a work under such a test, it is further unlikely that a person who is “ below the line” and part of the “ invisible labor” that goes into a film, would either. Despite the frequency with which the Aalmuhammed test is employed by numerous courts when discussing various entertainment works, the test has proven inapplicable to certain categories within the modern entertainment industry.

B. The Aalmuhammed test negatively affects entertainment industry employment

The inability of the Aalmuhammed test to adapt to the entertainment industry’s practices yields unintended consequences on employment. Generally, joint authorship is an important intellectual property right that may play a part in establishing whether or not an individual can obtain future employment opportunities. If an individual meets the requirements to qualify as a joint author, then she will be able to register and obtain a copyright in the work. This further enables her to receive future benefits associated with the copyright, such as the opportunity to be involved with any sequel or a work tangentially related to the original work. This will then increase her exposure, and therefore her potential to garner greater recognition or reputation in the industry. Being known as a joint author may also signify to other creators her potential as a collaborator for future projects, thereby opening the door to future employment opportunities. Without recognition of joint authorship, a co-author may be deprived of the employment opportunities arising from or incidental to copyright ownership.

There are several more specific ways in which the Aalmuhammed test is unable to best serve the interests of those seeking employment and those currently employed in the entertainment industry. First, the audience appeal factor disincentivizes participants from contributing to a work because it does not adhere to the fluidity of the creative process. This is evidenced in Ford, where the plaintiff did not provide enough evidence upfront that his contributions throughout the creative process impacted the audience appeal of the work. Creators may therefore take additional precautions external to the creative process itself in order to abide by the types of inquiries a court may make upon reviewing whether an artist contributed enough to qualify for authorship rights. The creative process is a lengthy one, involving a broad spectrum of contribution ranging from the first spark of creativity to the final product. Drafts, shot suggestions, lighting changes, deletions, and many other steps that blur the process frequently occur. Requiring or expecting parties involved in the music creative process, for example, to memorialize agreements and document material information at all stages of production is a large burden, especially given the prevalence of oral contracts in the entertainment industry. The difficulty of proving specific contributions after the fact arguably discourages collaboration between creators, and, as a consequence, may adversely affect the labor market within the industry as well.

Second, the Aalmuhammed test reinforces the monopolization of the creative process evident in Hollywood and in doing so, inhibits the opportunities for employment growth in entertainment. A recent deal between the Walt Disney Company and 21st Century Fox, in which the former acquired the latter’s assets for $52.4 billion epitomizes how the “ Hollywood heavyweights” continue to maintain and exert the upper hand in the broader entertainment industry.[118] By expanding its empire even further, Disney undoubtedly will gain a strong foothold over 21st Century Fox’s operations, employees included, and even “ more control over some of the content that fuels [Disney’s] business.”[119]

Even for smaller scale projects in entertainment, employers normally will be able to fulfill positions needed for a music video or a film because prospective creators are eager for their “ big break” in Hollywood and would not normally pass down a shot at fame. Those with established reputations such as Steven Spielberg and Quentin Tarantino, then, are able to maintain the upper hand in terms of bargaining power (and the subsequent copyrights), making it much more difficult for individuals to make a name for themselves. Different hiring structures or novel approaches to the distribution of responsibility will not be implemented as a result of the cemented status quo and monopoly over employment. Opportunities for employment growth and diversification are thus decreased in ways that harm those trying to break into the industry.

Third, as a consequence of increased litigation over co-authorship claims, courts have experimented with different approaches to applying the Aalmuhammed standard, resulting in inconsistent results and ambiguity for jobseekers and employers alike. As a result, proper fair notice is not offered to those who would benefit from awareness of the requirements of co-authorship under the Aalmuhammed test. This lack of prospectivity with respect to authorship rights will have a detrimental effect on entertainment industry employment because those who are hiring will not know what protocol or processes to put in place regarding the allocation of degree of control and distribution of responsibilities. That courts may or may not accord proper weight to the audience appeal factor leaves entertainment industry employers with little guidance for dealing with the co-authorship issue and avoiding potential lawsuits over joint authorship claims.

Similarly, creators or authors will lack the necessary guidance for understanding whether or not they have to direct their contributions in a way that will affect the audience appeal of the work. One major incentive a participant to a project may have in agreeing to provide services is the potential recognition he or she could gain as a result of being listed as a co-author to a work. The incentive may be lost among the confusion caused by the unclear rules and judicially constructed obstacles blocking the path to legal recognition of joint authorship. Both parties involved may then lose the incentives needed for a successful employment relationship, thus decreasing the potential for collaborative creation.

Fourth, a false impression has been created that any standard other than the Aalmuhammed test will result in the overbroad dissemination of copyrights. This, in turn, makes those with more bargaining power overprotected while they employ others. In Garcia v. Google, for example, the Ninth Circuit relied on Aalmuhammed when addressing Google’s argument that an actress hired to say a few lines in a film posted on YouTube did not qualify as a joint author to the work.[120] The court cited Aalmuhammed for the proposition that a work cannot be defined based on “ some minimal level of creativity or originality” because such a definition “ would be too broad and indeterminate to be useful.”[121] Further, the court noted that “ its animating concern” in Aalmuhammed was that such a definition of “ ‘work’ would fragment copyright protection for the unitary film . . . into many little pieces” where anyone might qualify as an author.[122] The Google court feared this would make “ Swiss cheese of copyrights.”[123] The court similarly stated that “ treating every acting performance as an independent work would not only be a logistical and financial nightmare, it would turn cast of thousands into a new mantra: copyright of thousands.”[124] The Ninth Circuit took a risk-averse approach, fearing that any alternative to the Aalmuhammed test may open the floodgates to copyright claims of joint authorship by every cast and crew member that worked on a film. By establishing Aalmuhammed as the only alternative to a world in which joint authorship claims run amok, the desire to keep Aalmuhammed as the standard is increased, as are concurrently, the limits on employment.

C. The Aalmuhammed test has a negative effect on entertainment unions, which, in turn, further harms industry workers’ employment opportunities

The Aalmuhammed test may also yield detrimental effects on employment by indirectly diluting the efficacy of entertainment unions to adequately represent workers. Specifically, the inability of the test to properly identify rights may lead to union strikes which, as has happened in the past, could cause networks to restructure their employment processes such that opportunities for post-production creation will be limited.

One of the unions in the entertainment industry that may be negatively affected by this phenomenon is the Writers Guild of America West (WGAW). The WGAW “ is a labor union composed of the thousands of writers who write the television shows, movies, news programs, documentaries, animation, videogames and new media content that keeps audiences constantly entertained and informed.”[125] The Guild covers the rights of writers with respect to derivative works, specifically “ minimums, residuals, credits, pension and health contributions, separated rights, and more.”[126] Strikes are an effective union bargaining tactic. Normally, union members strike in an effort to convince or coerce employers to change their employment practices and conditions. Common wisdom may suggest that the potential of the Aalmuhammed test to strip authors of the rights and recognition they deserve might encourage members of the WGAW, for example, to strike. However, striking in response to being denied recognition rights can harm union members’ prospects for future employment.

The 2007 Writers Guild of America (WGA) strike against ABC Studios exemplifies the potential negative effects of a labor strike on the writers. Writers for the hit television show Lost went on strike in order to receive better recognition and compensation for their creations of derivative digital content tied to the original show.[127] During this time, innovations in technology had led to the proliferation of derivative works, including “ blogs, alternate reality games (ARGs), and mobisodes[128] to promote the series.”[129] Recognizing the rights issues that arose from the growth of these markets, the WGA increased its support for enterprising authors or writers who sought to challenge the traditional studio employment model.[130] For example, the WGA website stated, “ [i]n the age of YouTube, Hulu, Crackle, and MyDamnChannel, new media outlets and digital technologies provide writers increased opportunities to become true creative entrepreneurs, armed with the tools and distribution channels necessary to connect directly with audiences – and often without studio/network intervention – like never before.”[131] While the WGA was increasing its support for such “ entrepreneurial” writers, Denise Mann points out that following the strike:

[R]ather than embrace the mass collaborative approach to television explored during the Lost moment, the ‘big three’ networks appear to have bolstered their traditional bureaucratic fortresses to maintain singular control over all aspects of the broadcast business. In particular, ABC fired several of the principal executives involved in the Lost franchise and reabsorbed many of the functions that the Lost writers had performed in collaboration with the network’s programming, marketing, licensing, and merchandising divisions. Since 2011, the big three have reverted to more conventional programming choices (reality shows, sit-coms, episodic dramas), reasserted control over their licensed properties (computer games, novels, board games, and the like), and expanded their in-house digital marketing divisions to create and manage digital promotions tied to their series — all, it seems, in an effort to maintain stricter controls over their industry in the post-strike environment.[132]

Mann further observes that, since the strike, studios and networks have turned to low-cost in-house labor in what appears to be an effort to escape the demands of writers seeking greater recognition and remuneration for their contributions.[133] She concludes that powerful networks like ABC have displayed an “ unwillingness to collaborate with creative partners to the degree seen during the Lost moment.”[134] More troublesome, however, is that “ all of the networks have shown a reluctance to grant additional power to showrunners who wish to engage fans by embracing social media as a viable component of the television experience in the digital age.”[135] She describes that the networks are hesitant about working with outside partners, such as “ WGA-represented transmedia czars, PGA-represented transmedia producers, super-fans who wish to contribute to the television experience, or the type of creative entrepreneurs that YouTube is hiring for its 100-channel partnerships.”[136] Overall, Mann noticed a trend of studios and networks trying to minimize the threat of collaboration by “ taking a giant step backward toward their analog past.”[137]

The treatment of the producers for the hit television show Heroes further demonstrates how networks have responded to the post-production creative boom in unfavorable ways. In 2007, the show’s producers proposed to network executives the formation of a “ transmedia team” to be tasked with managing the continuity of any extensions of the show’s narrative as it expanded beyond its traditional television medium and format through the creation of bonus material made available on other media platforms. The responsibilities of this team would be divided into three main roles: “ merchandising,” “ coordinat[ing] all narrative mobilizations of the property across comics, the Internet, and the like,” and “ work[ing] with the stars of the series to secure their participation in promotions and content made for these new media.”[138] However, the network ultimately dismissed the proposed model for maintaining narrative continuity and fidelity. This is a paradigm example of what Derek Johnson classifies as the “ contradiction between singular authorship and the decentralized creativity of networked production cultures.”[139] Regrettably, it appears that participants who seek alternative means of employment are only met by rejection.

Operating as an obstacle to the recognition of judicial rights and providing limited guidance on creators’ legal rights, the Aalmuhammed test arguably contributes to a growing dissatisfaction among entertainment industry unions. The outcome has been debilitating in certain instances. Although some minimums or residuals may be established in the short-term, networks may put up more barriers in response to such proposed changes in the long-term, as exemplified by the Lost strike and the doomed Heroes proposal. Given the weaker bargaining position of union members, big networks are well-placed to structure their operations in a manner that promotes their interests, much to the detriment of entertainment creators. Union mobilization would be disrupted and potentially paralyzed if networks take advantage of the Aalmuhammed test protections. Employee creators would struggle to avail themselves of joint authorship status under the three-factor Aalmuhammed test, especially in the emerging context of derivative supplemental and promotional content designed to “ keep the story alive” as it migrates across consumption platforms. Creators would be hard-pressed to show that they contributed to the original story, that all relevant parties manifested mutual intent to be co-authors, and that they contributed to audience appeal.

Writers are also directly burdened by the adverse effects the Aalmuhammed test has on entertainment unions. Given the monopolization of the creative process, writers may struggle to obtain a writing credit for projects they worked on despite their sizable contributions. The WGA has a system in which “ any writer whose work represents a contribution of more than 33% of a screenplay shall be entitled to screenplay credit. One exception exists for original screenplays in which any subsequent writer or writing team must contribute 50% to the final screenplay.”[140] In order to determine whether a writer meets this threshold, an arbiter is brought in to consider various elements of their contribution to the work at issue, including dramatic construction, original and different scenes, characterization or character relationships, and dialogue.[141] Each work is looked at independently and there are no formulas an arbiter must use. The WGA, in contrast to the Aalmuhammed test, approaches the authorship analysis with much needed flexibility.

In Hollywood, receiving a writing credit signifies the writer’s accomplishment and can bolster her reputation. A writing credit is an important touchstone of the writer’s professional development because as “ writers move from project to project, a career is created as people move from credit to credit.”[142] Further, a writing credit “ facilitates the assessment of talent in a high-velocity labor market” and provides “ residuals . . . [to] . . . compensate writers during periods of slack employment, thus keeping their human capital in the industry.”[143] One way in which a residual provides better compensation to writers is through the allocation of a credit bonus, which is “ a provision of an individual hiring contract stating that the writer will receive a bonus if the writer is determined to get screen credit.”[144] By minimizing the writer’s ability to legally establish co-authorship, the Aalmuhammed test may have adverse economic repercussions on writers by minimizing their ability to gain writing credit and its associated benefits.

Overall, the Aalmuhammed test-or-nothing approach creates obstacles to joint authorship status and future employment opportunities for creative contributors.

V. Proposals for Reform

That brings the analysis to the next question: what should be the standard by which joint authorship is determined?

Professor Balganesh et al. recommend implementing a test that analyzes the contributions made and the “ mutual intent among the collaborators that they be joint authors,” rather than focusing on the control element of the work.[145] Such inquiry as to mutual intent will only arise when the “ parties have not expressly contracted ownership.”[146] This approach would allow “ all principal creative collaborators in the production of a motion picture, sound recording, or other collaborative work” to “ qualify as joint authors”.[147] While the professors qualify that this would not result in every single contributor being granted joint authorship, this standard remains problematic with respect to the entertainment industry. For example, agreements to take on certain roles often do not go as far as establishing a mutual intent to be joint authors given the time constraints of production on set and the prevalence of oral contracts.

Professor Christopher Jon Sprigman proposes another systematic approach that imposes a default rule that parties may contract around to reach the idealized authorship relationship.[148] In practice, a default rule would establish which party would automatically be granted joint authorship status. Sprigman further suggests making the default rule “ painful” so that the parties, in objecting to it, will have to provide information about which contributors they would prefer to credit with joint authorship status.[149] Parties have the ability to opt-out of the default rule and can negotiate against the background of clear rules. While this proposal offers an organized, approachable solution, it may not pair well with the elasticity and flexibility of the creative process. Specifically, there may not be time to engage in the contracting, nor would there be widespread knowledge of a default rule one would need to contract out of in the first place. Further, it places the parties in a situation where a negotiation, or at least a discussion, would have to take place from the outset. This may disincentivize collaboration, as creators may wish to avoid the trouble and cost of bargaining over authorship ex ante.

Another area of the law that has been proffered as a solution to the authorship question is the work made for hire doctrine. The Copyright Act defines a work made for hire as “ a work prepared by an employee within the scope of his or her employment” or a work that is commissioned specifically to be used in a variety of listed works, including, in most relevant part, “ as a part of a motion picture,” with the parties “ expressly agree[ing] in a written instrument signed by them that the work shall be considered a work made for hire.”[150] In Community for Creative Non-Violence v. Reid, the Supreme Court delineated a list of factors that can be used to determine whether or not a work qualifies as a work made for hire by identifying the status of the person hired.[151] Hiring status becomes determinative to the extent that the copyright lies with the employer if the creative contributor is an employee or independent contractor with the requisite signed written agreement. To determine the employment status of a hired party, the Supreme Court looked to the following factors:

the hiring party’s right to control the manner and means by which the product is accomplished, the skill required, the source of the instrumentalities and tools, the location of the work, the duration of the relationship between the parties, whether the hiring party has the right to assign additional projects to the hired party, the extent of the hired party’s discretion over when and how long to work, the method of payment, the hired party’s role in hiring and paying assistants, whether the work is part of the regular business of the hiring party, whether the hiring party is in business, the provision of employee benefits, and the tax treatment of the hired party.[152]

The proposed solution, then, is to have the employers tailor the course of employment to fit the Reid factors such that the person hired would affirmatively qualify as an employee. Employers would arguably use this strategy when they could anticipate a potential copyright claim. As evidenced by the fluctuations inherent to the entertainment industry in both hiring and throughout the creative process itself, predicting copyright claims or even framing the employment relationship to adhere to the Reid factors is, for the most part, too difficult to execute.

While more compatible with the functions and operations of the modern entertainment industry, a case-by-case analysis lacks the guidance courts may seek when addressing such joint authorship issues. Stripping courts of judicial guidance on the one hand, and rote application of an obsolete and poorly designed three-part test on the other, are both not ideal. Instead, a middle-ground approach should be adopted in order to provide the courts with necessary assessment tools and background knowledge, while still allowing them the freedom to embrace the fluidity they may encounter in the various fact patterns presented.

First, guidelines should be tailored to the subject or content at issue, whether it takes the form of films, television shows, musicals, music videos, or the like. Tests with distinct factors may be developed for the various kinds of content, as long as they allow for the flexibility inherent to the entertainment industry and permit creators to freely roam within its carefully crafted boundaries. For example, a test for motion pictures could account for creative contributions by film consultants by analyzing the quantum of control they exercised over the film. Under such a test, a court may determine that if Aalmuhammed had rewritten or had a significant impact on more than one half of the total scenes, then he should have been awarded joint authorship status. Musical works, too, should be analyzed in ways that account for more than stereotypical elements that are considered by a court to make a “ rap” song popular or a “ country” song identifiable. Instead, an objective approach should be implemented whereby the stereotypes that Judge Gregory had relied on in Flowers are removed.

Second, plaintiffs or defendants may introduce as evidence market studies that describe which aspects of an entertainment work served as most attractive to its audience members. By relying on what consumers actually think, this may reduce the potential for mistaken attributions of ownership and disputes over whose contributions made the biggest impact with respect to revenue or audience appeal. Market studies of this kind can also assuage the concerns of judges who fear the theoretical onslaught of endless co-authorship claims. This middle-ground approach will thus avoid any slippery slope, or as Google’s attorneys would like to say, “ the Swiss cheese of copyrights.”[153]

Another player that can improve the current framework surrounding joint authorship is Congress. Congress can amend the Copyright Act in ways that provide guidance not just to judges, but also to the creators themselves. For example, the Section 101 definition of “ joint work,” as well as the Copyright Act in general, currently lack a definition of “ author.” Congress can narrowly revise the Copyright Act to account for idiosyncratic works to which there are many contributors, by qualifying that “ an author, particularly in creative works in which many different collaborators contribute, may not necessarily fit the ideal or traditional characteristics one would normally associate with the title.” Such a revision would clarify overbroad or vague language, while also loosening the standards to which collaborators may have felt previously bound. Such guidance could function as the proper notice collaborators should be able to rely on, thus diminishing any reluctance on the part of would-be collaborators to enter into a collaborative creative relationship, reducing the likelihood of authorship battles erupting, and ultimately improving the overall efficiency of the entertainment industry.


Then he burst out crying. ‘Sorry,’ he said, regaining his composure. ‘It’s been difficult because there’s been a lot of criticism, obviously, and that was never the intention.’”[154] – Steven Jan Vander Ark

The judicial system’s current treatment of those who seek vindication of their claims for joint authorship paints a rather bleak picture: one where enthusiastic creators like Steven Jan Vander Ark are reduced to tears on the stand. Vander Ark is joined by film consultants like Aalmuhammed, screenwriters like Richlin, sound engineers like Brown, and band members like Morrill who put their trust in the legal system to secure their authorship rights as creative contributors after their fellow collaborators denied them the credit and benefits they deserve. To their dismay, the Ninth Circuit’s three-part test to determine whether an individual qualifies as a joint author to a work was unaccommodating with respect to both the major creative roles they played as well as the modern functions of the entertainment industry.

Entertainment has proven to be one of the most booming industries in society today, providing high quality content in unprecedented ways across the globe. Yet while the industry continues to grow, the legal framework for the authorship question remains stuck in the past. Poorly suited for the expanded marketplace for collaboration and unchanged in the face of widespread innovative media technologies, the Ninth Circuit’s obsolete test hinders, rather than furthers Congress’s explicit constitutional duty to promote the growth of the arts. This paradox, where the judicial system’s treatment of authorship ultimately disincentivizes authors, the very people expected to create, heightens the necessity for reform. To improve employment in the modern entertainment industry and further the policy goals of copyright law, the Ninth Circuit’s test must no longer serve as the governing standard for evaluating joint authorship rights.

The judicial system must play its role in granting and reinforcing the allocation of intellectual property rights so that those who are deserving of credit in the creative process can do so without worrying about having their contribution minimized. This is a call to action for the courts to recognize its role in creating barriers to employment in the entertainment industry and craft more appropriately tailored authorship tests to better accommodate those who are employed and who seek to be employed in the entertainment industry. A more carefully articulated framework can be implemented to strike a fair balance between infinite copyrights on the one hand and a monopoly over copyrights on the other. It is not solely up to the courts; Congress must amend the Copyright Act to overcome the inherent vagueness underlying the statutory definition of “ author” as well. These reforms have been advanced by scholars and creators alike. In a multibillion dollar industry where success is highly correlated with authorship rights, it is simply too costly to leave the Vander Arks of the world stripped of their rightful legal protection.

*J.D. Candidate, New York University School of Law, 2019; B.A., Communication Studies, Northwestern University, 2016. The author would like to thank Professor Day Krolik and the JIPEL Notes Program participants: Jared Greenfield, Vincent Honrubia, Chloe Kaufman and James Yang.

[1] Aalmuhammed v. Lee, 202 F.3d 1227 (9th Cir. 2000).

[2] Id. at 1230.

[3] Id. at 1234.

[4] U.S. Const. art. I, § 8, cl. 8.

[5] Id.

[6] See generally, 17 U.S.C. §§ 101 et seq. (2012).

[7] Id. § 101.

[8] H.R. Rep. No. 94-1476, at 120 (1976), reprinted in 1976 U.S.C.C.A.N. 5659, 5736.

[9] 17 U.S.C. § 102.

[10] Id. § 101.

[11] Id.

[12] Aalmuhammed v. Lee, 202 F.3d 1227, 1229 (9th Cir. 2000).

[13] Id.

[14] Id. at 1230.

[15] Id.

[16] Id.

[17] Id.

[18] Id.

[19] Id. at 1231.

[20] Id. at 1231-32.

[21] Id. at 1232.

[22] Id. at 1236.

[23] Id. at 1232.

[24] Id.

[25] Id.

[26] Id.

[27] Id.

[28] Id.

[29] Id.

[30] Id.

[31] Burrow-Giles Lithographic Co. v. Sarony, 111 U.S. 53, 54 (1884).

[32] Id. at 57-59.

[33] Aalmuhammed, 202 F.3d at 1233.

[34] Id.

[35] Thomson v. Larson, 147 F.3d 195, 207 (2d Cir. 1998).

[36] Id.

[37] Aalmuhammed, 202 F.3d at 1234.

[38] Id.

[39] Id.

[40] Id.

[41] Id.

[42] Id. at 1235.

[43] Id.

[44] Id.

[45] Id.

[46] Id. at 1235-36.

[47] See, e.g., Complaint at 1, Horror, Inc. v. Victor Miller, No. 3:16-cv-01442 (D. Conn. Aug. 24, 2016) (Victor Miller, the author of the screenplay Friday the 13th, issued termination notices to the film’s producers to notify them that he was reclaiming his copyright in the original screenplay); Ron Dicker, Filmmaker Says ‘Stranger Things’ Creators Stole His Ideas in New Lawsuit, Huffington Post (Apr. 4, 2018, 10:39 AM), (Charlie Kessler filed suit against Matt and Ross Duffer alleging they stole his ideas for a sci-fi series called “ Montauk” in their popular Netflix series “ Stranger Things”).

[48] See, e.g., Complaint at 4, Brittle v. Time Warner, Inc., No. 3:16-cv-00908-JAG (E.D. Va. Nov. 11, 2016) (Gerald Brittle sued Time Warner and other entertainment company defendants over the exclusive motion picture rights to the film The Conjuring based on the book he wrote called The Demonologist).

[49] Complaint, Frank Ocean v. Om’Mas Keith, No. 2:18-cv-01383 (C.D. Cal. Feb. 20, 2018) (Musical artist Frank Ocean brought a declaratory action of copyright non-ownership against music producer Om’Mas Keith, who claimed that he had co-written musical compositions with Ocean for the singer’s album titled Blonde).

[50] Aalmuhammed, 202 F.3d at 1234. (“ [T]he audience appeal of the work must turn on both contributions and the share of each in its success cannot be appraised.”) (internal quotation marks omitted).

[51] Richlin v. Metro-Goldwyn-Mayer Pictures, Inc., 531 F.3d 962, 970 (9th Cir. 2008).

[52] Id.

[53] Id.

[54] Id.

[55] Aalmuhammed v. Lee, 202 F.3d 1227, 1235 (9th Cir. 2000).

[56] Brown v. Flowers, 196 F. App’x 178, 184 (4th Cir. 2006).

[57] Id.

[58] Id.

[59] Id. at 184.

[60] Id. at 181

[61] Id. at 191 (Gregory, J., concurring in part and dissenting in part).

[62] Id.

[63] Id.

[64] Ford v. Ray, 130 F. Supp. 3d 1358, 1359 (W.D. Wash. 2015).

[65] Id.

[66] Id.

[67] Id. at 1364.

[68] Id. at 1363.

[69] Id. at 1364.

[70] Id. at 1363-64.

[71] Id. at 1364.

[72] Id.

[73] Morrill v. Smashing Pumpkins, 157 F. Supp. 2d 1120 (C.D. Cal. 2001).

[74] Id. at 1121.

[75] Id.

[76] Id. at 1126.

[77] Id. at 1125 (internal quotation marks omitted).

[78] Id.

[79] Eagle Rock Entm’t, Inc. v. Coming Home Prods., Inc., No. CV 03-571 FMC(AJWx), 2004 WL 5642002, at *1 (C.D. Cal. Sept. 1, 2004).

[80] Id. at *3.

[81] See generally, Harry Potter Wish List, (last visited Nov. 11, 2018) (providing an example of an extrinsic encyclopedia and fan-site dedicated to J.K. Rowling’s book series).

[82] John Eligon, Rowling Wins Lawsuit Against Potter Lexicon, N.Y. Times, Sept. 8, 2008, at B3.

[83] Warner Bros. Entm’t Inc. v. RDR Books, 575 F. Supp. 2d 513 (S.D.N.Y. 2008).

[84] Eligon, supra note 82.

[85] Media and Entertainment Spotlight: The Media and Entertainment Industry in the United States, Selectusa, (last visited Nov. 11, 2018).

[86] May 2017 National Industry-Specific Occupational Employment and Wage Estimates, Sector 71: Arts, Entertainment, and Recreation, U.S. Dep’t of Lab.,

[87] Duncan Crabtree-Ireland, Labor Law in the Entertainment Industry Supplemental Payments, Intellectual Property Rights, and the Role of Unions, 31 Ent. & Sports L. 4, 4–5 (2014).

[88] Id.

[89] Cynthia Baron, The Modern Entertainment Marketplace, 2000-Present, in Acting 144 (Claudia Springer & Julie Levinson eds., 2015) (quoting author Stephen Keane).

[90] Id.

[91] Dastar Corp. v. Twentieth Century Fox Film Corp., 539 U.S. 23 (2003).

[92] Id. at 35.

[93] William T. Bielby & Denise D. Bielby, Organizational Mediation of Project-Based Labor Markets: Talent Agencies and the Careers of Screenwriters, 64 Am. Soc. Rev. 64, 66 (1999).

[94] Gary M. McLaughlin, Oral Contracts in the Entertainment Industry, 1 Va. Sports & Ent. L.J. 101, 119 (2001).

[95] Id. at 120.

[96] See Jay M. Spillane, Lawsuits Over "Handshake Deals" Are As Old As the Entertainment Industry (and Can Be Easily Avoided), 11 Ent. & Sports L. 15, 16 (1993).

[97] Main Line Pictures, Inc. v. Basinger, No. B077509, 1994 WL 814244, at *1 (Cal. Ct. App. Sept. 22, 1994).

[98] Id. at *4.

[99] Id. at *5.

[100] Id. at *2.

[101] See Mihaela Mihailova, Collaboration Without Representation: Labor Issues in Motion and Performance Capture, 11 Animation: An Interdisciplinary J. 40, 46 (2016).

[102] Id.

[103] Dean Goodman & Edwin Chan, Natalie Portman Leaps to Oscar for ‘Black Swan’, (Feb. 27, 2011, 11:20PM),

[104] Mihailova supra note 101, at 46.

[105] See, e.g., Aalmuhammed v. Lee, 202 F.3d 1227, 1232 (9th Cir. 2000)

[106] Id.

[107] Brief of Amici Curiae Professors Shyamkrishna Balganesh et al. Supporting Neither Party, Garcia v. Google, Inc., 786 F.3d 733 (9th Cir. 2015) (No. 12-57302), at 17.

[108] Id. at 13.

[109] Id.

[110] Id. at 13-14.

[111] See generally Homer Croy, How Motion Pictures Are Made, (Goemaere Press 2009) (1918) and J.S. Rudsenske, Music Business Made Simple: A Guide to Becoming a Recording Artist, (J.P. Denk ed. 2004).

[112] See generally Jack Viertel, The Secret Life of the American Musical: How Broadway Shows Are Built (Sarah Crichton Books 2017) (2016).

[113] See generally James R. Martin, Create Documentary Films, Videos, and Multimedia: A Comprehensive Guide to Using Documentary Storytelling Techniques for Film, Video, the Internet and Digital Media Nonfiction Projects (Real Deal Press 2010); Croy, supra note 111.

[114] See Gail Resnik & Scott Trost, All You Need to Know About the Movie and TV Business 36-37 (Simon & Schuster 1996).

[115] Id.

[116] Id.

[117] Balganesh et al., supra note 107, at 14.

[118] See generally Cynthia Littleton & Brian Steinberg, Disney to Buy 21st Century Fox Assets for $52.4 Billion in Historic Hollywood Merger, Variety (Dec. 14, 2017 4:04 AM),

[119] Id.

[120] Garcia v. Google, Inc., 786 F.3d 733, 736 (9th Cir. 2015).

[121] Id. at 742 (quoting Aalmuhammed v. Lee, 202 F.3d 1227, 1233 (9th Cir. 2000)).

[122] Id.

[123] Id.

[124] Id. at 743.

[125] See generally What is the Writers Guild of America West?, WGA, (last visited Nov. 11, 2018).

[126] See Denise Mann, The Labor Behind the Lost ARG: WGA’s Tentative Foothold in the Digital Age, in Wired TV: Laboring Over an Interactive Future 118, 124 (Denise Mann ed., 2014).

[127] Id. at 118-19.

[128] A mobisode is an episode of television intended for viewing on a mobile device. See generally Scott Ruston, Televisual Narratives in the Palm of Your Hand, Understanding Mobisodes, Producing Television, Spring 2008,

[129] Mann, supra note 126, at 118.

[130] Id. at 135.

[131] Id.

[132] Id. at 119-20

[133] Id. at 122.

[134] Id. at 136.

[135] Id.

[136] Id.

[137] Id.

[138] See Derek Johnson, Authorship Up for Grabs: Decentralized Labor, Licensing, and the Management of Collaborative Creativity, in Wired TV: Laboring Over an Interactive Future 32, 40–41 (Denise Mann ed., 2014).

[139] Id. at 34.

[140] See Writers Guild of America, Screen Credits Manual 16 (2010), (defining the term “ writer”).

[141] Id.

[142] Catherine L. Fisk, The Role of Private Intellectual Property Rights in Markets for Labor and Ideas: Screen Credit and the Writers Guild of America, 1938-2000, 32 Berkeley J. Emp. & Lab. L. 215, 250 (2011).

[143] Id. at 248.

[144] Id. at 264.

[145] Balganesh et al., supra note 107, at 15-16.

[146] Id. at 16.

[147] Id.

[148] Christopher Jon Sprigman, Remarks for Advanced Copyright Class at New York University School of Law (Spring 2018).

[149] Id.

[150] 17 U.S.C. § 101 (2012).

[151] Cmty. for Creative Non-Violence v. Reid, 490 U.S. 730, 751-52 (1989).

[152] Id. at 752.

[153] Garcia v. Google, Inc., 786 F.3d 733, 742 (9th Cir. 2015).

[154] Anemona Hartocollis, Sued by Harry Potter’s Creator, Lexicographer Breaks Down on the Stand, N.Y. Times, Apr. 16, 2008, at B1.

“Prime Health” and the Regulation of Hybrid Healthcare*

<p>“Prime Health” and the Regulation of Hybrid Healthcare<a name="_authorref1" href="#_author1">*</a></p>

By: Nicolas P. Terry**

Download a PDF version of this article here

This article examines the possible constructs behind the announcement that Amazon, Berkshire Hathaway, and JPMorgan Chase & Co. are jointly building a new healthcare entity for their employees. In this article, I provide context by discussing and comparing the healthcare ambitions of the three largest information technology companies before arguing that various forms of hybrid entities will increase their footprint in healthcare data and delivery. The core of this discussion is a thought experiment about the nature of what I term “Prime Health.” That analysis is based initially on observations about Amazon’s existing culture and business model of Amazon. Thereafter I examine both what Prime Health could and should be. I argue that it will likely go beyond the pedestrian model of a very large self-funded group insurance plan; will disintermediate traditional healthcare insurers; and attempt to bring consumers and healthcare providers together into some type of online marketplace—an updated, privatized version of managed competition. In the final parts of the article I delve into the regulatory environment that hybrid healthcare generally, and Prime Health in particular, will face. This analysis includes federal device and data protection laws, a few idiosyncratic state laws, and a brief discussion of the problems inherent in the limited regulation of hybrid healthcare entities.



Healthcare is no stranger to dramatic headlines. However, a short press release in January 2018 was not only the mother of all healthcare stories but also desperately short on detail. In a joint press release, Amazon, Berkshire Hathaway, and JPMorgan Chase & Co. announced they were founding an independent company focusing on “technology solutions that will provide U.S. employees and their families with simplified, high-quality, and transparent healthcare at a reasonable cost.”[1] According to Berkshire Hathaway Chairman and CEO, Warren Buffett:

The ballooning costs of healthcare act as a hungry tapeworm on the American economy. Our group does not come to this problem with answers. But we also do not accept it as inevitable. Rather, we share the belief that putting our collective resources behind the country’s best talent can, in time, check the rise in health costs while concurrently enhancing patient satisfaction and outcomes.[2]

Commentators and interviewees responded to this announcement with heady optimism about healthcare reform while the stock prices of traditional healthcare stakeholders came under pressure.[3] The Economist hailed the announcement with the headline “A digital revolution in health care is coming.”[4]

A few months before the press release, there were rumors that Amazon had a secret “skunkworks” healthcare team codenamed 1492.[5] But even today, there is very little information about what Amazon and its two partners—Berkshire Hathaway and JPMorgan Chase & Co.—intend, let alone whether it portends a “revolution.” We do know it will be independent of its founding companies, non-profit, and based in Boston.[6] Still, at the time of this writing, it doesn’t even have a name. Herein the corporate entity will be referred to as “ABJ” and the service/product as “Prime Health.”

Nevertheless, there is some evidence that Amazon and its partners are not just building a better healthcare service but are considering a novel healthcare model designed to eliminate or minimize some well-known problems with U.S. healthcare. These problems include access difficulties (particularly for the very poor and the poor); high and increasing prices and costs (including insurance costs, prescription drug costs, and cost-shifting); substandard care coordination; an incoherent delivery model involving multiple types of entities and financing or reimbursement models; and severe deficiencies in data management and sharing.

At root, many of these problems are a function of friction, as multiple stakeholders exhibit inefficiencies, create indeterminacies, and create informational problems all while generally failing to coalesce with their fellow stakeholders. For a company like Amazon, which has figured out a way to reduce friction and combine its digital and physical presence, U.S. healthcare’s complexity and resistance to reform must be frustrating. And, now perhaps, ABJ sees an opportunity.

This article provides a framework for understanding how the largest technology companies view healthcare and their roles in healthcare’s future. I contrast their approaches and their relationships with traditional healthcare stakeholders, label them as hybrids, and contrast their healthcare interests. Amazon emerges from this analysis as a positive outlier because of its existing role as a retailer of physical healthcare goods, its unique approach to combining digital and physical products, and its participation in ABJ’s ambitious healthcare financing and delivery project. I argue that Prime Health will be more than the sum of its parts and that it will attempt to remove (i.e. disintermediate) some intermediary stakeholder from the healthcare value chain, thereby establishing a new healthcare marketplace. (That is, that Prime Health will cut out certain middle persons from various healthcare transactions.) Further, I analyze how this imagined Primed Health will raise issues of over-regulation and under-regulation.

Part II provides context, discussing the businesses and likely conceptions of healthcare among the major technology companies that have been explicit about their healthcare ambitions: Apple, Alphabet, and Amazon. That discussion leads to a tentative conclusion that their healthcare plays will result in differentiated hybrid entities that combine disruptive technologies with novel approaches to brick-and-mortar businesses. Part III projects the properties of Prime Health, assuming it is based on the existing culture and business model of Amazon. Part IV, working with minimal evidence and relatively unrestrained conjecture, examines both what Prime Health could and should look like if it departs from Amazon’s existing culture and business model. This part argues that it will (i) go beyond the pedestrian model of a very large self-funded group insurance plan, (ii) disintermediate traditional healthcare insurers, and (iii) attempt to bring consumers and healthcare providers together into an online marketplace (i.e. an updated, privatized version of managed competition). Part V discusses how current healthcare regulation (which was drafted with traditional, brick-and-mortar healthcare entities in mind) will apply to these new hybrid entities, with particular attention paid to Prime Health. Part VI highlights some specific concerns about the under-regulation of high technology companies as they broaden their grip on economic activities and healthcare in particular.

I. Hybrid Typology

The involvement of high technology companies in healthcare is not new. While some mythologize the progress of traditional health information technologies such as electronic health records,[7] information technology companies have been part of the health puzzle for a while, albeit without notable effect.[8]

Recently, however, strong evidence has surfaced that the largest technology companies are targeting healthcare with increased interest. Technology companies are investing more heavily in health-related ventures,[9] services such as Uber are participating in non-emergency medical transportation,[10] and key functions such as ICU monitoring are being outsourced to remote providers.[11] At the highest level, some seek to be agents of change and insert their technologies or business strategies into healthcare. While the “big three”—Apple, Amazon, and Alphabet (Google’s parent company)—have been explicit, others such as Facebook are believed to have similar ambitions.[12]

In decoding the announcements and activities of these companies, it is clear that there is not one model of technologically-mediated care but several. While Apple, Amazon, and Google may all be seeking to disrupt, or at least transform, healthcare, their business models and strategies are quite different. As the New York Times notes, “[e]ach tech company is taking its own approach, betting that its core business strengths could ultimately improve people’s health.”[13]

As a result, these interventions into the healthcare space or even attempted disruptions of incumbents by technologically-equipped market entrants cannot be reduced to a binary. It remains unlikely that either the brick-and-mortar business typified by traditional healthcare systems or the pure technological model favored by Apple will emerge victorious. It is more probable that the conquering business will be a hybrid, as was true for their disruption of other markets. For example, Amazon built its disruptive business on a digital platform for selling books yet has now opened some physical bookstores.[14] Similarly, AmazonFresh began as a product category, an extension of Amazon’s delivery service,[15] but its strategic potential changed after the company’s purchase of the brick-and-mortar Whole Foods.[16]

A.  Apple

Apple is first and foremost a consumer electronics company[17] that builds software and hardware, particularly the iPhone. Like other smartphones, the iPhone has become “a portal for managing daily life.”[18] Apple software and services are designed to put the consumer in control of their information (including health information) in a trusted, secure environment.[19]

Although Apple is opening medical clinics for its employees,[20] its primary healthcare focus is on consumer-facing hardware and software that monitors health[21] or empowers consumers to safely curate their own health information.[22] In that latter context, David Blumenthal and Aneesh Chopra argue such an approach removes patient dependency “on the bureaucracies of big health systems or on understaffed physician offices to make their own data available for further care.”[23] With the data liberated and under patient control, “consumer-facing applications . . . have the potential to revolutionize patient-provider interactions and empower consumers in ways never before imagined in the history of medicine.”[24] The Apple strategy does not seem to compete with incumbent healthcare. It creates an additional level of data, derived directly from the patient via its own apps that are secure and patient-centric. Increasingly, however, Apple is facilitating the import of data created in traditional healthcare settings into its ecosystem.[25] Apple is also implementing actions based on the healthcare data it collects, for example, by providing health alerts derived from such data. Progressively, therefore, the data generated or imported by Apple’s customers will resemble or  become intertwined with more traditional healthcare services (and, potentially, intertwined with traditional healthcare regulation).

B.  Alphabet

Alphabet takes an orthogonal approach, concentrating on patient and pre-patient data and using AI to predict, inter alia, risk-factors.[26] The company has invested heavily in numerous health technology companies.[27] Verily, Alphabet’s primary healthcare subsidiary, is developing tools and devices that collect and analyze health data to improve interventions and care management, typically in partnership with existing stakeholders.[28] Recently Alphabet announced a re-organization and consolidation of its healthcare properties into a new Google Health unit.[29]

DeepMind, Alphabet’s UK-based AI subsidiary, has shown a particular interest in health data.[30] Indeed, DeepMind’s close working relationship with Royal Free NHS Foundation Trust, in which the latter supplied DeepMind with the clinical records of 1.6 million patients, drew a stern rebuke from the UK Information Commissioner’s office.[31] Notwithstanding Alphabet’s generalist and data-driven approach, it does have one subsidiary concentrating on direct care initiatives: Cityblock. This subsidiary creates community-based clinics (“Neighborhood Hubs”) in underserved urban areas.[32]

Looking at the future of healthcare through the lenses of these and other companies, we can imagine at one end of the spectrum vertically integrated legacy companies (incumbents) exemplified by the combination of Aetna (the third largest U.S. health insurer) and CVS (the largest U.S. pharmacy and pharmacy benefit manager).[33] At the other end of the spectrum, Apple is building relationships with health researchers and large providers but essentially focusing on consumer products and secure management of patient-controlled health information. In the middle, but closer to Apple, is Alphabet: data-focused but still a hybrid because of the deep relationships it is building with researchers and providers to gain access to their health data.

C.  Amazon

As it did for Amazon Web Services (“AWS”), Amazon may well start Prime Health with the goal of being its “first and best customer.”[34] But, thereafter, it is much more likely it will look to scale and attract other employers and eventually even individuals. It has to do so in order to fulfill its prime directive, extracting profit from all transactions.

A common misperception is that the Prime Health partners have no ambition beyond leveraging their buying power to cut better deals with healthcare incumbents.[35] Indeed, it has been argued that of the big three, Apple and Alphabet will have a greater short-term impact than will Prime Health. Here, the argument is that the initial, internal roll-out of Prime Health will only cover ABJ’s approximately one million employees, whereas Apple and Alphabet may be quicker to make their health insights available to hundreds of millions of external customers.[36]

Others will take that course. For example, Cisco Systems, Intel, and Walmart have all built direct relationships with providers.[37] In contrast, it is simply not Amazon’s way to take such a reactively pedestrian path.

The Amazon pathology around this idea of its own first and best customer is quite consistent. It launched a sophisticated and scalable data system to track its own customers and logistics (Amazon was the first and best customer of AWS) before opening up AWS to outsiders and, not incidentally, many billions in annual revenue.[38] Similarly, Amazon is investing in aircraft and trucks to build its own delivery service infrastructure. Initially, Amazon will be its own first and best customer and no doubt find itself with leverage in negotiations with UPS and FedEx,[39] but inevitably thereafter, third parties will be able to ship via “Amazon Parcel” or however it will be branded. Finally, AmazonFresh lacked a first and best customer within the Amazon portfolio—the solution? Amazon purchased Whole Foods, “the first-and-best customer that will instantly bring its grocery efforts to scale.”[40] Regarding its healthcare ambitions, circumstantial evidence shows that Amazon is exploring aspects of traditional healthcare such as prescription drug distribution.[41]

One of the difficulties in discerning the properties of, much less answering structural or delivery questions about, Prime Health is that it is difficult to see where Amazon’s internal healthcare needs end and Prime Health’s ambitions begin. For example, Prime Health could be a relatively modest effort based on a very large, self-funded ABJ group health plan with “innovation” restricted to somewhat pedestrian (albeit worthy) build-outs of company-owned employee health clinics.[42] That model is supported by some recent hiring by Amazon (rather than Prime Health). For example, Amazon hired a former U.S. FDA chief health informatics officer[43] and a well-known cardiologist.[44] At the moment, it is impossible for outsiders to see the vector between Amazon’s own healthcare ambitions and Prime Health (i.e. what it is building for its employees internally compared with what it is building for users externally). It is possible that Amazon itself does not yet know.

Neither an ambition to build Prime Health nor the supposition that it will likely adopt a hybrid construct completely answer questions such as what Prime Health will look like or how it will function. For example, will Amazon, in the vein of its Whole Foods acquisition, purchase brick-and-mortar providers such as hospitals?[45] Additionally, while some technological “disruptions” of healthcare (for example, Apple enabling individuals to curate their own health information) will add new layers to the existing healthcare industry without  immediately threatening incumbents, Amazon initiatives are typically unfriendly to incumbents. Their entrance suggests there will be some net losers. The next two sections address these structural questions, first by keying in on some of the cultural and business practices that define Amazon and, second, by examining the various ways that Prime Health could be structured and how it could deliver access to care (or even provide actual care).

II. How the Amazon Model May Frame Prime Health

The next two parts use a thought experiment to analyze Prime Health, which proceeds as follows. Part III highlights the foundational properties of Amazon that are likely to migrate to Prime Health. It relies on identifiable properties of the Amazon business model (“Amazon’s DNA”) such as data-leveraging and being customer-driven. Part IV expands that vision, imagining what Prime Health could or should achieve if not constrained by Amazon’s existing culture and business model.

Since the pioneering work of Kenneth Arrow, it has been well established that healthcare fails to obey basic market rules.[46] In part, this explains the political interest in more radical disruptions of healthcare, with internally-driven reforms or externally-fashioned, market-driven reforms deemed as unlikely.[47] U.S. businesses built around well-functioning markets who shop for healthcare or health insurance for their employees are likely confused or at least frustrated by an ecosystem strewn with path-dependent practices and misaligned incentives. Not surprisingly, therefore, mainstream businesses generally have not expanded into healthcare while technology-harnessing disruptive businesses have continued their relentless targeting of even the most unlikely brick-and-mortar incumbents.[48]

However, Amazon, often to the befuddlement of financial analysts, is a very different type of company and somewhat counter intuitively may prove to be a formidable competitor in healthcare. For example, Amazon “in an accelerated and innovative way . . . continues to invest in areas of growth at the expense of profitability, something most other retailers (and other firms) can’t afford to do.”[49] And Amazon is also a patient company, “willing and able to build [Internet] businesses with the patience that will be necessary to wait for the old order to collapse.”[50] These characteristics support the imagining of Prime Health as a long game disruption of U.S. healthcare.

A.  Data Driven

Data may or may not be the new oil[51] or the new coal.[52] However, it is clear that the promise of integrating data is a major driver of merger and acquisition activity. As Wendy Epstein noted in the context of the CVS-Aetna combination, “in a world where consumer costs have dominated the discussion, it is at least worth thinking more about the impact that the integration of payor and provider data might have.”[53] Further, according to Dan Mendelson, “[a] lot of the basic wiring of the health care system is now complete—a result of federal investment and lower technology prices. The need now is to harness the power of analytics to improve care. . . . [T]his drives mergers and acquisition.”[54] Data is also an important demand-side driver. Patients lacking knowledge and control is a major problem in our current healthcare system; empowering them with access to data may provide such knowledge and control, thereby increasing consumer demand for healthcare.[55] Prime Health’s recent hiring of COO Jack Stoddard, who has a strong track record in digital health, further suggests the extent to which Prime Health will be data-driven.[56]

B.  Combining the Digital with the Physical

The “big three” technology companies already possess almost unimaginable amounts of customer data, some of it health information and much of it medically-inflected data. They have also built some of the most sophisticated analytics and logistics platforms. Further, Amazon broke new ground by combining the digital and physical in a way never before seen.[57] As Michael J. Coren notes, “Amazon’s unprecedented logistics and delivery infrastructure, paired with access to personal data about Americans’ purchasing habits, means it is unique in the history of global commerce.”[58]

It has been more than a decade since Steven Spear argued that healthcare should adopt the technologies and practices of successful businesses and “tightly couple the process of doing work with the process of learning to do it better as it’s being done.”[59] Yet, how many healthcare providers have heeded or implemented that message? In contrast, any Prime Health service providers would become customers of key strategies and technologies, rapidly improving areas where healthcare institutions suffer from under-investment or lack of scale, such as procurement, secure IT, and data analytics.

Amazon already possesses many healthcare properties, of which Prime Health could potentially take advantage.[60] For example, Amazon already sells medical supplies and equipment,[61] offers one-hour delivery on some OTC drugs in Seattle,[62] sells prescription drugs in Japan,[63] has reportedly been acquiring expertise in selling prescription drugs in the U.S.,[64] and recently acquired PillPack, an online pharmacy specializing in pre-sorted medication doses.[65] Further, in Whole Foods, Amazon has a collection of strategically placed locations that are already being fitted with pickup and return lockers for Amazon customers. Without a doubt, these could be expanded to include brick-and-mortar pharmacies or clinics. The company even holds a patent for “anticipatory package shipping,” meaning they can ship goods to consumers before such goods are ordered, potentially reducing the costs and delays associated with the online market.[66] If Amazon fully deploys to Prime Health its properties in the healthcare space, Prime Health could fundamentally alter the industry.

C.  Customer-Facing

Just as it is unimaginable that Prime Health would transmit data using fax machines, “the cockroach of American medicine,”[67] it is also hard to believe it would tolerate brick-and-mortar anachronisms. For example, Prime Health surely would have no tolerance for waiting rooms that “put [] a high cost on a medical provider’s time while valuing the patient’s time at essentially zero.”[68]

Amazon continually works on the customer experience.[69] As an organization, it deploys considerable resources to discover and correct barriers to growth (or other “invisible asymptotes”[70]), such as consumer dislikes. It also has a culture of high standards and continual improvement. According to Amazon CEO Jeff Bezos:

One thing I love about customers is that they are divinely discontent. Their expectations are never static—they go up. It’s human nature. . . . People have a voracious appetite for a better way, and yesterday’s ‘wow’ quickly becomes today’s ‘ordinary.’ . . . I sense that the same customer empowerment phenomenon is happening broadly across everything we do at Amazon and most other industries as well. You cannot rest on your laurels in this world. Customers won’t have it.[71]

In contrast, healthcare stakeholders continue to disagree over the importance of consumer-patient satisfaction.[72]

Amazon also has some intriguing customer-facing assets that have been built on top of and cemented the Prime membership lock-in. Echo devices, which allow access to Prime products, are now installed in millions of homes. It should be relatively simple to add new “Alexa skills” that respond to health inquiries and medical emergencies.[73] Also, because Echo-Alexa is a platform, its voice commands are now tightly integrated with myriad home automation products. For example, imagine if Prime Health were to build a team of first responders. When Prime Health first responders arrive at a Prime customer’s home, they could request that its front door be remotely unlocked and house lights turned on. To many, this is a dystopian vision, but Amazon has convinced its most loyal Prime customers that it deserves their trust either because of an instrumental trade-off between privacy and convenience or the dubious proposition that Amazon does not sell data to third parties and merely exploits data for its own purposes. Amazon also sells the Echo Look,[74] a fashion-oriented body camera, and the Echo Show,[75] a video communication device. Little re-tooling should be required to make these the basis for an effective telemedicine platform. Recently, Amazon has even announced that its Alexa AI engine will be acquiring some basic emotional awareness.[76]

D.  Reading the Gawande Tea Leaves

Notwithstanding early reports that ABJ was experiencing difficulty in finding its first CEO,[77] in June 2018, ABJ announced that Dr. Atul Gawande, the well-known surgeon, writer, and academician, will lead the new company.[78]

There are several reformist themes readily discernible from Gawande’s books, popular press, and peer-reviewed literature. His best known writing on the healthcare system appeared in three articles in the New Yorker from 2010-12: The Cost Conundrum,[79] Cowboys and Pit Crews,[80] and Big Med (aka the Cheesecake Factory).[81] Gawande’s themes are relatively consistent: over-utilization (particularly of surgical procedures and procedures with debatable cost-effectiveness), the need for reimbursement to be based on quality not quantity, inadequate attention paid to the actual needs and satisfaction of patients, lack of care coordination needing to be met with teamwork, increased standardization and checklists, and a recognition of “the reality that medicine’s complexity has exceeded our individual capabilities as doctors.”[82] In medical literature, he is probably best known for his (and his research center, Ariadne Lab’s) work in designing the World Health Organization surgical checklist[83] and its validation studies both globally[84] and in an important South Carolina study.[85] Just after his announcement as ABJ’s CEO, while speaking at the annual meeting of America’s Health Insurance Plans, Gawande opined that surgery is the single largest healthcare cost.[86]

Gawande seems to hold relatively mainstream views touched on by other centrist reformers,[87] such as advocating for reasonable cost containment strategies.  (To a certain degree, some fully vertically-integrated providers already implement such strategies.[88]) Gawande was a defender of the ACA’s incremental approach to cost containment,[89] earning a rebuke from Alain Enthoven for defending Congress’ “unmistakable signal that it is unable or unwilling to control health expenditures and the fiscal deficit.”[90]

Other than suggesting a centrist approach and a sensitivity towards health equity, we know little about how Gawande will approach his leadership role at Prime Health. That said, a commitment to patient safety, patient satisfaction, and concerns about tertiary healthcare costs suggest Gawande is a leader well-versed in the current challenges facing healthcare in the U.S. and someone compatible with Amazon’s DNA.

III. What Prime Health Could or Should Be

This section builds on the foundational properties described above. It is more speculative, relying on the work of business analysts, previous disruptions by technology companies, and the potential for disintermediation of some healthcare incumbents. Unlike Part III, this Part IV imagines what Prime Health could look like if not constrained by Amazon’s existing culture and business model. Questions about what Prime Health could be are interwoven with what it should be. The conceit of this article is that Prime Health could be built off the disintermediation of some stakeholders and the commoditization of the products and services of others.[91] Therein also lies the normative point; what Prime Health should be is an attempt to demonstrate vitality and innovation in the employer role in providing access to healthcare services.

These questions include the likely customers and where the enterprise will identify structural and delivery deficiencies that can be improved upon. There is also a fundamental question of whether Prime Health will be based upon some existing construct or will be built from whole cloth. [92]  According to the Economist, “[t]here are two broad routes into health care. The first is doing business with hospitals and health-care companies in the existing system . . . . A second route is for tech firms to use their various platforms to create entirely new channels through which medical care can be delivered to patients.”[93] These are accurate descriptions of the disruption-lite we are seeing from Alphabet and Apple, with the former tending to build data relationships with existing providers on top of which it can build analytics products and the latter tending to build new channels and tools atop of existing clinical and research relationships.

A.  Strategic Options

There are some imperfect parallels to an expansive sketch of Prime Health. For example, the idea of commoditizing healthcare products and services is somewhat reflected in the announced strategy of large healthcare networks to create their own drug company to manufacture off-patent drugs.[94] Essentially, the providers intend to commoditize off-patent drugs to counter the practice of pharmaceutical companies continuing to sell such drugs at high prices because of “a reduction in the number of suppliers, consolidation of production volumes, and a concentration of market pricing power.”[95]

Vertically-integrated providers may provide another preview (albeit imperfect) of what Prime Health could theoretically look like. Some non-governmental, vertically-integrated providers, such as affordable care organizations (“ACOs”), report positive results, although generally with regard to quality rather than reduced costs.[96] Overall, evidence about the performance of fully vertically-integrated providers is mixed.[97] An additional parallel can be drawn to the Veterans Health Administration (“VA”), a fully vertically-integrated provider. The VA, for obvious reasons, “owns” its consumers and it purchases healthcare goods and services (their clinical staff are employees) without using intermediaries. The VA uses comparative effectiveness research to determine clinical practices and which drugs to include in its formulary.[98] It is certainly within the realm of probability that Prime Health would adopt a limited formulary like the VA does[99] or like Massachusetts has proposed for its Medicaid program.[100]

B.  The Customers

The ABJ parent companies have a national presence and employ approximately one million people in worldwide.[101] A fundamental question is whether ABJ will build Prime Health exclusively for its own employees or build it out as a generally available product. If the endeavor hues to Amazon’s traditional practices, Prime Health will begin with ABJ employees (its first and best customers[102]) and then extend outwards, maybe to other large employers and ultimately even to individuals.

In June 2018, at an event planned before his appointment at ABJ, Atul Gawande gave a lengthy interview at the Aspen Ideas Festival.[103] He spoke about the demographics of ABJ employees, beginning with Amazon’s fulfillment center workers:

Most of [Amazon’s fulfillment center workers], people who probably are there only about a year or so. These are people who have very unstable health care, been in and out . . . . Then you get to JPMorgan Chase, where their largest employment group are bank tellers. So you’re talking about people at the 30th, 40th, 50th percentile of income. They fall between the people who get Medicaid and the people who get Medicare.[104]

Gawande was also sensitive to health inequities present in our current system, noting how middle income individuals pay “taxes for people to get Medicaid. That’s better coverage than they could ever get. No co-pays, no premiums, no deductibles.”[105] Finally, he reflected on the growing problem of underinsurance, stating that “typically, in private sector employment, you’re getting up to $1,000, $2,000 deductibles these days.”[106]

It is certainly tempting to believe that a company led by Dr. Gawande— someone with a sophisticated understanding of the societal costs of access stratification and the detrimental impact of health insurance policies’ declining actuarial values—has ambitions beyond its own employees. If Prime Health were to expand beyond its own employees, it would likely be facilitated by Amazon’s existing “ownership” of more than 100 million “Prime” members.[107] Among those are persons with lower incomes who are subsidized by Amazon,[108] a perhaps tenuous link to the Affordable Care Act (ACA) model of making individual insurance affordable on the exchanges.[109]

C.  Friction and Intermediaries

U.S. healthcare has become a zero-sum game, shifting costs rather than creating new value.[110] This does not have to be the case. Healthcare incumbents could expand into new businesses that are not directly related to clinical care, such as fitness and healthcare management.[111] To an extent, this is the market that Apple is beginning to dominate, at least for more affluent consumers. Such consumers spend additional, discretionary “healthcare” dollars to have their health information collected by an Apple Watch and securely stored on an iPhone. In this scenario, Apple is unlikely to pull revenue away from traditional healthcare businesses, although it may disrupt the watch and direct-to-consumer (“DTC”) medical device markets.

However, Prime Health is likely to play out in a zero-sum environment. If Prime Health is going to succeed, some other healthcare businesses are going to suffer or fail, just as booksellers, network television, department stores, and shopping malls have failed over the last twenty years because of competition from Amazon.[112]

The U.S. healthcare system has many idiosyncratic properties, most of which either increase cost or make it harder to decrease costs. Many of these properties can be attributed to accidental intermediaries or rent-seekers who arguably add little value to the supply chain that connects patients to healthcare services and products. In contrast, technology companies thrive on eliminating friction from transactions and reducing barriers between the consumer and the goods or services they want.[113]

Employers fall into the “accidental intermediary” category. One of healthcare’s original sins was the placement of employers in the middle of healthcare markets, a mistake made soon after the Second World War ended. U.S. healthcare has been path dependent ever since. After all, how many employees want their employers involved in their wellness or healthcare? Why, pre-ACA, should an employer’s business decision about the structuring of a health insurance benefit (self funded vs. fully funded) be determinative of whether state or federal law applies? Even when employers are not involved, their ghosts interfere. For example, the historical tying of health insurance to employment is in part responsible for the recent movement to making work a requirement for Medicaid.[114]

These negative externalities have been offset somewhat by the preeminence of group health insurance in reach, cost, and coverage due to the strong bargaining position of employers. As the percentage of Americans covered by employer health insurance continues to gradually decline,[115] the premiums paid by employees continue to rise.[116] Moreover, copays and other out-of-pocket (“OOP”) cost-shifting mechanisms reduce actuarial value, creating a cohort of underinsured employees.[117] The continuing role of employers seems less of a positive. Indeed, the fact that employers can deduct the cost of employer-sponsored health insurance (part of the “accident” in employer involvement in healthcare) arguably encourages overspending on healthcare[118] and led to the ACA’s high-cost plan tax (the so-called “Cadillac Tax”).[119] Whether the high-cost plan tax is progressive or regressive is still under debate.[120] Politically unpopular, the tax has been repeatedly delayed[121] and may never be applied.

Notwithstanding the federal government’s promotion of health information technologies[122] or the promise of Apple and Google’s forays into health-related gadgets and data analytics respectively, healthcare has proven difficult to disrupt.[123] According to disruption guru Clay Christensen and his colleagues, “[t]hird-party reimbursement systems sap motivation for innovation—particularly disruptive innovation—out of the system.”[124] It is therefore possible that the predicate for Prime Health disruption is to remove third-party reimbursement from the picture.

Collectively, health insurers have reported annual revenues of more than three-quarters of a trillion dollars since 2013[125] and continue to report record profits.[126] Intuitively, they should function to reduce costs. However, Elisabeth Rosenthal notes that health insurers have very little incentive to negotiate better prices with suppliers of healthcare services and products.[127] What, then, is their value as intermediaries or infomediaries?

Developments in the 1990s hinted at a positive answer. For a few years, managed care reversed the trend of healthcare inflation as consumers moved to health maintenance organizations (“HMOs”).[128] However, backlash from consumers and physicians soon put a stop to this reversal,[129] as healthcare continued to devour more and more of the economy, except during a short blip during the Great Recession.[130] A general summary of the failure is this: managed care attacked volume despite the fact that the real issue, of course, was (and remains) price,[131] including drug prices,[132] hospital charges,[133] and physician salaries.[134]

Historian Christy Ford Chapin argues that “[t]he problem with American health care is not the care. It’s the insurance.”[135] Health insurers’ failure to control prices (for which they are not necessarily responsible) explains two provisions in the ACA: (1) the Medical-Loss Ratio (“MLR”)[136] and (2) the Independent Payment Advisory Board (“IPAB”).[137] Further, the Washington Post Editorial Board noted, in reference to the IPAB, that “[e]very bit of waste is some companies’ profit, and the industry wasn’t going to let it go without a fight—though of course it pretended to be fighting on behalf of patients.”[138] This sentiment can also be applied to the MLR. The MLR was designed to force insurers to spend a greater percentage of premium income on healthcare rather than administrative costs or executive salaries.[139] Few ACA provisions have had as much vitriol spilled on them as the IPAB. IPAB was a potential check on Medicare overspending that, if successful, would likely have influenced private insurance spending. It was despised by healthcare companies[140] and finally laid to rest in the 2018 Spending Bill.[141] In its obituary, the Washington Post opined that “[t]he IPAB . . . represented Congress’s peak effort at serious spending restraint on health care, which is probably why it had few champions and a long list of enemies. Now, before ever beginning its work, IPAB has been smothered.”[142]

D.  Managed Competition: HillaryCare for the Digital Age

According to business analyst Ben Thompson, “the most important consequence of the Internet . . . was the reduction of the cost of distribution to effectively zero.”[143] Amazon had a novel approach to the cost of distribution. It made the cost of distribution appear to be free by having its customers pre-pay through Prime. At that point, big box stores, grocery stores, bookstores, etc., lost their distribution monopolies. According to Thompson, once a distribution monopoly is lost, disruption occurs. First, Internet-based companies can take over the ownership of customers. Second, once products are no longer protected by the cost-based distribution monopoly they are easier to commodify.[144] It is that commodification of health services and products that may be at the core of the Prime Health construct.

The lack of insurer interest in engaging in meaningful cost containment and the absence of government negotiation[145] combine to keep healthcare prices high. High margins and unhelpful intermediaries are anathema to Amazon’s view of commerce. After all, as noted by Malcolm Harris, “[t]he world’s biggest store doesn’t use suggested retail pricing; it sets its own.”[146] Disintermediating insurers could be the first step in a “disintermediate-commoditize-build out” strategy.

As to the second step, Thompson speculates that Prime Health will build out “interfaces” for its employees to access existing healthcare services (presumably treatment, care, equipment, pharmacies, etc.). Then it will build “infrastructure for those healthcare suppliers, requiring them to serve Amazon’s employees using a standard interface.”[147] That infrastructure will then morph into a marketplace where services compete to serve the employees. Essentially, if these hypotheses play out, Prime Health will have disintermediated health insurance companies and set the stage for the commoditization of healthcare services.[148]

Once health insurers are disintermediated, Prime Health will be free to build the marketplace of its choice. At first sight, ABJ may take the Obama-era construct of the Health Insurance Marketplace, designed to link patients to insurers (the traditional sine qua non for healthcare),[149] but re-engineer it to link patients to providers. However, the Prime Health construct arguably may go further. If ABJ is seeking to be an agent of change and provide a structure whereby those providing healthcare services at all levels compete for ABJ dollars, the Prime Health construct may resemble the Clinton managed competition plan of the 1990s more than the managed care of the 1980s or the ACA’s individual exchange marketplaces.[150] In this iteration, the non-profit Prime Health would replace the federal government in “managing” the competition. And if successful, the strategy will lead to the third step, building out Prime Health as a product to be offered outside of ABJ. ABJ’s founding companies would be the first and maybe best customers of Prime Health, but eventually not the only ones.

E.  Delivery and Services

Assuming the existence of some form of marketplace, questions arise as to the extent ABJ will build out its own services and how disaggregated the non-ABJ services offered on the marketplace will be. For example, it is obvious that Amazon already has efficient interfaces to somewhat commoditize OTC pharmaceuticals, and Amazon seems to be expanding into Rx pharmaceuticals. Equally, Amazon is already an established seller of Durable Medical Equipment (“DME”).[151] After these are established, the difficult question will be what to build and what to outsource in terms of services.

One answer would be to outsource the entire services bundle: primary, secondary, tertiary, and quaternary care.[152] This is an approach increasingly favored by large employers. For example, General Motors has contracted directly with the Henry Ford Health System to offer a healthcare option for its non-union employees under which the hospital system will provide all services.[153] As part of its “Centers of Excellence” program, Walmart contracted with the Cleveland Clinic Heart & Vascular Institute for some tertiary and quaternary services (heart, transplant, and spine care) for its employees.[154]

However, these relationships, like other direct-care models, are not as disruptive as would first appear. In these models, health insurers are still used as intermediaries, providing Administrative Services Only (“ASO”) in much the same way as in fully-funded employer-provided group insurance.[155]

Accordingly, Prime Health is unlikely to follow a simple outsourcing model. ABJ already has considerable negotiating weight and would not need to build a new non-profit corporate structure to execute such a strategy. The answer could be partial outsourcing, building out primary care through either free-standing or Whole Foods-based clinics,[156] and using an ASO model for higher tier services. This also feels like a half-hearted attempt at change that fails to reflect Amazon’s DNA. In the end, the purchase of a hospital chain or some other conventional healthcare asset is possible, if not probable.[157]

F.  Caveats

Prime Health could head in many different directions. Certainly, there are many issues and possible hurdles to monitor. The weakest link in the disintermediate-commoditize-build out model is the second—persuading those who sell high margin healthcare products and services to enter the Prime Health “trap.” This will be challenging, especially as the likely endgame is that their products or services will be commodified (or at least their competition will be “managed”). After all, unlike the German healthcare systems, there is no government mandate that regional providers negotiate as a unit.[158] Some trend lines could play into ABJ’s hands. Increased premiums, escalating OOP costs, and the rise of limited benefit health plans could make healthcare increasingly unaffordable for consumers. Alternatively, there may be a point at which the federal government successfully caps its healthcare expenditures by reducing Medicare services and converting Medicaid to block grants. In short, there are scenarios where having Amazon pick up excess capacity could be attractive to service and product suppliers. Waiting for this moment is where Amazon’s renowned patience may come in to play.

Another issue is over-utilization. A primary motive behind moving towards value-based care and away from fee-for-service reimbursement[159] is to move towards consumers “pulling” necessary and patient-centric services and  away from physicians “pushing” services (i.e. provider-induced consumption).[160] In its search to make the consumer purchasing friction-free, Amazon has built an array of consumer-facing tools (i.e. Prime, Subscribe & Save, Dash Buttons, Echo devices, etc.) that are designed to encourage consumers to buy (i.e. pull) more and more products. To control utilization ABJ will need to mimic the re-conceptualization currently being asked of healthcare and tweak its algorithms to move from quantity to value.

No doubt Prime Health is developing alternative strategies and its strategic focus will likely adapt and shift over time. It is, however, difficult to imagine an Amazon-linked product that does not focus on consumer satisfaction, extensively leverage data, build out its own profit centers to replace services it has historically paid for, and reduce transactional friction.

IV. Regulating Hybrid Healthcare

The past decade or so has witnessed an interesting race: whether healthcare providers would become technology-driven businesses more quickly than technology companies would learn how to provide healthcare services.[161] Increasingly, it seems that the technology companies will win that race. Healthcare technologies, from apps, to robotics, to AI are developing far faster than conventional healthcare can reinvent itself. At the same time, technology companies do not have to deal with innovation-sapping third-party reimbursement. Nor have technology companies had to cope with the complex regulatory structures faced by healthcare insurers, providers, and researchers.[162]

As already noted, the major technology companies investing heavily in healthcare are choosing different strategies, stressing different types of technologies (for example, apps rather than AI), and concentrating on different markets (for example, consumers rather than healthcare entities).[163] Whatever the technology-conventional healthcare “mix” they settle on, these technology companies will likely be operating some type of hybrid healthcare entities.

This evolving hybrid typology raises some interesting regulatory questions. The assumption is that the current mixture of federal and state regulatory frameworks will apply, although often without success.

A. Applicability of Federal Law

1.  Drug and Device Regulation

As discussed above, there is little reason to believe that ABJ will enter the pharmaceutical production space (primarily because of research and development and intellectual property costs), although it might be interested in the distribution of generics made by others. It is more likely that ABJ will look to reduce Prime Health drug costs by disintermediating not only insurers but also pharmacy benefit managers and leveraging its purchasing power to create a cost-effective formulary. As such, ABJ will need to comply with state pharmacy licensure regulations,[164] an area in which Amazon has expertise through its experience with[165] and its current pharmacy properties.[166]

Devices, however, may be a different matter. Amazon designs and sells an array of intelligent assistant products[167] for use inside and outside of the home.[168] Alexa’s AI engine is now incorporated in 20,000 devices manufactured by a diverse set of companies, from home thermostats, to alarm systems, to automobiles.[169] In the near future it is likely to find its way into more advanced products such as home robots.[170]

Intelligent assistants are unlikely to be a core part of Prime Health’s initial roll-out (the priority should be the marketplace). It is hard, however, to imagine that technologically-mediated care will not be a major part of Prime Health. That raises the likelihood that Alexa-enabled devices will start processing health data. These devices could make physician appointments, request simple prescription renewals, prompt medication adherence, and compile competitive, value-driven offers for surgical interventions. To the extent that Alexa-enabled devices are used to process health data, they may fall within the purview of the Food, Drug, and Cosmetic Act (FDCA), which defines a “device,” in part, as an instrument “intended for use in the diagnosis of disease or other conditions, or in the cure, mitigation, treatment, or prevention of disease, in man.”[171] This is to say that Alexa-enabled devices will likely cross the line that separates consumer devices from FDA-regulated medical devices at some point.

As per FDA-issued guidance, the FDA exercises discretion with regard to very low risk devices and has typically excluded from regulation many fitness devices and some mobile medical apps.[172] The 21st Century Cures Act of 2016 formally extended that exclusion to some fitness software.[173] However, neither that reform nor the reissued draft guidance on app software[174] is likely to protect Amazon from a “device” characterization if Alexa is medicalized.

Finally, a recent initiative, the FDA’s Precertification (Pre-Cert) Program, is designed to better align regulatory and technology iteration cycles by using a surrogate device approval based on approving manufacturers and their safety-testing protocols.[175] Other healthcare hybrids, including Apple and Verily, are part of that program; Amazon is not.[176] Amazon signaling interest in Pre-Cert would signal how Alexa’s AI may be deployed within Prime Health.

2.  Data Protection

Amazon has avoided serious scrutiny of its data practices even while other major digital companies such as Facebook have been rife with scandal.[177] While Amazon shares data with a limited number of third parties and mines customer data to make purchase recommendations, so far Amazon seems to view customer data as a closely held asset rather than something to be sold.[178] The primary reason Amazon does not have a data protection problem has less to do with its data policies than it does with the fact that, at least in the U.S., it (along with Facebook, Google, and Apple) exists in a severely under-regulated space.[179] In broad terms, companies holding vast stores of consumer data are essentially unregulated so long as they comply with their own privacy policies.[180]

Unlike consumer data regulation, U.S. data protection for health information is relatively robust.[181] The Privacy and Security Rules promulgated under the Health Insurance Portability and Accountability Act of 1996 (HIPAA) broadly apply to both traditional healthcare providers and their business associates holding individually identifiable health information.[182] Subject to some exceptions, the HIPAA Privacy Rule prohibits the sharing of personal health information with those outside the healthcare domain.[183] For hybrid companies to be drawn into the HIPAA web, it would not be enough for them to merely hold or process personal health information.[184] Rather, the hybrid would have to satisfy the HIPAA Privacy Rule’s definition of “covered entity.”[185] Included in this definition is a “health plan,” which the Act defines as an “individual or group plan that provides, or pays the cost of, medical care.”[186] In short, Prime Health would be covered by HIPAA if it were held to be an “insurer” or a “group health plan.” The former is possible, as discussed below, but the latter is probable.[187] The specifics of HIPAA compliance for group insurance plans[188] should be no surprise to Amazon, as this is the model Amazon currently uses to provide coverage to its employees.[189]

While Amazon is capable of achieving HIPAA-class security,[190] its current Alexa-enabled devices are unlikely to survive a HIPAA audit.[191] Further, Prime Health would require a privacy model (and culture) that would be distinct from Amazon’s own, and ABJ would need to resist attempts by its partners to cross-market or data-mine using Prime Health data.

3.  Insurance, ERISA, and Fiduciary Duties

Assuming the premise that Prime Health will disintermediate existing health insurers (at least for ABJ employees), the question arises whether it will be considered an “insurer” engaging in the “business of insurance,” within the meaning of the McCarran-Ferguson Act,[192] for regulatory purposes. Although that could be a fascinating inquiry and there are some indications that Amazon is entering insurance markets in other countries,[193] this is not a crucial inquiry given the current regulatory model in the U.S.

The McCarran–Ferguson Act reserves insurance regulation to the states.[194] If Prime Health is structured as a fully insured plan (i.e. ABJ pays premiums to a health insurer to cover its employees), it will be subject to diverse state insurance regulations. These include, for example, licensing, benefits, consumer protection, and dispute resolution regulations. However, this would be extremely unlikely if Prime Health is not structured as a self-funded plan in which the employer takes the risk subject to stop-loss reinsurance. In general terms, self-funded plans are viewed as employee benefits subject to the Employee Retirement Income Security Act of 1974 (ERISA).[195]

ERISA plays two roles. First, it applies some substantive federal regulation over self-funded plans, such as imposing fiduciary duties on the employer-insurer.[196] Second, ERISA preempts state insurance regulation for self-funded plans,[197] essentially deregulating employers’ self-funded plans. Some of  that deregulation was countered by HIPAA Subtitle A and the ACA.[198] A broad preemptive approach to ERISA has been adopted by the United States Supreme Court,[199] and as Congress and the Trump Administration continue to deprecate ACA protections, the ERISA shield may become important again. ERISA (and the continued viability of the ACA) should provide some stable, national framework for the regulation of Prime Health insofar as Prime Health is viewed as an insurer or, more likely, as a self-funded health plan.

B.  Applicability of State Laws

As mentioned, pursuant to the McCarran-Ferguson Act, Prime Health will be subject to state insurance regulations if it is structured as a fully insured plan. Outside of insurance regulation, ABJ will also face a broad array of state healthcare-related legislation and regulation requiring navigation.

ABJ’s employees are scattered across the country. Amazon alone has 140 fulfillment centers and a brick-and-mortar presence in an increasing number of states.[200] As their physical presence expanded into more and more states, Amazon began collecting state sales taxes on purchases, albeit often in exchange for tax credits or other state largesse encouraging Amazon to set up shop locally.[201] (In fact, they stopped relying on the Quill Corp. v. North Dakota[202] sales tax loophole long before South Dakota v. Wayfair, Inc.[203]) Currently, Amazon collects state sales tax on purchases in all forty-five states that have a statewide sales tax.[204] In his Aspen Institute interview, Atul Gawande noted ABJ employees “are across the entire country.”[205]

Of course, there are broad policy disagreements among the states, creating differences in state laws relating to Medicaid expansion, scope of practice,[206] telemedicine,[207] and prescription drug cost controls.[208] To the extent Prime Health begins to resemble traditional insurers or healthcare providers, ABJ will likely have to contend with a myriad of similar, but not identical, state licensing statutes,[209] regulations,[210] and regulators[211] that may increase the cost of providing interstate care. Judicial regulation of healthcare through tort and contract models is almost exclusively a matter of state law, featuring differences in the standard of care,[212] the disclosure model for informed consent,[213] and the rules defining the “corporate” liability of healthcare entities.[214]

Large corporations operating in multiple states will generally recognize and “price-in” such differences between state laws. This is expected. However, state laws that have anti-competitive effects may have more of a surprising result. Two laws that qualify are the corporate practice of medicine doctrine (“CPM”) and certificate of need laws (“CON”).

1.  Corporate Practice of Medicine Doctrine

The CPM doctrine is an offshoot of the historical licensing of persons in the “practice of medicine.”[215] Corporate entities such as HMOs did not enter the healthcare provider space until the 1970s, and corporate consolidation of healthcare entities did not become pronounced until the 1980s.[216] Not surprisingly, the licensing processes and criteria that developed in the late nineteenth century were framed by and for physicians and not for then unforeseen clinicians such as nurse practitioners or healthcare corporate entities.[217]

Broadly stated, the CPM doctrine prohibits corporations from either practicing medicine (a variant is to require entities to be owned by physicians) or employing physicians to do the same. In contemporary practice, the justifications for the continued existence of the doctrine are the primacy of individual physician judgment[218] and quality of care.[219] Perhaps dubious policy justifications aside, one of the major problems with CPM is that the doctrine varies on a state-by-state basis. In the words of the Supreme Court of Illinois:

[S]ome jurisdictions refused to adopt the prohibition against the corporate practice of medicine altogether. . . . [O]ther jurisdictions determined that the corporate practice of medicine doctrine is inapplicable to nonprofit hospitals and health associations on the basis that the public policy arguments supporting the doctrine do not apply to physicians employed by charitable institutions. . . . [T]he remainder of jurisdictions have determined that the prohibition against the corporate practice of medicine does not apply to hospitals which employ physicians because hospitals are authorized under other laws to provide medical treatment to patients.[220]

According to Nicole Huberfeld, “[t]he corporate practice of medicine doctrine is a relic; a physician-centric guild doctrine that is at best misplaced, and at worst obstructive, in the present incarnation of the American healthcare system.”[221] It is, however, just the kind of relic that a radically restructured hybrid health entity may have to navigate.

2.  Certificate of Need Laws

CON laws are another historic relic. The Hospital Survey and Construction Act, better known as the Hill–Burton Act of 1946,[222] was partly designed to modernize the post-war hospital system without overbuilding in particular areas.[223] States responded by passing CON laws designed to scrutinize proposals to build healthcare facilities. As explained by the National Conference of State Legislatures, “[t]he basic assumption underlying CON regulation is that excess capacity stemming from overbuilding of health care facilities results in health care price inflation.”[224] The federal construct expired decades ago, yet most states have maintained some type of legislative or regulatory structure for determining whether proposed healthcare construction qualifies for a certificate of need.[225]

The ABJ partners should rightly conclude that CON laws are anticompetitive. As noted by the FTC, “[b]y interfering with the market forces that normally determine supply of services, CON laws can suppress competition and shield incumbent health care providers from competition from new entrants.”[226] The Court of Appeals for the Fourth Circuit, however, is less receptive to the argument that CON laws are anticompetitive:

Appellants’ free market arguments also overlook the fact that the health care market has its own idiosyncrasies. . . . Squeezed by insurers, regulation, and obligations to provide indigent care at a financial loss, providers lack the customary freedom of a seller of services to set its price. Unprofitable but vital medical services do not reap providers the usual market rewards. Many of the classic features of a free market are simply absent in the health care context, and that fact counsels caution when courts are urged to dismantle regulatory efforts to counter perceived gaps and inefficiencies in the healthcare market.[227]

A company like Amazon that is accustomed to large tax breaks from states hoping to attract new fulfillment centers or even its second headquarters[228] would no doubt be bewildered when faced with a state requirement that it justify the construction of Prime Health facilities.

V. (How) Should Hybrid Healthcare Be Regulated?

The preceding observations about the application of federal and state regulatory models to hybrid healthcare entities is primarily descriptive. Due to pragmatic constraints, I have assumed that these novel entities will not be met by a new all-embracing regulatory model.  After all, it is far too early to predict what the hybrids will look like or whether they will be sufficiently similar to make common regulation even possible.

Still, such a pragmatic conclusion conceals some potential concerns. For example, Frank Pasquale argues that “[major digital firms] are no longer market participants. Rather, in their fields, they are market makers, able to exert regulatory control over the terms on which others can sell goods and services.” [229] He sees an even more dystopian regulatory future as these firms “aspire to displace more government roles over time, replacing the logic of territorial sovereignty with functional sovereignty.”[230] However, the regulators are in retreat. Almost a generation of work building robust health insurance regulation culminating in the ACA and responsible consumer protection through the Consumer Financial Protection Bureau seems to be running on life support.[231]

Some of the hurdles (for example, CPM and CON) that Prime Health may have to navigate are state laws that themselves have anticompetitive effects. However, the more important competitive question may be how major technology companies themselves impact markets. In August 2018, Amazon became the second company to reach a trillion-dollar market valuation,[232] a few weeks after Apple achieved the same feat.[233] The healthcare ambitions of the two largest companies have not been lost on incumbents. Indeed, the CVS-Aetna merger,[234] itself an attempt to remake healthcare delivery, was  likely partially driven by the threat of Prime Health[235] and the government’s blocking of horizontal mergers between insurers.[236] The combined company could leverage CVS’s retail health clinics to provide low cost primary care and, by making it easier to reach patients, enable preventative care.[237] This would enable Aetna to partially disintermediate physician practices, particularly those who refer patients to their networked (and expensive) secondary and tertiary care providers.

Healthcare has experienced unprecedented levels of deal making in the years following the passage of the ACA. Between 2013 and 2017, almost twenty percent of U.S. hospitals merged with or were acquired by another hospital.[238] The result is that healthcare is arguably already over-concentrated and requires far more regulatory scrutiny to avoid declines in quality, cost containment,[239] and local services.[240]

A very real question arises as to how antitrust regulators will react to large hybrid companies entering the healthcare space. Amazon’s competitors and analysts act as though the company holds monopoly power. For example, after a report that Amazon had obtained wholesale pharmacy licenses, the stock prices of other drug distribution companies plunged.[241] But neither this “Amazon Effect” nor the company’s unprecedented wealth and market share make Amazon a dictionary definition monopoly.[242] Indeed, it prompted Lina Khan to argue that the prevailing “antitrust doctrine [that] views low consumer prices, alone, to be evidence of sound competition” [243] is a flawed model for assessing modern businesses such as Amazon. Khan argues that “a company’s power and the potential anticompetitive nature of that power cannot be fully understood without looking to the structure of a business and the structural role it plays in markets.”[244]

Khan’s observations are particularly salient for Prime Health. Will success in reducing costs relative to traditional healthcare businesses give ABJ a pass or perhaps even some sort of award for finally bending the healthcare cost curve? Or will incumbents successfully argue that (allegedly) predatory pricing, customer data leveraging, and horizontal and vertical integration require enhanced regulatory scrutiny? It appears that the current Department of Justice position is that there is no evidence that the large technology companies are harming competition.[245]

Antitrust aside, there are (or should be) three specific concerns regarding the Prime Health construct imagined herein. First, earlier it was noted how some technology firms working in the health space have apparently been granted special treatment through the FDA’s Pre-Cert program.[246] Overall, this seems to be an appropriate way to explore nimbler models of regulation that are better suited to rapidly innovating products. However, as the FDA’s use of Pre-Cert, mobile app enforcement discretion,[247] and other expedited regulatory processes (for example, de novo review applied to the ECG app in the new Apple Watch Series 4[248]) expands, the FDA will need to publish stronger guardrails to maintain our trust in the agency’s safety and efficacy reviews.

Second, as discussed above, there has always been a path-dependent, almost accidental quality to the involvement of employers in providing access to healthcare for their employees.[249] This structure can cause concerns or even conflicts of interest when employers opt to retain the insurance risk and self-fund their employees’ health insurance and, for example, use an insurer for ASO. Pre-ACA ERISA protections were designed to cabin some of the greatest risks associated with direct employer involvement in benefits decisions.[250] Similarly, the HIPAA Privacy Rule has specific provisions applying to employers who provide health insurance,[251] including the requirement that such employers build data “walls” between people or departments dealing with health insurance and, say, human resources.[252] Prime Health (and other businesses building out direct care models) may, however,  involve building a far closer relationship between employers and the provision of healthcare to employees. There is danger of information acquired for healthcare purposes surfacing elsewhere—as occurred, for example, during the outing of the AOL “distressed babies.”[253]

Third, and perhaps most importantly, there are growing concerns about how technology firms are subject to relatively little data protection regulation. This is especially alarming considering that technology companies will edge closer to our specific clinical data rather than “merely” aggregate healthcare data. The U.S. has a tradition of exceptional protection for healthcare data, surpassing that provided for data in other domains.[254] Because the regulation is triggered by domain participants rather than data type, technology companies are not required to protect healthcare data to the same degree as healthcare providers.[255] Often, domain-specific regulation has invited regulatory arbitrage.[256] Recently, Apple CEO Tim Cook went so far as to criticizes some of his technology company competitors as a “data industrial complex” engaged in surveillance.[257]

Our lack of long-term, imaginative policymaking combined with narrow, domain-specific legislation has created a fragmented, often incoherent regulatory environment. The optimal solution to the data protection issues raised by hybrid healthcare lies in the enactment of a U.S. analogue to the General Data Protection Regulation (GDPR), but the odds of that happening are low.[258] Modern legislation or regulation tends to be of the band-aid variety lest markets should become jittery, while building modern legislative and regulatory structures designed to deal with contemporary problems seems anathema to our political processes. Even the expansive California Consumer Privacy Act of 2018, the single most advanced, general (i.e., not domain-specific) privacy law enacted in the U.S. in a generation,[259] is embroiled in some debate as to whether it does or should apply to healthcare providers.[260] Notwithstanding, other recent California legislation dealing with the cybersecurity of connected devices does appear to cover the healthcare “information of things” that are being embraced by hybrid healthcare providers.[261] Although there has been some current interest in federal data protection legislation,[262] it is difficult to imagine a current Congressional majority that would favor anything more rigorous than the relatively light regulatory model involving enforceable codes of conduct that was proposed by the Obama Administration in 2015.[263]

There is much to like about the companies behind hybrid healthcare. We are enthralled by their innovation and the optimism they bring into the space. For now, we may even trust them (or at least some of them). But the fundamental regulatory truth is that we lack a “Plan B” if they decide to be evil.[264]


As both Microsoft founder Bill Gates[265] and President Donald Trump[266] have remarked, healthcare is “complicated. ” The Economist notes that “[i]t is worth remembering that the prospect of technology firms transforming health care has been heralded in the past, only to disappoint.”[267] Not surprisingly, commentators have greeted the new venture with equal amounts of skepticism[268] and cautious optimism.[269]

The U.S. healthcare ecosystem and econosystem are so complex that outsiders making even slight contact will face indeterminacies and barriers to entry that have their roots in both market and regulatory failures. As hybrid entities, particularly Amazon, enter the healthcare space and offer healthcare financing and healthcare services, they will also be sucked into healthcare’s regulatory morass.

Prime Health and the other hybrid insurgents will also experience pushback and even competition from incumbents. We will see more concentrations, primarily of the vertical type like the CVS-Aetna merger.[270] Others may be combinations of insurers and providers,[271] while others yet may be combinations of insurers, PBMs, and health systems.[272] There will also be insurgents other than technology companies.[273] However, in the case of the traditional healthcare entities, some skepticism is appropriate as to whether they will change their ways or whether they are merely building defensive positions by combining two different levels of healthcare entity to increase customer lock-in. As the New York Times opined, “[they’re] taking a zebra and a zebra. . . . What they want to become is a unicorn.”[274]

Given its corporate provenance, Prime Health cannot, and indeed should not, take the path well-trodden. As the ACA limps along, we will not see attempts at major health reform any time soon. This is the time for those with extraordinary market power to step forward, to be audacious and to design a true functioning healthcare marketplace. If Prime Health fails, it will lend support to those who argue that consumer-directed and competitive solutions do not work. If it succeeds, Prime Health’s architects will be elevated to the health policy Parthenon to stand next to Otto von Bismarck and William Beveridge.

David Blumenthal’s qualified enthusiasm seems to strike the appropriate tone: “The excitement about these . . . bold new health care arrangements says as much about the desperation with our current health care systems as it does about the promise of the mergers themselves.”[275] Hybrid healthcare, particularly Prime Health, may well be remembered as a final attempt to make employer-provided healthcare affordable and sustainable when the better path forward is arguably for employers to persuade the federal government that employment-based insurance is not the answer; the preferable solution is government-provided insurance. However, until that epitaph is written, it will be a fascinating journey to observe.

* © 2018 Nicolas Terry. All rights reserved. I express my thanks to Seema Mohapatra, Efthimios Parasidis, and Valerie Blake for their thoughtful comments on an earlier draft and to Emily Beukema for her editorial and research assistance.

** Hall Render Professory of Law, Executive Director, Hall Center for Law and Health, Indiana University Robert H. McKinney School of Law. Email:

[1] Press Release, Amazon, Berkshire Hathaway & JPMorgan Chase & Co., Amazon, Berkshire Hathaway and JPMorgan Chase & Co. to Partner on U.S. Employee Healthcare (Jan. 30, 2018),

[2] Id.

[3] Carolyn Y. Johnson, Amazon, Berkshire Hathaway and JP Morgan Chase Join Forces to Tackle Employees’ Health-care Costs, Wash. Post: Wonkblog (Jan. 30, 2018),

[4] Doctor You, Economist, Feb. 3, 2018, at 11.

[5] Eugene Kim & Christina Farr, Amazon 1492: Secret Health Tech Project, CNBC (July 26, 2017, 5:30 PM),

[6] Jessica Bartlett, Health Care Venture Co-Founded by Amazon to Be Based in Boston, Bos. Bus. J. (June 20, 2018, 10:09 AM),

[7] Vindell Washington et al., The Hitech Era and the Path Forward, 377 New Eng. J. Med. 904 (2017).

[8] See Nicolas P. Terry, Information Technology’s Failure to Disrupt Healthcare, 13 Nev. L.J. 722 (2013).

[9] Natasha Singer, How Big Tech Is Going After Your Health Care, N.Y. Times (Dec. 26, 2017), (“In the first 11 months of [2017], 10 of the largest tech companies in the United States were involved in health care equity deals worth $2.7 billion, up from just $277 million for all of 2012.”).

[10] Ann M. Marciarille, Uberizing Nonemergency Medical Transportation, PrawfsBlawgs (Jan. 19, 2018, 8:31 PM),

[11] See Chiedozie Udeh et al., Telemedicine/Virtual ICU: Where Are We and Where Are We Going?, 14 Methodist DeBakey Cardiovascular J. 126 (2018).

[12] E.g., Christina Farr, Facebook Sent a Doctor on a Secret Mission to Ask Hospitals to Share Patient Data, CNBC (Apr. 5, 2018, 2:01 PM), (examining evidence from several sources on Facebook’s desire to link their data with healthcare data to improve services).

[13] Singer, supra note 9, at 2.

[14] Jake Swearingen, Why Is Amazon Building Brick-and-Mortar Bookstores? N.Y. Mag.: Intelligencer (June 1, 2017),

[15] J.J. McCorvey, The Race Has Just Begun, Fast Company, Sept. 2013, at 68 (discussing how AmazonFresh grocery products can be purchased online and delivered along with other goods from Amazon’s non-grocery product selection).

[16] Nick Wingfield & Michael J. de la Merced, Amazon to Buy Whole Foods for $13.4 Billion, N.Y. Times (June 16, 2017),

[17] Either Apple has morphed from a computer company into a consumer electronics company or computers have become consumer electronics.

[18] Surgical Intervention, Economist, Feb. 3, 2018, at 59.

[19] Id. at 60.

[20] Christina Farr, Apple Is Launching Medical Clinics to Deliver the ‘World’s Best Health Care Experience’ to Its Employees, CNBC (Feb. 27, 2018),

[21] See, e.g., Alex Webb, Apple Is Developing an EKG Heart Monitor for Its Smartwatch, Bloomberg (Dec. 21, 2017, 12:00 PM),

[22] See Press Release, Apple, Apple Announces Effortless Solution Bringing Health Records to iPhone (Jan. 24, 2018), (discussing an updated to Apple’s Health app which allows users to view their medical records on their iPhone); see also Apple, Apple Opens Health Records API to Developers (Jun. 4, 2018),

[23] David Blumenthal & Aneesh Chopra, Apple’s Pact with 13 Health Care Systems Might Actually Disrupt the Industry, Harv. Bus. Rev. (Mar. 23, 2018),

[24] Id.

[25] Lucas Mearian, Apple’s Health Record API Released to Third-Party Developers; Is it Safe?, Computerworld (June 7, 2018, 3:11 AM), (discussing Apple’s API which allows users to transfer their electronic health records to iPhone and share between providers).

[26] Surgical Intervention, supra note 18.

[27] Jillian D’Onfro, Alphabet’s Bets on Health Tech Are Paying Off as Portfolio Companies Go Public, CNBC (Mar. 30, 2018, 1:44 PM),

[28] Verily, (last visited Oct. 18, 2018).

[29] Christina Farr, The New Google Health Unit Is Absorbing Health Business from DeepMind, Alphabet’s AI Research Group, CNBC (Nov. 13, 2018, 11:15 AM),

[30] About DeepMind Health, DeepMind, (last visited Oct. 18, 2018).

[31] Royal Free – Google DeepMind Trial Failed to Comply with Data Protection Law, ico (Jul. 3, 2017),

[32] Our Story, Cityblock, (last visited Oct. 18, 2018).

[33] Brian Fung, CVS’s $69 Billion Merger with Aetna Is Approved in Deal that Could Transform Health-Care Industry, Wash. Post (Oct. 10, 2018), (discussing the CVS-Aetna merger).

[34] Ben Thompson, The Amazon Tax, Stratechery (Mar. 15, 2016), (using the term “first and best customer” to describe the practice wherein Amazon built AWS for internal use first before making AWS available to customers externally).

[35] See, e.g., Robert Langreth & Zachary Tracer, Buffett-Bezos Health Plan Will Hinge on Buying Power, Technology, Bloomberg (Jan. 30, 2018, 10:39 AM),

[36] Surgical Intervention, supra note 18.

[37] Caroline Humer, Fed Up with Rising Costs, Big U.S. Firms Dig into Healthcare, Reuters (Jun. 11, 2018, 6:07 AM),

[38] Jason Del Rey, Amazon’s AWS and Advertising Businesses Are Fueling Its Retail Dominance, recode (Apr. 26, 2018, 6:55 PM),; Ron Miller, How AWS Came to Be, TechCrunch (July 2, 2016),

[39] Laura Stevens, Amazon to Launch Delivery Service That Would Vie with FedEx, UPS, Wall St. J. (Feb. 9, 2018),

[40] Ben Thompson, Amazon’s New Customer, Stratechery (June 19, 2017),

[41] See, e.g., Carolyn Y. Johnson, What Amazon Could Do to the Business of Selling Prescription Drugs, Wash. Post (May 17, 2017),; Robert Langreth & Spencer Soper, This Is How Amazon Could Invade the Pharmacy, Bloomberg (Nov. 7, 2017, 10:10 AM),

[42] See Christina Farr, Amazon Has Plans to Open Its Own Health Clinics for Seattle Employees, CNBC (Aug. 9, 2018, 12:46 PM),; see also Christina Farr, Amazon Just Hired a Top Seattle Doctor Who Ran a Network of Health Clinics, CNBC (Jan. 19, 2018, 5:14 PM),

[43] Christina Farr, Amazon is Hiring a Former FDA Official to Work on Its Secretive Health Tech Business, CNBC (Mar. 16, 2018, 5:14 PM),

[44] Christina Farr, Amazon Hires a Star Cardiologist to Help Its Push into Health, CNBC, (Aug. 20, 2018, 2:10 PM),

[45] Nicolas P. Terry, Prime Health: Should Amazon Purchase a Hospital Chain?, Medium (July 30, 2017),

[46] Kenneth J. Arrow, Uncertainty and the Welfare Economics of Medical Care, 53 Am. Econ. Rev. 941 (1963).

[47] See, e.g., Yoni Blumberg, 70% of Americans Now Support Medicare-For-All—Here’s How Single-Payer Could Affect You, CNBC (Aug. 28, 2018, 1:05 PM),

[48] See, e.g., Carrie Hojnicki, How Casper Disrupted the Mattress Industry, Architectural Dig. (Mar. 21, 2017),; Warren Shoulberg, What’s Next for Mattress Firm After Bankruptcy and Putting 700 Stores to Sleep?, Forbes, (Oct. 5, 2018, 12:54 PM),

[49] See Michael K. Spencer, W*F is Amazon?, Medium (Nov. 1 2017),; see also Shira Ovide, How Amazon’s Bottomless Appetite Became Corporate America’s Nightmare, Bloomberg (Mar. 14, 2018),

[50] Ben Thompson, Amazon Health, Stratechery (Jan. 31, 2018),

[51] The World’s Most Valuable Resource Is No Longer Oil; Regulating the Data Economy, Economist, May 6, 2017, at 9.

[52] Alex Hern, Why Data Is the New Coal, Guardian (Sept. 27, 2016, 6:26 AM),

[53] Wendy N. Epstein, The CVS/Aetna Deal: The Promise in Data Integration, Bill of Health (Dec. 13, 2017),

[54] Robert J. Samuelson, Opinion, Is This the Future of Health Care?, Wash. Post (Nov. 1, 2017),

[55] Doctor You, supra note 4.

[56] Christina Farr, The Amazon-Berkshire-JPM Health Venture Led by Atul Gawande Has a COO: Jack Stoddard, CNBC (Sept. 4, 2018, 2:28 PM),

[57] Michael J. Coren, There’s Precedent for Amazon Competing with So Many Companies. It Doesn’t End Well., Quartz (Oct. 28, 2017),

[58] Id.

[59] Steven J. Spear, Fixing Healthcare from the Inside, Today, Harv. Bus. Rev., Sept. 2005, at 1.

[60] Terry, supra note 45.

[61] Department of Medical Supplies & Equipment, Amazon, (last visited Nov. 3, 2018).

[62] Introducing Prime Now Free 2-Hour Delivery, Bartell Drugs, (last visited Nov. 3, 2018) (offering non-prescription drugs through Prime Now).

[63] Shusuke Murai, Amazon Launches Same-Day Delivery Service for Food and Medicine, Japan Times (Apr. 19, 2017),

[64] See Christina Farr, Amazon is Hiring People to Break into the Multibillion-Dollar Pharmacy Market, CNBC (May 16, 2017, 8:23 PM), (discussing the variety of actions Amazon has taken to develop pharmaceutical industry expertise).

[65] Press Release, Amazon, Amazon to Acquire PillPack (Jun. 28, 2018),

[66] See Coren, supra note 57; see also Method and Sys. for Anticipatory Package Shipping, U.S. Patent No. 8,615,473 (filed Aug. 24, 2012).

[67] Sarah Kliff, The Fax of Life: Why American Medicine Still Runs on Fax Machines, Vox (Jan. 12, 2018, 3:53 PM),

[68] A. Gary Shilling, Health Care’s Waiting Room Problem, Forbes (Apr. 20, 2016),

[69] See, e.g., Eric Feinberg, How Amazon Is Investing in Customer Experience by Reimagining Retail Delivery, Forbes (Jan. 4, 2018, 9:00 AM),

[70] See Eugene Wei, Invisible Asymptotes, Remains of the Day (May 22, 2018), (discussing how Amazon’s investment in customer-focused researched uncovered consumer dislike for paying shipping fees; by “eliminating” shipping fees with Amazon Prime, the company drastically raised their growth asymptote).

[71] Letter from Jeff Bezos, Chief Exec. Officer,, Inc., to Amazon’s Shareholders (Apr. 18, 2018) (on file with the SEC), available at

[72] See Paul Rosen, The Patient as Consumer and the Measurement of Bedside Manner, NEJM Catalyst (Mar. 20, 2017), (discussing research which suggests that physicians, at least compared to service providers in other industries, place relatively less weight on the importance of customer satisfaction and experience); see also Augusta Meill & Gianna Ericson, The Trouble with Treating Patients as Consumers, Harv. Bus. Rev. (Jan. 9, 2012),

[73] See generally Alexa Skills, Amazon, (last visited Oct. 31, 2018).

[74] Echo Look, Amazon, (last visited Oct. 31, 2018).

[75] Echo Show, Amazon, (last visited Oct. 31, 2018).

[76] Tom Simonite, Amazon Wants Alexa to Hear Your Whispers and Frustration, Wired (Sept. 20, 2018), (“[Amazon] is experimenting with giving Alexa a rudimentary form of emotional awareness, enabling it to listen for the sound of frustration in a person’s voice.”).

[77] Christina Farr, The Bezos-Buffett-Dimon Joint Venture to Save Health Care Is Struggling to Find a CEO, CNBC (May 16, 2018),

[78] Press Release, Amazon, Berkshire Hathaway & JPMorgan Chase & Co., Amazon, Berkshire Hathaway and JPMorgan Chase Appoint Dr. Atul Gawande as Chief Executive Officer of Their Newly-Formed Company to Address U.S. Employee Healthcare (June 20, 2018),

[79] Atul Gawande, The Cost Conundrum, New Yorker, June 1, 2009.

[80] Atul Gawande, Cowboys and Pit Crews, New Yorker (May 26, 2011),

[81] Atul Gawande, What Big Medicine Can Learn from the Cheesecake Factory, New Yorker (Aug. 13, 2012),

[82] Gawande, supra note 80.

[83] See Surgery: WHO Safe Surgery Checklist, Ariande Labs (last visited Oct. 31, 2018).

[84] Alex B. Haynes et al., A Surgical Safety Checklist to Reduce Morbidity and Mortality in a Global Population, 360 New Eng. J. Med. 491, 491 (2009).

[85] See Alex B. Haynes et al., Mortality Trends After a Voluntary Checklist-Based Surgical Safety Collaborative, 266 Annals Surgery 923, 923 (2017); see also George Molina et al., Perception of Safety of Surgical Practice Among Operating Room Personnel from Survey Data Is Associated with All-Cause 30-day Postoperative Death Rate in South Carolina, 266 Annals Surgery 658, 658 (2017).

[86] Deena Beasley, Surgery, Not Pharma, Is Biggest Healthcare Cost Worry: Gawande, Ins. J. (June 22, 2018),

[87] See, e.g., Donald Berwick et al., The Triple Aim: Care, Health, and Cost, 27 Health Aff. 759, 759-69 (2008); Donald Berwick et al., No Toyota yet, but a Start, 35 Modern Healthcare 18, 18-19 (2005).

[88] See, e.g., Jesse Pines et al., Brookings Inst., Kaiser Permanente–California: A Model for Integrated Care for the Ill and Injured (2015),

[89] Atul Gawande, Testing, Testing, New Yorker, Dec. 14, 2009, at 34.

[90] Alain C. Enthoven, Would Reform Bills Control Costs? A Response to Atul Gawande, Health Aff. (Dec. 22, 2009),

[91] Thompson, supra note 50.

[92] See, e.g., Drew Altman, Don’t Overhype the New Health Care Venture, Axios (Feb. 8, 2018),

[93] Surgical Intervention, supra note 18.

[94] Reed Abelson & Katie Thomas, Fed Up With Drug Companies, Hospitals Decide to Start Their Own, N.Y. Times, Jan. 18, 2018, at B1.

[95] Leading U.S. Health Systems Announce Plans to Develop a Not-for-profit Generic Drug Company, Intermountain Healthcare (Jan. 18, 2018),

[96] See, e.g., Wenke Hwang et al., Effects of Integrated Delivery System on Cost and Quality, 19 Am. J. Managed Care e175, e178-82 (2013).

[97] See Austin Frakt, The Performance of Integrated Delivery Systems, AcademyHealth (July 20, 2016), (discussing several studies showing that integrated health service networks often fail to reduce costs while only occasionally improving quality of care).

[98] See David Atkins et al., The Veterans Affairs Experience: Comparative Effectiveness Research in A Large Health System, 29 Health Aff. 1906, 1906-07 (2010); see also Gretchen A. Jacobson et al., Cong. Research Serv., RL33802, Pharmaceutical Costs: A Comparison of Department of Veterans Affairs (VA), Medicaid, and Medicare Policies 8-14 (2007).

[99] U.S. Dept. Veterans Affairs, VHA Formulary Management Process (2009),

[100] See Katie Gudiksen, State Medicaid Programs Are a Tool to Address Rising Drug Costs, Source on Healthcare Price & Competition: Source Blog (May 8, 2018), In June 2018, CMS refused Massachusetts’ Medicaid waiver request. See Letter from Angela D. Garner, Dir., Div. Sys. Reform Demonstrations, Ctrs. Medicare & Medicaid Servs., to Daniel Tsai, Assistant Sec’y & Dir. MassHealth, Mass. Exec. Office Health & Human Servs. (Oct. 23, 2018) (on file with Ctrs. Medicare & Medicaid Servs.), available at

[101] In 2017, Amazon employed 566,000 people worldwide. See Kurt Schlosser, Amazon Now Employs 566,000 People Worldwide—a 66 Percent Jump from a Year Ago, GeekWire (Feb. 1, 2018), As of June 2018, JPMorgan Chase & Co. employed 252,539 people. See Company Profile for JPMorgan Chase & Co., Forbes, (last updated June 6, 2018). As of November 2017, Berkshire Hathaway employed approximately 377,291 people. See Company Profile for Berkshire Hathaway, Inc., Bloomberg, (last updated Nov. 20, 2018).

[102] See supra Part I.C.

[103] A Conversation with Surgeon, Author, and Researcher Atul Gawande, Aspen Institute (June 23, 2018) [hereinafter Conversation with Atul Gawande],

[104] Id.

[105] Id.

[106] Id.

[107] Bezos, supra note 71.

[108] Nick Wingfield, Pitching Deal, Amazon Bids for Shoppers on a Budget, N.Y. Times, Mar. 8, 2018, at B2 (discussing discounted Amazon Prime membership for low-income Americans).

[109] 26 U.S.C. § 36B (2012) (establishing tax credits designed to help low-income individuals and families afford health insurance purchased through the Health Insurance Marketplace).

[110] See, e.g., Michael E. Porter & Elizabeth Teisberg, Redefining Competition in Health Care, Harv. Bus. Rev., June 2004, at 66.

[111] Dave Chase, Overcoming Healthcare Organizations’ Dangerous Zero Sum Thinking, Forbes (July 12, 2015, 4:45 PM),

[112] See, e.g., Lydia DePillis, It’s Amazon’s World. We Just Live in It., CNN (Oct 4. 2018),

[113] David Pogue, Technology’s Friction Problem, 306 Sci. Am. 28, 28 (2012).

[114] See generally MaryBeth Musumeci et al., Medicaid and Work Requirements: New Guidance, State Waiver Details and Key Issues, Henry J. Kaiser Family Found. (Jan. 16, 2018),

[115] Elise Gould, Econ. Pol’y Inst., A Decade of Declines in Employer-Sponsored Health Insurance Coverage 2 (2012),

[116] Zachary Tracer, Rising Health-Insurance Costs Are Eating Into Employees’ Paycheck Gains, Bloomberg (Sept. 19, 2017, 12:30 PM),

[117] Sarah R. Collins et al., Commonwealth Fund, How Well Does Insurance Coverage Protect Consumers from Health Care Costs? 4-6 (2017),

[118] William H. Frist, Obamacare’s ‘Cadillac Tax’ Could Help Reduce the Cost Of Health Care, Forbes (Feb. 26, 2014, 7:37 AM),

[119] Id.; 26 U.S.C. § 4980I (2012).

[120] Sherry Glied & Adam Striar, Looking Under the Hood of the Cadillac Tax, 15 Commonwealth Fund 1, 4-6 (2016).

[121] See, e.g., Bruce Japsen, Big Employers Win Delay for Obamacare’s Cadillac Tax Once Again, Forbes (Jan. 23, 2018, 4:30 PM),

[122] See, e.g., Nicolas P. Terry, Pit Crews with Computers: Can Health Information Technology Fix Fragmented Care?, 14 Hous. J. Health L. & Pol’y 129, 139-40, 162–64 (2014).

[123] See generally Terry, supra note 8, at 754, 756.

[124] Clayton M. Christensen et al., Seeing What’s Next: Using the Theories of Innovation to Predict Industry Change 197 (2004).

[125] Jack Curran, IBISWorld, Health & Medical Insurance in the US 38 (2018).

[126] Bertha Coombs, As Obamacare Twists in Political Winds, Top Insurers Made $6 Billion (Not That There Is Anything Wrong with That), CNBC (Aug. 5, 2017, 9:39 AM),

[127] Elisabeth Rosenthal, An American Sickness: How Healthcare Became Big Business and How You Can Take It Back 20-21 (2017).

[128] Alain C. Enthoven & Sara J. Singer, The Managed Care Backlash and the Task Force in California, 17 Health Aff. 95, 96-97 (1998).

[129] See Robert J. Blendon et al., Understanding the Managed Care Backlash, 17 Health Aff. 80, 83, 90 (1998); see also Managed Care: What Went Wrong? Can It Be Fixed?, Stan. Graduate Sch. Bus. (Nov. 1, 1999),

[130] Amitabh Chandra et al., Is This Time Different? The Slowdown in Health Care Spending, Brookings Papers on Econ. Activity, Fall 2013, at 272-74.

[131] Gerard F. Anderson et al., It’s the Prices, Stupid: Why the United States Is SO Different from Other Countries, 22 Health Aff. 89, 90-103 (2003).

[132] Robert Langreth, Drug Prices, Bloomberg (Jan. 25, 2016),

[133] Tara Bannow, Report Finds Hospitals Drive Overall Healthcare Price Growth, Mod. Healthcare (Apr. 13, 2018),

[134] Dean Baker, The Problem of Doctors’ Salaries, Politico, (Oct. 25, 2017, 5:03 AM),

[135] Christy F. Chapin, How Health Care Went Wrong, N.Y. Times, June 19, 2017, at A19.

[136] Patient Protection and Affordable Care Act § 2718, 42 U.S.C. § 300gg-18 (2012).

[137] Id. § 3403, 42 U.S.C. § 1395kkk-1 (repealed 2018).

[138] Editorial Board, The So-called Obamacare Death Panel Meets Its Unfortunate End, Wash. Post, (Feb. 9, 2018),

[139] Explaining Health Care Reform: Medical Loss Ratio (MLR), Henry J. Kaiser Family Found. (Feb. 29, 2012),

[140] Sahil Kapur, Why Republicans Are Fighting to Repeal Obama’s Medicare Cost-Cutting Board, Talking Points Memo (Feb. 28, 2012, 12:29 AM),

[141] Mike DeBonis & Erica Werner, Brief Government Shutdown Ends as Trump Signs Spending Bill, Wash. Post: Powerpost (Feb. 9, 2018),

[142] Editorial Board, supra note 138.

[143] Ben Thompson, The Great Unbundling, Stratechery (Jan. 18, 2017),

[144] Id.

[145] See 42 U.S.C. § 1395w–111(i) (2012); see also Tony Abraham, Azar Says Admin ‘Does Not Believe in’ Medicare Drug Price Negotiations, Healthcare Dive (May 16, 2018),

[146] Malcolm Harris, The Singular Pursuit of Comrade Bezos, Medium (Feb. 15, 2018),

[147] Thompson, supra note 50.

[148] Id.

[149] Patient Protection and Affordable Care Act § 1311(b), 42 U.S.C. § 18031 (“Each State shall . . . establish an American Health Benefit Exchange . . . .”). See generally What is the Health Insurance Marketplace?, U.S. Dep’t Health & Hum. Servs., (last visited Nov. 1, 2018).

[150] Robert Pear, Clinton’s Health Plan: The Overview; Congress is Given Clinton Proposal for Health Care, N.Y. Times, Oct. 28, 1993, at A1. See generally Alain C. Enthoven, The History and Principles of Managed Competition, 12 Health Aff. 24 (1993).

[151] Durable Medical Equipment, Amazon, (last visited Oct. 22, 2018).

[152] See generally Trisha Torrey, Primary, Secondary, Tertiary, and Quaternary Care, verywellhealth (Sep. 18, 2018), (defining primary, secondary, tertiary, and quaternary care).

[153] See Anna W. Mathews, GM Cuts Different Type of Health-Care Deal, Wall St. J. (Aug. 6, 2018, 9:52 PM),

[154] Brie Zeltner, Walmart to Send Employees to Cleveland Clinic for Heart Care, Plain Dealer, Oct. 12, 2012, at B2.

[155] See generally Andrew Austin & Thomas L. Hungerford, Cong. Research Serv., R40834, The Market Structure of the Health Insurance Industry 22, 28, 36 (2010).

[156] See Thompson, supra note 40.

[157] See generally Terry, supra note 45.

[158] Reinhard Busse & Miriam Blümel, Health Systems in Transition, 16 Ger.: Health Sys. Rev. 1, 44 (2014) (describing the German system wherein physicians accredited to treat patients under the German statutory health insurance scheme are organized into mandatory regional associations).

[159] See e.g., Michael E. Porter et al., Redesigning Primary Care: A Strategic Vision to Improve Value by Organizing Around Patients’ Needs, 32 Health Aff. 516 (2013).

[160] See generally Mark Graban, #Lean: Clarifying Push, Pull, and Flow in a Hospital; the Patient “Pulls, Lean Blog (Feb. 24, 2014), (“‘Pull’ is the concept of basically building in response to actual customer demand . . . . Pull reduces costs, improves quality, and makes sure the customer gets what they want, when they need it, in the right quantities.”).

[161] See generally Nicolas P. Terry, Appification, AI, and Healthcare’s New Iron Triangle, 20 J. Health Care L. & Pol’y 117, 178-79 (2018).

[162] See Robert I. Field, Why is Health Care Regulation So Complex?, 33 Pharmacy and Therapeutics 607 (2008).

[163] See supra Part I.

[164] See e.g., Information for Pharmacists, Pharmacies, and Pharmacy Technicians, Texas State Board of Pharmacy, (last visited Oct. 22, 2018).

[165] Taylor Soper, Walgreens to Shut Down, 4 Years After $429M Acquisition, GeekWire, (July 28, 2016, 10:14 AM),

[166] See Spear, supra note 59.

[167] See Bezos, supra note 71.

[168] Scot Hornick & Shri Santhanam, How Smart Speakers Are Poised to Reinvent the Travel Industry, Harv. Bus. Rev. (Aug. 7, 2018),

[169] Ben F. Rubin, Amazon’s Alexa Assistant Now Works with Over 20K Devices, cnet (Sept. 1, 2018, 6:41 AM),

[170] See generally Terry, supra note 161.

[171] Federal Food, Drug, and Cosmetic Act § 201(h), 21 U.S.C. § 321(h)(2) (2012).

[172] See generally Nicolas P. Terry, Mobile Health, 147 CHEST J. 1429 (2015); Nicholas P. Terry & Tracy D. Gunter, Regulating Mobile Mental Health Apps, 36 Behav. Sci. L. 136 (2018).

[173] 21st Century Cures Act § 3060, 21 U.S.C. § 360j(o).

[174] U.S. Food & Drug Admin., Changes to Existing Medical Software Policies Resulting from Section 3060 of the 21st Century Cures Act: Draft Guidance for Industry and Food and Drug Administration Staff (2017),

[175] See generally Digital Health Software Precertification (Pre-Cert) Program, U.S. Food & Drug Admin., (last updated Sept. 27, 2018); Nathan G. Cortez et al., Questions About The FDA’s New Framework For Digital Health, Health Aff.: Health Aff. Blog (Aug. 16, 2017),

[176] See Digital Health Software Precertification (Pre-Cert) Program, supra note 175.

[177] See, e.g., Dylan Curran, Are you ready? Here is all the data Facebook and Google have on you, Guardian (Mar. 28, 2018, 6:00 AM),; Hillary Grigoris, 9 Things to Know About Facebook Privacy and Cambridge Analytica, Digital Trends (Apr. 5, 2108, 8:06 AM),; Alyssa Newcomb, A Timeline of Facebook’s Privacy Issues—and its Responses, NBC News (Mar. 24, 2018, 7:02 AM),

[178] Jack Morse, Facebook Isn’t the Only Tech Company with Too Much of Your Data, Mashable (Apr. 14, 2018),

[179] See generally Nicolas P. Terry, Regulatory Disruption and Arbitrage in Health-Care Data Protection, 17 Yale J. Health Pol’y L. & Ethics 143, 173-84 (2017) (noting the consumer healthcare domain is less regulated than the professional domain).

[180] See generally Privacy and Security Enforcement, Fed. Trade Comm’n, (last visited Oct. 27, 2018).

[181] See Terry, supra note 179, at 168-73; see also Nicolas P. Terry & Lindsay F. Wiley, Liability for Mobile Health and Wearable Technologies, 25 Annals Health L. 62 (2016); Nicolas P. Terry, Will the Internet of Things Transform Healthcare?, 19 Vand. J. Ent. & Tech. L. 327 (2016).

[182] 45 C.F.R pts. 160, 162, 164 (2018).

[183] See, e.g., 45 C.F.R. § 164.502.

[184] A small exception exists for non-healthcare data custodians who hold personal health records who are subject to a special data notification rule. See Complying with the FTC’s Health Breach Notification Rule, Fed. Trade Comm’n, (last visited Oct. 27, 2018).

[185] The HIPAA Privacy Rule defines a “covered entity” as follows:

Covered entity means:

(1) A health plan.

(2) A health care clearinghouse.

(3) A health care provider who transmits any health information in electronic form in connection with a transaction covered by this subchapter.

45 C.F.R § 160.103.

[186] Id.

[187] The HIPAA Privacy Rule defines “group health plan” as follows:

Group health plan . . . means an employee welfare benefit plan . . . including insured and self-insured plans, to the extent that the plan provides medical care . . . , including items and services paid for as medical care, to employees or their dependents directly or through insurance, reimbursement, or otherwise, that:

(1) Has 50 or more participants . . . ; or (2) Is administered by an entity other than the employer that established and maintains the plan.


[188] See, e.g., 45 C.F.R. § 164.504(f).

[190] See HIPPA, Amazon Web Services, (last visited Oct. 17, 2018).

[191] Jack Murtha, Amazon’s Alexa Really Isn’t Ready For Healthcare, Healthcare Analytics News (May 24, 2018),

[192] See McCarran-Ferguson Act, 15 U.S.C. §§ 1011-15 (2012).

[193] Maria Terekhova, Amazon Pushes Further into Insurance with its Latest Investment, Bus. Insider (Jan. 5, 2018, 9:20 AM),

[194] 15 U.S.C. §§ 1011-12.

[195] Employee Retirement Income Security Act of 1974, Pub. L. No. 93-406, 88 Stat. 829 (codified as amended in scattered sections of 26 and 29 U.S.C.).

[196] 29 U.S.C. § 1104. See generally U.S. Dep’t. Lab., Emp. Benefits Security Admin., Understanding Your Fiduciary Responsibilities Under a Group Health Plan (2015),

[197] 29 U.S.C. § 1144.

[198] See generally Abbe R. Gluck et al., ERISA: A Bipartisan Problem for the ACA and the AHCA, Health Aff.: Health Aff. Blog (June 2, 2017),

[199] Gobeille v. Liberty Mut. Ins. Co., 136 S. Ct. 496 (2015).

[200] Dennis Green & Anaele Pelisson, This Map of Amazon’s Warehouse Locations Shows How It’s Taking Over America, Bus. Insider (Sept. 27, 2017, 12:14 PM),

[201] Thad Rueter, Amazon to Open Another Distribution Center in Indiana, Digital Com. 360 (Mar. 27, 2012), (noting $2 million in tax credits and $300,000 in training grants from the Indiana Economic Development Corp.).

[202] Quill Corp. v. North Dakota, 504 U.S. 298, 305-08 (1992) (finding states could not collect taxes from sellers lacking a physical presence in the state).

[203] South Dakota v. Wayfair, Inc., 138 S. Ct. 2080 (2018) (overruling Quill Corp., 504 U.S. 298).

[204] Chris Isidore, Amazon to Start Collecting State Sales Taxes Everywhere, CNN (Mar. 29, 2017, 2:59 PM),

[205] Conversation with Atul Gawande, supra note 103.

[206] See Credentialing, Licensing, and Education, Nat’l Ctr. for Complementary and Integrative Health, (last visited Oct. 28, 2018) (discussing state to state variations in credentialing and licensing standards);  see also N.C. State Bd. of Dental Exam’rs v. FTC, 135 S. Ct. 1101 (2015).

[207] See, e.g., Telemedicine Laws and Developments: A State-by-State Analysis, Becker’s Hosp. Rev. (Aug. 22, 2014), (last visited Nov. 21, 2018).

[208] See, e.g., Health, Nat’l Conf. of St. Legislatures, (last visited November 4, 2018).

[209] See, e.g., Ind. Code §§ 16-18-2-179, 16-21-2, 25-22.5-1-1.1, 27-8-10-1 (2018).

[210] See, e.g., 844 Ind. Admin. Code 5-1-1 et seq. (2018).

[211] See, e.g., Ind. Code § 25-22.5-2-1; see also Bever v. State Bd. of Registration for Healing Arts, No. WD 57880, 2001 WL 68307 (Mo. Ct. App. Jan. 30, 2001) (state board disciplinary inquiry into alleged substandard care).

[212] Compare Sheeley v. Mem’l Hosp., 710 A.2d 161 (R.I. 1998) (national standard of care), with Hagedorn v. Peterson, 690 N.W.2d 84 (Iowa 2004) (retaining the “locality” rule).

[213] Compare Largey v. Rothman, 540 A.2d 504 (N.J. 1988) (prudent patient standard), with Eady v. Lansford, 92 S.W.3d 57 (Ark. 2002) (physician standard), and Ind. Code §§ 34-18-12-2 to -3 (rebuttable presumption of consent from signing consent form).

[214] Compare Larson v. Wasemiller, 738 N.W.2d 300 (Minn. 2007) (explaining the limited duty of “negligent credentialing”), with Thompson v. Nason Hosp., 591 A.2d 703, 707-08 (Pa. 1991) (explaining the broad application of corporate negligence). See generally Barry R. Furrow, Enterprise Liability and Health Care Reform: Managing Care and Managing Risk, 39 St. Louis U. L.J. 79 (1994).

[215] See generally William C. Tait, The Legal Definition of the Practice of Medicine, 2 Cal. St. J. Med. 119, 119–21 (1904).

[216] See Paul Starr, The Social Transformation of American Medicine 428-36 (1982).

[217] See id. at 102-11.

[218] Corporate Practice of Medicine, Med. Board. Cal., (last visited Nov. 5 2018) (“The policy . . . is intended to prevent unlicensed persons from interfering with or influencing the physician’s professional judgment.”).

[219] For example, California considers the following to be unlicensed practice of medicine if performed by an unlicensed person:

Determining what diagnostic tests are appropriate for a particular condition.

Determining the need for referrals to, or consultation with, another physician/specialist.

Responsibility for the ultimate overall care of the patient, including treatment options available to the patient.

Determining how many patients a physician must see in a given period of time or how many hours a physician must work.


[220] Carter-Shields v. Alton Health Inst., 777 N.E.2d 948, 956 (Ill. 2002).

[221] Nicole Huberfeld, Be Not Afraid of Change: Time to Eliminate the Corporate Practice of Medicine Doctrine, 14 Health Matrix 243, 243-45 (2004).

[222] Hospital Survey and Construction (Hill-Burton) Act, Pub. L. No. 79-725, 60 Stat. 1040 (1946) (codified as amended at 42 U.S.C. §§ 291 to 291o1-1 (2012)).

[223] CON-Certificate of Need State Laws, Nat’l Conf. of St. Legislatures (Aug. 17, 2018), See generally The Hospital Survey and Construction Act, 132 J. Am. Med. Ass’n 148 (1946).

[224] CON-Certificate of Need State Laws, supra note 223.

[225] See id; see also Matthew D. Mitchell & Christopher Koopman, 40 Years of Certificate-of-Need Laws Across America, Mercatus Ctr. at Geo. Mason U., (Sept. 27, 2016),

[226] Fed. Trade Comm’n, Joint Statement of the Federal Trade Commission and the Antitrust Division of the U.S. Department of Justice to the Virginia Certificate of Public Need Work Group (2015),

[227] Colon Health Ctrs. of Am., LLC v. Hazel, 813 F.3d 145, 158 (4th Cir. 2016) (challenging unsuccessfully Virginia’s CON law invoking the dormant commerce clause).

[228] Elizabeth Weise, Amazon Second Headquarters Search Has Become a Cultural Meme, A Year After It Began, USA Today (Sept. 5, 2018, 5:00 AM),

[229] Frank Pasquale, From Territorial to Functional Sovereignty: The Case of Amazon, L. & Pol. Econ. (Dec. 6, 2017),

[230] Id.

[231] See, e.g., Jonathan Oberlander, The Republican War on Obamacare—What Has It Achieved?, 379 New Eng. J. Med. 703, (Aug. 23, 2018); Alan Rappeport, What’s in a Name? Consumer Bureau to Find Out, N.Y. Times (Jun. 12, 2018),

[232] Laura Stevens & Amrith Ramkumar, Amazon Hits $1 Trillion Valuation, Wall St. J., (Sept. 4, 2018, 4:11 PM),

[233] Sara Salinas, Apple Hangs onto Its Historic $1 Trillion Market Cap, CNBC (Aug. 2, 2018, 11:48 AM),

[234] Press Release, CVS Health, CVS Health Announces Management Team for Combined Company Following Close of Aetna Acquisition Transaction (June 6, 2018),

[235] Reed Abelson, As Health Care Changes, Insurers, Hospitals and Drugstores Team Up, N.Y. Times, (Nov. 26, 2017),

[236] Compare Press Release, Aetna, Aetna and Humana Mutually End Merger Agreement (Feb. 14 2017),, and Delaware Judge Denies Anthem Injunction, Effectively Ending Cigna Merger, CNBC (May 12, 2017, 7:02 AM),, with Cecilia Kang & Edmund Lee, AT&T-Time Warner Deal Approval Gets Justice Department Challenge, N.Y. Times (July 12, 2018), (showing vertical mergers seem to trouble regulators less).

[237] Paul R. La Monica, What the CVS-Aetna Deal Means for Consumers, CNN (Dec. 5, 2017, 7:51 AM),

[238] Steven Findlay, Can A Community Hospital Stay True to Its Mission After Sale to Large Corporation?, Kaiser Health News, (July 23, 2018),; see also US Health Services Deals Insights: Q3 2018, PwC, (last visited Nov. 5, 2018); Melanie Evans & Anna W. Mathews, Hospital Giants in Talks to Merge to Create Nation’s Largest Operator, Wall St. J. (Dec. 10, 2017, 8:15 PM),; Keshia Hannam, Ascension and Providence St. Joseph in Talks to Form U.S.’s Largest Hospital Operator, Fortune (Dec. 11, 2017),

[239] Brent D. Fulton, Health Care Market Concentration Trends in the United States: Evidence and Policy Responses, 36 Health Aff. 1530 (2017).

[240] Casey Ross, Paying More and Getting Less: As Hospital Chains Grow, Local Services Shrink, STAT (Jan. 24, 2018),

[241] Rani Molla, Amazon’s Move into Wholesale Pharmaceuticals Sent Pharmacy Stocks Plunging, recode (Oct. 27, 2017, 12:00 PM),

[242] Coren, supra note 57.

[243] Lina M. Khan, Amazon’s Antitrust Paradox, 126 Yale L.J. 710, 716 (2017).

[244] Id. at 716-17.

[245] See David McLaughlin, Trump Antitrust Chief Says No Sign of Competitive Harm from Tech, Bloomberg (Sept. 28, 2018, 12:56 PM),

[246] See supra note 172 and accompanying text.

[247] Examples of Mobile Apps for Which the FDA Will Exercise Enforcement Discretion, U.S. Food & Drug Admin., (last updated Aug. 1, 2016).

[248] U.S. Food & Drug Admin., Classification Order on ECG App (Sept. 11, 2018), available at

[249] Conversation with Atul Gawande, supra note 103.

[250] C.f. Karl Polzer & Patricia A. Butler, Employee Health Plan Protections Under ERISA, 16 Health Aff. 93 (‎1997) (stating that ERISA was designed to establish uniform federal standards for employee benefit plans that are offered through private employers and unions).

[251] 45 C.F.R. § 164.504 (2018).

[252] 45 C.F.R. § 164.504(f)(2).

[253] See Amy D. Sorkin, Whose Distressed Baby Is It?, New Yorker (Feb. 11. 2014),

[254] See generally Terry, supra note 179, at 168-73.

[255] Id.

[256] See generally id. at 173-84.

[257] Natasha Lomas, Apple’s Tim Cook Makes Blistering Attack on the ‘Data Industrial Complex, TechCrunch (Oct. 24, 2018),

[258] See General Data Protection Regulation (GDPR), Regulation 2016/679, 2016 O.J. (L119) 1 (EU). See generally David Meyer, GDPR Attacks: First Google, Facebook, Now Activists Go After Apple, Amazon, LinkedIn, ZDNet (May 29, 2018, 1:44 PM),

[259] California Consumer Privacy Act of 2018, Cal. Civ. Code §§ 1798.100 to 1798.198 (West 2018).

[260] See S.B. No. 1125, 2017-2018 Leg., Reg. Sess. (Cal. 2018).

[261] S.B. No. 327, 2017-2018 Leg., Reg. Sess. (Cal. 2018).

[262] Examining Safeguards for Consumer Data Privacy Before the Senate Committee on Commerce, Science, and Transportation, 115th Cong. (2018) (announcement of the hearing available at

[263] See Dan Tynan, Silicon Valley Finally Pushes for Data Privacy Laws at Senate Hearing, Guardian (Sept. 26, 2018, 7:33 PM),; see also David McCabe, Tech Companies Want Privacy Rules, But on Their Own Terms, Axios (Sept. 26, 2018), See generally Nicolas P. Terry, Should Health Lawyers Pay Attention to the Administration’s Privacy Bill?, Health Aff.: Health Aff. Blog (Mar. 13, 2015),

[264] Kate Conger, Google Removes ‘Don’t Be Evil’ Clause from Its Code of Conduct, Gizmodo (May 18, 2018, 5:31 PM),

[265] Megan Thielking, Bill Gates Is Glad Amazon Is Getting into Health Care—But Cautions It’s Complicated, STAT (Apr. 30, 2018),

[266] Kevin Liptak, Trump: ‘Nobody Knew Health Care Could be So Complicated, CNN (Feb. 28, 2017, 4:10 AM),

[267] Surgical Intervention, supra note 18, at 59, 60.

[268] See, e.g., Anna W. Mathews et al., JP Morgan Plan Frets Some, Wall St. J., Feb. 4, 2018, at B2; Margot Sanger-Katz & Reed Abelson, Can Amazon and Friends Handle Health Care? There’s Reason for Doubt, N.Y. Times (Jan. 30, 2018),; Spencer Soper & Caroline Chen, Amazon Hasn’t Figured Out Drugstores Yet. But It Will Have To, Bloomberg (Dec. 18, 2017, 7:00 AM),

[269] See, e.g., David Blumenthal, Can Three of America’s Most Innovative Business Minds Really Transform Health Care?, Hill (Feb. 5, 2018, 7:30 AM),

[270] See generally PwC Health Res. Inst., The New Health Economy in the Age of Disruption: Novel Combinations Attempt to Remake the Health System 2-5 (Apr. 2018),

[271] David Blumenthal, Is M&A the Cure for a Failing Health Care System?, Harv. Bus. Rev. (Dec. 14, 2017),

[272] Abelson, supra note 235.

[273] John Bowden, Walmart in Early Talks to Buy Humana: Report, Hill (Mar. 29 2018, 7:50 PM),

[274] Reed Abelson, Hospital Giants Vie for Patients in Effort to Fend Off New Rivals, N.Y. Times (Dec. 18, 2017), (quoting Thomas Cassels, a consultant at the Advisory Board).

[275] Blumenthal, supra note 271.

Using the Economic Espionage Act to Protect Trade Secrets in Baseball

Using the Economic Espionage Act to Protect Trade Secrets in Baseball

By: Brette Trost*

Download a PDF version of this article here

In 2016, Christopher Correa, a former employee of the St. Louis Cardinals, was sentenced to forty-six months in prison for violating the Computer Fraud and Abuse Act when he accessed a Houston Astros database without authorization. However, these were not the only charges Correa could have faced. This note uses the Correa case to illustrate how the Economic Espionage Act can be used to prevent trade secret theft in Major League Baseball. More specifically, this note asserts that the sabermetric data systems used by MLB teams to evaluate and track players are legally protectable trade secrets. Furthermore, due to the fluid nature of the baseball analytics talent pool and barriers to civil prosecution inherent in baseball’s structure, the Economic Espionage Act presents the best way to combat the misappropriation of this information. The note goes on to distinguish between teams’ off-field and on-field tactics and discusses how, if at all, this framework should apply to the collection and use of biometric data.



Sports are the paradigm of competition. They are perhaps the arenas of business in which winning is most objectively quantifiable and competition is on display every night. On the field, competitive tactics are expected and gamesmanship is routine. Yet behind the scenes, there is an army of data scientists who are competing in their own way. Their competition does not revolve around which team collects the most runs after nine innings but rather around who can discover the most effective means of evaluating the players on the field.

This facet of the game is no secret. However, the extent to which some are willing to go to gain a competitive edge became strikingly apparent in 2016, when Christopher Correa, a member of the St. Louis Cardinals’ baseball operations staff, received a forty-six-month prison sentence for hacking into a Houston Astros’ database.[1] The database, known as “Ground Control,” was built by the Astros’ baseball operations department to house scouting reports, trade discussions, proprietary statistical analysis, injury histories, projections for players, contract information, and more.[2]

Major League Baseball (“MLB”) has undergone a major transformation over the last two decades. A game that once largely relied on subjective analyses and gut instincts to assess players, professional baseball—through the collection and study of statistical data—is now obsessed with an objective search for truth.[3] This objective analysis, or sabermetrics as it is commonly known, began as a hobby held by a few people scattered throughout the baseball world,[4] but it has since turned into an industry-wide practice, rapidly becoming the fixation of nearly every team in the league.[5] Teams now hire the most technical and scientific minds in the country, such as NASA engineers, data scientists from leading statistical software companies, and PhDs in cognitive neuroscience, applied statistics, and machine learning, in order to gain any slight competitive edge in discovering the most intricate details of a player’s ability.[6]

Sabermetrics, named after the Society of American Baseball Research (“SABR”), is defined as “advanced statistical collection and analysis to develop objective knowledge about baseball for use in player evaluation and tactical decision-making.”[7] Collecting certain statistics, such as batting average and earned run average, has been a part of the game since baseball’s inception.[8] However, for most of the twentieth century, the examination of more granular data was only performed by “amateur statisticians from outside the baseball establishment” and “statistically-inclined fans.”[9] By the end of the century, several companies, such as Baseball Prospectus and STATS LLC, began to collect more extensive data, including the speed and type of every pitch thrown during a game. Nonetheless, while baseball has been played in the United States since 1840, it was not until 2003, when Michael Lewis published the book Moneyball: The Art of Winning an Unfair Game,[10] that baseball industry insiders awoke to the potential of using analytical techniques to assess talent. Lewis’ book focused on one team, the Oakland Athletics, as it embarked on what was seen at the time as a unique and innovative process.[11] Now, every team relies at least to some extent on the use of analytics.[12]

Baseball teams own many of the same types of information as that which traditional businesses own, such as customer lists, pricing data, and marketing strategies. These categories of information are generally considered trade secrets when companies take reasonable measures to protect them.[13] Unlike traditional businesses, however, teams collect and store a plethora of data specific to the baseball industry, including statistical analyses (such as compilations and algorithms for new metrics),[14] scouting reports, trade proposals or discussion notes, playbooks, verbal or hand signals used on the field, player skill techniques, player training techniques, dietary and nutritional regimens, physical therapy techniques, psychological assessment techniques, and biometric analyses.[15] Many people in the baseball industry assert that such baseball-specific-data, which teams store and collect, constitute trade secrets.[16]

Despite the many potential trade secrets, there have not been any cases that discuss what material qualifies as a trade secret in baseball. Although Correa misappropriated information from Ground Control, a system that housed almost all of the Astros’ proprietary information, Correa was instead prosecuted under the Computer Fraud and Abuse Act (CFAA)[17] for hacking Ground Control.[18] What was criminalized was the fact that he accessed the information “without authorization,”[19] not the misappropriation of the information he obtained, and likely used, from the hacking. Due to the lack of court decisions (criminal or civil), there is no direct precedent holding that these types of analytics databases are in fact trade secrets. Nor is there extensive analysis of how teams keep this data secret and whether those controls are effective. Further, strategies the industry and public accept as part of the competitive nature of sports, such as on-field tactics to gain a competitive advantage like “stealing signs,” could be more intensely scrutinized if the legal system is used to police what should be considered fair competition in baseball.

Part I of this note will argue that the sabermetric data systems used by MLB teams to evaluate and track players are legally protectable trade secrets. Part II will examine the fluid nature of the baseball analytics talent pool, and will suggest that because of this aspect of the industry, the best way to prevent the misappropriation of these trade secrets is through criminal prosecution under the Economic Espionage Act of 1996 (EEA).[20] Part III will discuss on-field strategies, arguing that although the improper acquisition of on-field plays through tactics like sign-stealing may, in certain cases, technically meet the definition of theft of trade secrets under the EEA, this behavior does not warrant the imposition of criminal sanctions. Finally, Part IV will briefly analyze future questions on the proprietary nature of baseball data, noting that the focus will be less on sabermetric statistical systems and more on the collection, compilation, and ownership of biometric data.

I. Trade Secret Law and Its Application in Baseball[21]

Since teams deal with many different types of information, Lara and Nathaniel Grow surveyed the general counsels of teams across the four major North American professional sports leagues—baseball, basketball, football, and hockey—on what they believed to be trade secrets.[22] The survey, which received responses from nineteen teams, including two in MLB, revealed that 89.47% claimed that their scouting reports were trade secrets, 78.95% asserted trade secret protection over trade proposal or discussion notes, 73.68% asserted trade secret protection over statistical analyses, and 52.63% asserted trade secret protection over player skill development techniques and biometric analyses.[23] Variations among the general counsels’ responses is likely due to the different information-collection practices between the four major North American sports leagues—that is, differences in the amount and type of data collected in one sport compared to the other three sports and differences in how biometric data is relied upon in one sport compared to the other three sports.[24]

Given the general counsels’ apparent zeal for believing that their scouting reports, trade proposals and discussion notes, statistical analyses, player skill development techniques, and biometric analyses constitute trade secrets,[25] it is worthwhile to analyze whether such information actually satisfies the EEA’s requirements for trade secret protection. Using baseball as a case study, this note begins by exploring whether sabermetric data systems fall within the EEA.

Under the EEA, a trade secret is defined as “all forms and types of financial, business, scientific, technical, economic, or engineering information, including patterns, plans, compilations, program devices, formulas, designs, prototypes, methods, techniques, processes, procedures, programs, or codes, whether tangible or intangible,” provided that the “owner . . . has taken reasonable measures to keep such information secret,” and the information “derives independent economic value, actual or potential, from not being generally known to, and not being readily ascertainable . . . by, another person who can obtain economic value from the disclosure or use of the information.”[26] While people in the baseball industry have asserted that the data they collect and the systems they create are trade secrets,[27] there are almost no legal precedents that deal directly with this issue. Though Correa was not charged with violating any trade secret laws, his case provides insight into how baseball data could be subject to trade secret protection and potential criminal prosecution. This note argues that much of the content stored on sabermetric data systems, especially scouting reports and statistical analyses of player talent, can and should receive trade secret protection under the EEA.

A. An Overview of Trade Secret Law

Though laid out in its current form above, how the law, specifically the criminal law, defines a trade secret has changed over the last decade. To help clarify and strengthen trade secret protection, Congress amended the EEA through the enactment of the Theft of Trade Secrets Clarification Act of 2012[28] and the Defend Trade Secrets Act of 2016 (DTSA).[29]

In 1996, Congress passed the Economic Espionage Act to fill a hole in the statutory scheme. Lawmakers recognized the necessity of protecting the intangible assets of companies in the United States in response to the challenges prosecutors faced in fitting the misappropriation of these assets into statutes like mail and wire fraud,[30] the National Stolen Property Act,[31] and the CFAA, which were not designed for this type of prosecution.[32] President Bill Clinton acknowledged a growing need for a statute dedicated solely to the protection of these assets through the criminal law, noting that “[t]rade secrets are an integral part of virtually every sector of our economy and are essential to maintaining the health and competitiveness of critical industries operating in the United States.”[33]

The EEA provides a fine, a prison sentence of up to ten years, or both for individuals who steal or without authorization appropriate trade secrets as follows:

Whoever, with intent to convert a trade secret, that is related to a product or service used in or intended for use in interstate or foreign commerce, to the economic benefit of anyone other than the owner thereof, and intending or knowing that the offense will, injure any owner of that trade secret, knowingly steals, or without authorization appropriates . . . such information . . . shall . . . be fined under this title or imprisoned not more than 10 years, or both.[34]

Much of the jurisprudence that defines trade secrets relies on interpretations under the Uniform Trade Secrets Act (UTSA),[35] a model state law which as of January 2019 has been adopted in forty-seven states and the District of Columbia.[36] The UTSA and EEA provide largely identical definitions of a trade secret, especially following the enactment of the DTSA.[37] Judicial interpretations of trade secrets under the UTSA have provided a body of case law to guide the interpretation of the EEA.[38]

B. Definition of Trade Secrets Under the EEA

In order to be a trade secret under the EEA, the prosecutor or plaintiff must show three distinct elements: (i) the alleged trade secret falls within a listed type of information; (ii) the owner has taken “reasonable measures” to keep that information secret; and (iii) the information derives “independent economic value” from not being generally known or ascertainable through “proper means.”[39]

The threshold element, that the alleged trade secret falls within a listed type of information, is fairly simple to meet.[40] To fall within the EEA, the alleged trade secret must be “financial, business, scientific, technical, economic, or engineering information, including patterns, plans, compilations, program devices, formulas, designs, prototypes, methods, techniques, processes, procedures, programs, or codes, whether tangible or intangible.”[41]

In Nat’l Football Scouting, Inc. v. Rang,[42] the U.S. District Court for the Western District of Washington addressed the question of whether scouting reports fall within the listed types of information. Rang is the “only reported court decision considering the status of proprietary sports-related knowledge under trade secrecy law.”[43] In that case, National Football Scouting, Inc. (“National”) sued Robert Rang, a part-time sportswriter, and the website for which he wrote, Sports Xchange, for copyright infringement and misappropriation of trade secrets under the UTSA. National’s business involved providing scouting reports to NFL teams. The reports were compiled and produced by National’s own scouts. Twenty-one NFL teams had each paid $75,000 for access to the reports. The reports assigned each player an overall “Player Grade,” which was “a numerical expression representing National’s opinion of the player’s likelihood of success in the NFL.”[44] National sued Rang for writing articles which disclosed the Player Grades.

Rang argued that the Player Grades did not qualify as “information” within the meaning of the UTSA because they were subjective opinions, rather than “factual information.”[45] The court rejected this argument, saying “the fact that National has assigned a Player Grade to a certain player is not an idea or opinion.”[46] Instead, the Player Grades constituted “information” under the statute.[47] The court believed a factual dispute existed as to whether National reasonably kept the information secret and whether the grades had an independent economic value. This, the court held, was a question for the trier of fact. Ultimately, the parties settled.[48]

While the court held that the Player Grades were “information” under the UTSA, it did not take a stance on whether the reports would have constituted “information” had they merely comprised a scout’s thoughts on a given player, rather than assigning a Player Grade. It is common practice for scouts to provide a numerical grade when assessing baseball players.[49] However, would scouting reports which lack numerical player values also qualify as “information” under the EEA? The plain meaning of the term “information” and the function of scouting information in relation to the business of running a professional sports team suggest that scouting reports which lack numerical player values would likely still qualify as “information” under the EEA.[50]

The compilation of baseball statistics would also qualify as “information”’ under the EEA. For example, Inside Edge, a baseball analytics company,[51] reviews at-bats of every player to identify and compile specific indicia useful in determining what percentage of those at-bats lead to “well-hit” balls.[52] The EEA expressly includes “compilations,” as long as they meet the statute’s other prerequisites. Further, the “method” of compiling that data (i.e., through algorithms and code) and the “design” of that information, are also types of information listed in the EEA’s definition of a trade secret.[53] Finally, most of these analyses are performed with the aid of proprietary computer programs, which would undoubtedly qualify.

Under the EEA, the second element to qualify as a trade secret is that the owner must take “reasonable measures”[54] to keep the information secret. The DTSA addresses from whom the information must be kept secret to qualify as a trade secret under the EEA. Originally, the EEA stated that the information must be kept secret from “the public.”[55] The DTSA made the definition identical to the UTSA, such that the information must be kept secret from “another person who can obtain economic value” from the disclosure.[56] This narrowed the scope of the provision, as there might be information that is commonly known within an industry but not known to the public.[57]

What qualifies as a “reasonable measure”[58] to keep information secret? Determining reasonableness usually takes the form of cost-benefit analysis to find the optimal level of precaution that is not overly burdensome given the risk.[59] Although this would be fact-specific to each case, media reports reveal that teams use the same types of protections as other businesses in securing their materials, such as “walling off” information from those who do not need to know it, using computer security methods (i.e., passwords, firewalls, and surveillance), and having employees sign non-disclosure or non-compete agreements.[60] Contractual provisions can be especially important in this analysis, as a lack of a non-disclosure agreement “may alone defeat [a] trade secret claim.”[61]

Under the EEA, the third requirement for qualifying as a trade secret is that the information’s economic value derives from the fact that it is not “generally known to” or “readily ascertainable” by “another person who can obtain economic value” from the information.[62] Detailed scouting reports, statistical analysis, and other means of player evaluation help teams create a more competitive product on the field. If another team gains access to these methods of evaluation, it could recreate them at a lower cost. If a team knows what strategy its competitor is going to use, it could more precisely tailor its own strategy. If a competitor knows which players a team values via its scouting reports or the type of statistics the team measures, it could use that in trade negotiations or adopt those strategies if they prove successful and recognize talent before others. To a certain extent, the foregoing relies on the assumption that a more competitive team will lead to a more profitable franchise. Although this metric is slightly undercut by the fact that teams operate as part of a league, which has revenue sharing and as a whole may benefit from a more even playing field,[63] given the expenditures teams make on personnel to create analytics databases[64] and the fact that there are individual revenue streams that increase when a team is more competitive,[65] it seems fairly clear that there is economic benefit to having these secret programs.

It may be, at first, counterintuitive to think of scouting reports and sabermetric databases as trade secrets, especially given that all the action being observed and measured occurs in public and is largely preserved on video. However, the fact that the data, in the aggregate, comprises a compilation has important implications for evaluating its secrecy. Although each play is public information, the compilation transforms the constituent parts, which are public, into information that gives the team a competitive advantage and economic benefit, thereby becoming a trade secret.[66]

That is, the analysis that goes into the making of a statistic is what makes it a trade secret. While the Player Grades disseminated in Rang and the analysis provided by Inside Edge represent types of analytical compilations accessible to and bought by many teams, teams themselves create closely guarded compilations. For example, the Astros created an algorithm for determining when a player in the minor leagues is ready to be promoted to the major leagues. When the player meets the criteria in the algorithm, a green arrow appears next to that player’s name. A grey arrow next to the player signals that the player should be demoted, and a black arrow means the player should be cut.[67] This system is one example of the many ways in which teams create their own proprietary trade secrets. The team must decide what data to collect (i.e., speed, direction, distance, angle), how to collect it (human review, cameras, or software), and how to combine and present it (numbers, graphs, charts, graphics, computer programs, or symbols). Scouting reports, even if done through first-hand observation and annotation of results by scouts, contain some of the same compilation features as do statistics (i.e., what attributes of the player to write down and focus on, how to weigh each of those attributes, how to present the report, and how to measure the importance of each individual scouting report when assessing the overall performance of a player within a larger database). The creation of these evaluation systems all required time, money and effort, making them competitively valuable.[68]

C. The Interstate Commerce Requirement and Intent

Once the plaintiff has established that the information at issue is a trade secret, the EEA has two further threshold requirements for criminal prosecution. First, the trade secret must meet the statute’s interstate commerce requirement.[69] Second, the prosecution must establish a mens rea requirement—that the actions were taken “with intent.”[70]

The interstate commerce requirement of the EEA has been subject to some controversy. As the Act was originally written, the trade secret had to be “related to or included in a product that is produced for or placed in interstate or foreign commerce.”[71] The Theft of Trade Secret Clarification Act of 2012 revised this language to its current form, requiring the trade secret to be “related to a product or service used in or intended for use in interstate or foreign commerce.”[72] This amendment was passed in response to the Second Circuit’s holding in United States v. Aleynikov.[73] In Aleynikov, a Goldman Sachs employee stole source code for a high-frequency trading system, which was used to make large volumes of trades in securities and commodities. The court held that Aleynikov did not violate the EEA because the source code did not meet the interstate commerce requirement as it was not “produced for” or “placed in” commerce.[74] Much of the court’s reasoning in Aleynikov could have applied to the information at issue here (i.e., it was for internal use only and there was no intention to sell or license the product). However, Congress closed this loophole by expanding the statute to cover services (in addition to products) and by broadening the language to include products or services intended for use in interstate commerce.[75]

Here, the statistical databases and scouting reports relate to a “product” used in interstate commerce, namely the sport of baseball. Although baseball may not be a product in the tangible sense, it is surely a product in the same way that most forms of viewable entertainment are products. Professional athletes playing baseball is what the teams are marketing and selling to the public. Baseball is intended for public consumption through the attendance of live events and the viewing of television broadcasts. Given the congressional intent to broaden the EEA’s interstate commerce requirement, it is not a stretch to say that the systems are intended for use in baseball, which is a product used in interstate commerce. Further, though baseball has historically been subject to an antitrust exemption, which was rooted in a finding that the business of baseball was not a part of interstate commerce,[76] the United States Supreme Court later clarified in Flood v. Kuhn[77] that “[p]rofessional baseball is a business and it is engaged in interstate commerce.”[78]

Finally, the EEA distinguishes itself from its civil counterpart by including a high mens rea requirement for the remaining elements. The alleged thief must (i) intend to convert the trade secret to the economic benefit of someone other than the owner, (ii) intend or know that the theft will injure the owner of the trade secret, and (iii) knowingly misappropriate the trade secret through one of the delineated unauthorized acts.[79] Each element requires a fact-specific inquiry.

D. The EEA as Applied to Correa’s Case

As suggested above, Correa’s case provides an illustration as to how the EEA could apply to trade secrets in baseball. Correa was charged with violating five counts of the CFAA. The application of criminal law to the sports world is neither novel nor extreme, and there have been many other instances in which the government has taken a keen interest in criminal activity in the sports industry. For example, the federal government extensively investigated and prosecuted the use of performance enhancing drugs.[80] The New England Patriots’ involvement in the so-called “Spygate” incident garnered significant political interest, with many calling for criminal prosecution.[81] Currently, the Department of Justice is investigating MLB’s international signing practices.[82]

Correa worked for the Cardinals from 2009 until he was charged in 2015. During the beginning of his tenure with the Cardinals, Correa worked closely with Jeff Luhnow and Sig Mejdal. His relationship with Mejdal, in particular, was contentious—the two were considered “rivals” who engaged in “heated disputes.”[83]

In December of 2011, the Astros hired Luhnow as General Manager. In January of 2012, Luhnow brought Mejdal along to head the Astros’ analytics department.[84] Mejdal, a NASA engineer, was brought in to “make sense of all the new data that [was] becoming available for assessing ballplayers.”[85] When Mejdal left the Cardinals, he was directed to hand over his computer and password to Correa.[86] At the time, the Astros and Cardinals were division rivals.[87] While Luhnow and Mejdal were with the Cardinals, the analytics staff used a database tool called “Red Bird Dog,” and Luhnow and Mejdal “had clear ideas of what they wanted after using [that] system.”[88] At the Astros, the two went on to build Ground Control, which housed “a variety of confidential data, including scouting reports, statistics, and contract information, all to improve the team’s scouting, communication, and decision-making for every baseball-related decision.”[89] The system, which takes “variables and weights them according to the values determined by the team’s statisticians, physicist, doctors, scouts and coaches,” was referred to as the “repository of the organization’s collective baseball knowledge—the Astros’ brain.”[90]

When Mejdal left to join the Astros, he used a password similar to the one he had used while working at the Cardinals.[91] Correa guessed the new password and accessed Mejdal’s Ground Control and email accounts.[92] In March of 2013, Correa viewed scouting information, including the Astros’ scouts’ rankings of all players eligible for the 2013 Amateur Draft, a weekly digest page which listed statistics and notes on the performance and injuries of players whom the Astros were considering drafting, other web pages containing the Astros’ evaluations of the Cardinals’ prospects, and notes on trade discussions.[93] In June of 2013, the day before the 2013 Amateur Draft, Correa sorted the Astros’ draft page to see which prospects the Astros rated highest, as well as other scouting reports.[94] Before day three of the Draft, Correa viewed the draft page to look for players not yet drafted, including the page of Adam Nelubowich, whom the Cardinals drafted later that day, and three players the Cardinals had drafted the day before.[95] On July 31, 2013, the day of the non-waiver trade deadline, Correa again accessed Ground Control to view trade discussions between the Astros and other teams.[96]

On March 8, 2014, the Houston Chronicle published an in-depth article about Ground Control.[97] In response, the Astros enhanced their security precautions by changing Ground Control’s URL and requiring Ground Control users to change their passwords. The team reset the database to a system-wide default password, which was emailed to users. However, since Correa had access to Mejdal’s email, he also gained access to the new URL and default password. Correa used this information to access Luhnow’s account, viewing 118 web pages containing confidential information specifically relating to players the Astros were targeting in the 2014 Amateur Draft. Correa also viewed the “task page” for the Astros’ analytics department, which “listed the projects that the department was researching.”[98] In March of 2014, Correa allegedly leaked embarrassing confidential information about the Astros’ trade discussions to Deadspin, a sports blog. In so doing, Correa allegedly sought retaliation for a recent Sports Illustrated article, which praised Luhnow and Mejdal’s reportedly outstanding analytical methods and predicted that the Astros would win the 2017 World Series.[99] During these unauthorized intrusions, Correa used software to conceal his identity, his location, and the type of device he was using.

In December of 2014, Correa was promoted to Director of Scouting, where his duties involved scouting and the amateur draft—areas in which his access to Ground Control would have been particularly relevant. Though the government only charged Correa with accessing Ground Control on five occasions, the prosecution’s sentencing memo alleges that Correa in fact accessed Ground Control on forty-eight occasions, using the accounts of five different Astros employees.[100] The sentencing memo further states that Correa improperly accessed Mejdal’s email account over a two-and-a-half-year span.[101]

Correa claimed he was looking at Ground Control because he believed that the Cardinals’ proprietary data had been “improperly transferred to the Astros’ system by former Cardinals employees who had been hired by the Astros”[102] and asserted the Astros had replicated “key algorithms and decisions tools” created by the Cardinals.[103] No charges were ever brought against the Astros.

Correa waived indictment and pleaded guilty to five counts of “Unauthorized Access of a Protected Computer” under the CFAA for intrusively accessing the Astros’ database from March 2013 to June 2014.[104] Correa was sentenced to forty-six months in prison and ordered to pay criminal monetary penalties, including over $279,000 in restitution to the Astros.[105] In addition, the MLB Commissioner ordered the Cardinals to give the Astros their top two draft picks in the 2017 Draft and pay the Astros $2,000,000, the maximum punitive fine that the MLB Commissioner has the authority to direct pursuant to the MLB Constitution.[106]

While Correa pleaded guilty under the CFAA, could he have also been convicted under the EEA? Correa did not appropriate the operational code of Ground Control itself, nor did he appropriate Ground Control’s algorithms used to evaluate input data. Instead, he took the analytical conclusions generated by Ground Control—that is, the results produced by the system. It seems clear that such results would fit the EEA’s definition of a “trade secret.”

First, the content which Correa accessed falls within the types of information listed in section 1839(3). The rankings which the Astros assigned to players whom they were interested in drafting are analogous to those provided in Rang, and the scouting reports, trade discussions, and medical reports that Correa accessed would qualify as “business information” within the meaning of the EEA.

Second, the Astros took several “reasonable measures” to keep their information secret, as required by section 1839(3). Ground Control was not only protected by a password, but this password was reset after the Houston Chronicle article, showing that the organization was actively vigilant in protecting its system. Additionally, certain functions were only permitted to be used by certain employees. For example, Correa’s bouts of unauthorized access involved intruding into the accounts of two minor league players who, according to the government’s sentencing memorandum, had more limited access than other personnel.[107] Prior to Correa’s hacking, Luhnow said that the team was taking “as many measures as we can to protect the information,” such as walling off access, inhibiting the ability to download the data, and logging users’ activity on the system.[108]

Third, the information in Ground Control derived “independent economic value” from not being generally known or ascertainable through “proper means.” The government argued, and the court agreed, that “the deliverable from all of [the scouting] expenses was the information that they put in” Ground Control.[109] As the government noted, in order to diminish the strong likelihood that years and money will be fruitlessly invested in talented individuals who never end up graduating to major-league caliber, teams have poured increasingly massive amounts of resources into the consideration of which players to acquire.[110] The Astros’ proprietary data that was stored in Ground Control was only economically valuable because it was not generally known to other baseball teams. By developing its own tools and metrics, the Astros were able to better evaluate talent, thereby gaining a competitive edge over other teams. Analogizing the secrecy-based value of proprietary sabermetrics, one journalist contended that Correa’s actions were “tantamount to stealing the secret formula for Coke.”[111] The plea agreement asserts that the intended loss to the Astros was $1.7 million.[112]

Further, the data Correa accessed related to a product intended for use in interstate commerce. As discussed above, baseball, as a form of viewable entertainment in which tickets are sold and marketing is conducted throughout the country, is a product of interstate commerce.[113]

Compared to satisfying section 1839’s definition of a “trade secret” and section 1832’s interstate commerce requirement, the EEA’s mens rea element would likely be more difficult to prove. This may explain why the government refrained from pursuing charges under the EEA. Correa proffered that his intent was not to injure the Astros for his own benefit but to assess whether the Astros had stolen information from the Cardinals.[114] Had the government prosecuted Correa under the EEA and had his case proceeded to trial, Correa may have argued that he did not intend to injure the Astros or convert it for his or the Cardinals’ benefit.[115] There may have been no conclusive evidence that Correa intended to injure the Astros.

That said, such intent could be inferred from the fact that Correa allegedly leaked the Astros’ confidential trade discussions to Deadspin—a move which inflicted foreseeable reputational damage on the Astros and seemed to serve no purpose other than to injure and embarrass. Also, as the government pointed out in its sentencing memorandum, the information Correa looked at did not relate to the Cardinals, but rather included the Astros’ trade discussions with other teams. Such trade discussions had no bearing on whether the Astros stole information from the Cardinals, suggesting that Correa’s intent was to injure the Astros (and not to assess whether the Astros had stolen information from the Cardinals).

Moreover, Correa personally benefited from the hack insofar as he was promoted to Director of Scouting in 2014. The specific content Correa accessed in the Astros’ Ground Control database was directly related to drafting and scouting, which were areas core to Correa’s new job responsibilities. As the prosecution highlighted in its court filings, Correa’s access to Ground Control gave him the ability to “corroborate his judgment calls” by “check[ing] what another analytics-minded organization thought.”[116] In addition, Ground Control enabled Correa to know which projects the Astros found promising and which they discarded.

Two principle questions remain. First, why did the prosecution not bring charges against the Cardinals as well? The Commissioner clearly saw it fit to sanction the organization through a fine and loss of draft picks. Further, it would have been possible to introduce evidence that Correa acted within the scope of his employment, thus making the Cardinals liable pursuant to the doctrine of respondeat superior. Perhaps because the government knew that MLB had its own internal mechanisms for disciplining and fining clubs, there was less of a need for the government to impose its own sanctions.

Second, why did the government not prosecute Correa under the EEA? Certainly, the CFAA charge was the more straightforward claim to pursue since the EEA has a more intricate mens rea requirement. As previously mentioned, to succeed on an EEA charge, the prosecution would need to establish that the defendant (i) intended to convert the trade secret for the benefit of someone other than the owner; (ii) intentionally or knowingly injured the owner; and (iii) knowingly misappropriated the trade secret through one of the delineated unauthorized acts.[117] Further, the prosecution would have had to prove that the content which Correa accessed on the Astros’ Ground Control constituted a trade secret. It is possible that the Astros were reluctant to reveal information about Ground Control, especially given the media scrutiny. Indeed, the prosecution “agreed to a more restrained sentence,” including the decision not to add additional charges such as aggravated identity theft,[118] and noted that the plea agreement was “the product of extended negotiations between the parties, both of whom made concessions over several months.”[119] While the prosecution specifically noted that they chose not to charge Correa with aggravated identity theft, there is no discussion of the EEA. Declining to charge Correa under the EEA may have been part of the prosecution’s strategy of taking a lenient posture in order to reach a plea deal.

II. There Are Policy Reasons to Apply the Economic Espionage Act to Trade Secret Theft in Baseball

Ground Control is not the exception in the baseball industry. Many teams have similar databases that house information used to make player-related decisions, including the Cardinals (who have since moved on from Red Bird Dog but refuse to disclose the name of their new system),[120] the Boston Red Sox (Beacon),[121] and the Cleveland Indians (DiamondView).[122]

Correa was charged under the CFAA for accessing the Astros’ database “without authorization.” In so doing, the prosecution neglected the heart of the wrong Correa committed. The prosecution failed to address the true focus of Correa’s misdeeds—not the means of accessing the information (a problem which brings to mind questions of password sharing discussed in United States v. Nosal[123]), but the proprietary nature and use of the information itself. This point is underscored by the fact that Correa accessed Ground Control not via the use of technical skill but rather by receiving Mejdal’s password when Mejdal turned over his computer upon leaving the Cardinals. Because Mejdal failed to significantly change his password, Correa had an easy means of entry.

Correa’s case provides an important lesson concerning the nature of the intellectual property risks which baseball teams face. The main threat is not from “outside” hackers who illicitly access computer databases but from those already embedded within the industry who impermissibly use secret information. Accordingly, the EEA, which focusses on the impermissible use of secret information, addresses the risks faced by baseball teams more directly than does the CFAA, which focusses on the illicit access from outside hackers. The importance of this shift is driven home by a few considerations.

A. The Fluidity of Personnel in Baseball Creates a High Risk for Misappropriation

While employee turnover is a common feature of many industries, the fluidity of baseball operations staff is practically a definitional feature of the baseball industry.[124] Just as a player is traded from team to team, front office staff routinely move from team to team as well. The Milwaukee Brewers’ baseball operations department provides one example. The team’s current General Manager, David Stearns, joined the Brewers from the Houston Astros, and he had previously worked for the Cleveland Indians, New York Mets, and Pittsburgh Pirates. The team’s Assistant General Manager, Matt Arnold, had stints with the Tampa Bay Rays, Los Angeles Dodgers, Texas Rangers, and Cincinnati Reds. The team’s senior advisor, Doug Melvin, had prior experience with the Rangers, Baltimore Orioles, and New York Yankees. Taken together, Stearns, Arnold, and Melvin alone have inside experience with one third of the league, including two division rivals.[125]

An examination of the thirty General Managers at the start of the 2018 season reveals that nine have worked for four or more teams, and thirteen have worked for two or three teams.[126] While that leaves eight General Managers who have only worked for one franchise, every team has baseball operations department staff with experience working for multiple teams.[127]

This “incestuous shuffling of scouting and front office talent” poses a serious risk to teams that have developed proprietary data systems.[128] The information one team has in assessing players is directly applicable to the core business of a competitor team.

At first glance, increasing criminal enforcement of trade secret laws produces undesirable consequences, such as a restricting employee mobility. Limits on employee movement within an industry can have “detrimental effects on innovation, market competition, and economic growth,”[129] because preventing “talented individuals from standing upon the shoulder of giants, sharing knowledge, and making use of their human capital,” harms innovation.[130] Thus, perhaps using the CFAA would be less detrimental to employee mobility and the cross-pollination of ideas because the CFAA focuses on the access to that information rather than how it is used. However, as discussed above, Correa’s case illustrates why the CFAA is inadequate on other grounds. The statute’s vague notions of what constitutes “hacking” fails to address what society wishes to express as the true harm of Correa’s actions. We do not want to punish Correa solely because he guessed a password. Rather, we want to punish Correa because he used that password to give his team an illicit and illegal competitive advantage.

Further, concerns over the EEA restricting employee mobility in baseball are overstated. First, because the EEA includes such a high mens rea requirement, trade secret prosecutions would be brought sparingly in baseball. Under the EEA, the prosecution must establish as to each element of the crime that the defendant (i) intended to convert a trade secret to the economic benefit of someone other than the owner, (ii) intended or knew that such conversion would injure the owner of the trade secret, and (iii) knowingly misappropriated the trade secret through one of the delineated unauthorized acts.[131] Given the EEA’s demanding mens rea requirement, prosecutors will likely only go after those with a truly “evil-meaning mind.”[132] That is, employees moving between organizations without “evil-meaning minds” will not have to fear prosecution. Still, as with any criminal statute, prosecutorial discretion will ultimately reign supreme on when and whether these cases will be brought.

Second, baseball teams already have internal mechanisms in place to stifle employee fluidity and movement, meaning that any chilling effect on employee mobility from the EEA would be relatively unpronounced. Among other mechanisms, teams require employees to ask for permission before interviewing with another MLB team. These rules stem from the prohibition against tampering “with negotiations or dealings respecting employment” found in the Official Professional Baseball Rules Book.[133] The rule reads:

[T]here shall be no negotiations or dealings respecting employment, either present or prospective, between any player, coach or manager and any Major or Minor League Club . . . unless the Club or baseball employer with which the person is connected shall have, in writing, expressly authorized such negotiations or dealings prior to their commencement.[134]

On its face, the provision extends to “managers,” a term which, along with individual team policies, could be and has been broadly interpreted to encompass a host of employees.[135] Although individual teams’ employee policies are generally not public information, there have been some media reports of teams amending their employee policies in response to employees getting poached by other teams. For example, in 2011, the Toronto Blue Jays amended their employee policy so that employees in their baseball operations department would not be granted permission to interview with other teams for positions that did not represent a promotion from their current position.[136] Teams generally have a “widely observed policy of letting other clubs interview their employees for positions that would represent promotions.”[137] Even so, in some rare cases teams have exercised this power in restricting employees from interviewing with other teams even if the employee would be offered a promotion.[138] Given the general trend of vast movement of executives between teams, this system still seems largely perfunctory. Nonetheless, the system shows that the industry is trying to put its thumb on the scale against employee movement, thereby overshadowing any theoretical chilling effect the EEA may have on employee mobility.

Third, even assuming that the EEA would stymie employee mobility, this would not necessarily harm the baseball industry. Limiting employee fluidity may in fact be healthy for the industry. Sports are built on the notion of discovering who has the best competitive strategy and advantage. Sharing ideas between teams breaks down the fundamental competitive fabric and function of the system. Unlike industries which may provide for a more concrete connection to economic growth, public utility, or the public good, sports are a gratuitous demonstration of who can outcompete whom, who can come up with the best strategy, and who can win a game. Professional sports are built on the fundamental idea of secret gamesmanship. Unlike in other industries where employees working together across companies may enhance the public good, employees sharing secrets in sports undermines the gamesmanship of the sport, harms the public’s confidence in the integrity of the game, and reeks of collusion. Furthermore, the confined and unique nature of the sports industry lessens costs to innovation that may be more harmful in other settings.

The Correa case is one example of the effects of employee turnover, both from a psychological and competitive perspective. Correa’s psychological paranoia resulting from Luhnow and Mejdal’s departure allegedly led him to access Ground Control. It was not ludicrous of Correa to worry that Luhnow and Mejdal may have taken proprietary information with them¾as one commentator noted, “the secrets were inside their heads.”[139] Even a Houston Chronicle article that predated the Correa case alluded to this phenomenon, noting that “were a member of the Astros front office to leave, some of the team’s operating secrets would leave with them.”[140]

Moreover, the Correa case illustrates what a competitor can do once this type of data is acquired. Among numerous occasions, Correa accessed Ground Control on three key instances: right before and during the 2013 Amateur Draft and the day of the non-waiver trade deadline. By accessing Ground Control on these dates, Correa was able to see the players in which the Astros were interested as well as gain more information in assessing the Cardinals’ own picks. For example, Correa accessed scouting information for a pitcher, Marco Gonzales, who was the Cardinals’ first-round draft pick.[141]

It remains to be seen whether teams will take the Correa case as a cautionary tale. The Commissioner, Rob Manfred, insinuated that there must be a shift in the way teams think about guarding proprietary data, noting that “30 years ago intellectual property in this business was what some GM carried around in his head and he was going to take it with him when he left . . . . There wasn’t much you could do about that. Today the business has changed.”[142] Implicit in the Commissioner’s statement is an acknowledgment that some secrets cannot be kept simply due to the fluidity of the industry. His statement points to a worry of hackers accessing data, not leaks from a team’s own employees. However, the idea that the threat does not come from employees changing teams is misguided, as Correa was only able to gain access to the Astros’ database because Mejdal gave Correa his old password.

Luhnow himself condoned some type of misappropriation, saying “if someone leaves, they’re allowed to take . . . anything they remember in their head.” [143] The Director of Baseball Research for the Minnesota Twins echoed this sentiment, saying “if they can remember it you cannot stop them from taking it.”[144]

Accordingly, some argue that the EEA does criminalize “theft by memory.”[145] The idea of theft of trade secrets by memory is not wholly foreign. Under state law, several state courts have held that memorizing trade secrets constitutes a basis for civil liability.[146] For example, in Stampede Tool Warehouse, Inc. v. May,[147] former employees of an automotive equipment distributor argued their “taking” of the company’s customer list could not be a violation of the Illinois Trade Secrets Act[148] because they memorized the list instead of physically or digitally taking the information. The court disagreed, holding: “[a] trade secret can be misappropriated by physical copying or by memorization. . . . Using memorization to rebuild a trade secret does not transform that trade secret from confidential information into non-confidential information.”[149] Though state courts, under state trade secret laws, have imposed civil sanctions on those who misappropriate trade secrets via memorization, to date, criminal liability has been mostly limited to theft of information in a tangible medium.[150]

Nonetheless, the literal language of the EEA suggests that prosecuting theft by memorization could be even easier than prosecution under most state trade secret laws. First, the definition of trade secrets under the EEA is broader than that of the UTSA. The EEA says information can be a trade secret “whether or how stored, compiled, or memorialized,”[151] whereas the UTSA lacks such elaboration.[152] The fact that a trade secret need not be stored or memorialized under the EEA points to an expansive definition of intangible objects as trade secrets. Further, the EEA provides that one who “communicates[] or conveys such information” without authorization, has committed a prohibited activity.[153] This suggests there is no requirement that a person must physically or electronically take trade secrets to be prosecuted under the EEA. The UTSA contains no such language. Thus, the EEA seems to contemplate the risk of misappropriation via memorization more than state laws do. Further, despite the fact that the statute has undergone numerous amendments since its enactment, Congress has done nothing to clarify this potential ambiguity.

Still, although the language of the EEA is amenable to criminalizing the memorization and disclosure of trade secrets, the EEA—in practice—has not been used to prosecute such conduct (perhaps because criminal sanctions for this type of misappropriation would “unduly endanger legitimate and desirable economic behavior”[154]). Turning to the EEA’s legislative history, theft by memory was not the type of misappropriation Congress had in mind.[155] Although a section of the EEA was removed during reconciliation which said that “the general knowledge and experience that a person gains from working at a job is not covered,”[156] this language was removed because Congress found it “unnecessary and redundant.”[157] Remembering information from one’s previous job is often an incidental fact to employee movement, and society may not view this behavior as culpable enough to warrant criminal sanctions.

Coupled with the lack of prosecution under the EEA for trade secret theft by memorization, baseball industry executives have taken a seemingly permissive attitude towards such conduct, thereby creating uncertainty as to when society should deem this behavior wrongful. Limited information sharing is tolerated in baseball culture. For example, one unnamed R&D Director noted that scouts often trade advance reports in exchange for favors or simply as an act of kindness among industry friends.[158] Teams openly admit that the reason they hire analysts is often because of the projects said analysts have worked on with a competitor.[159] While baseball executives have deemed some information sharing impermissible, where they seem to draw the line (as to what trade secret misappropriation they consider wrongful versus what they consider permissible), they seem to do so arbitrarily with no grounding in any legal framework. For example, while one unnamed executive said that copying source code to a Dropbox would constitute prosecutable behavior, they opined that if a developer still had access to code in his or her email and used that for a new team, that would be a “gray area.”[160]

This permissive approach is misguided. Uncertainty as to conduct that companies deem improper has a detrimental effect on ex ante behavior and destroys any prospect for notice or ability to shape expectations as to what type of information teams value, what type of conduct is permitted, and what employees can take with them should they—or perhaps more accurately, when they—switch employers. The necessary normative guidance that shapes employee behavior is lacking in the baseball industry, so the threat of criminal prosecutions may be necessary to discourage misconduct that harms competition and fair play, on and off the field. Accordingly, the EEA can and should provide guidance to employees over what type of behavior should be considered wrongful.

B. How Disputes Are Resolved in the Absence of Criminal Sanctions

A wide variety of internal disputes in MLB are subject to arbitration clauses. If the controversy involves disciplining a player, the league is required to go to arbitration as prescribed in the collective bargaining agreement with the MLB Players Association.[161] For disputes involving two teams, the Major League Baseball Constitution (“MLB Constitution”) sets forth arbitration procedures. The latter is more applicable for cases of trade secret theft. The MLB Constitution states:

All disputes and controversies related in any way to professional baseball between Clubs . . . (including . . . owners, officers, directors, employees and players) . . . shall be submitted to the Commissioner, as arbitrator, who, after hearing, shall have the sole and exclusive right to decide such disputes and controversies and whose decision shall be final and unappealable.[162]

The Commissioner also has the separate power to impose punitive action on “Major League Clubs, owners, officers, employees or players” for any conduct “deemed by the Commissioner not to be in the best interests of Baseball.”[163]

Any action a team might seek to take against another team for the misappropriation of trade secrets by a former employee (i.e., under a state trade secrets law) would be subject to the arbitration clause of the MLB Constitution. Because teams are precluded from entering the courts to adjudicate these disputes, criminal law, where appropriate, could fill the gap. The fact that there is a separate system for internal discipline may lead some to believe that the need for criminal prosecution is reduced (or perhaps completely eliminated), as the league has come up with its own way for handling these types of disputes. However, the record of punishments imposed upon teams under the arbitration framework is sparse and opaque,[164] and the Commissioner is under no duty to disclose the punishments imposed.[165] Arbitration eliminates the advantage of the public process and transparency the legal system brings to the resolution of these disputes. Further, the standards in a criminal trial (i.e., beyond a reasonable doubt) in conjunction with the extensive mens rea requirements (especially for the EEA) allow for a more rigorous and thorough investigation of the issue than does private arbitration between teams.

Unlike civil disputes which fall within MLB’s mandatory arbitration rules, criminal prosecutions under the EEA would be adjudicated in the courts. In failing to prosecute EEA violations in the context of baseball, prosecutors have, in effect, empowered MLB to define the scope of trade secret law in baseball and to relegate such disputes to private arbitration. This is contrary to the will of the legislature, which has elected to make trade secret theft a crime. As discussed above, baseball teams—many which feel powerless to stop the sharing of proprietary information in the face of the industry’s employee fluidity—generally take a permissive attitude towards information leaving an organization when employees move teams. Where a private industry feels powerless to stop wrongful behavior is precisely where the criminal law should step in, not where the criminal law should stand down. Section 1832 of the EEA was written with this kind of misappropriation in mind. The importance of this information was underscored by Senator Herbert H. Kohl, when he said: “[B]usinesses spend huge amounts of money, time, and thought developing proprietary economic information . . . . This information is literally a business’s lifeblood. And stealing it is the equivalent of shooting a company in the head.”[166] Teams should not resign to letting their trade secrets, into which they have invested time and money, be taken to other teams. There may be more of a “league-think” attitude in baseball as opposed to other industries since each team is part of a larger collective, but undermining the competitive nature of the sport by allowing employees to bring proprietary information with them when they leave a team will eventually disincentivize teams from investing in these types of program and harm the league more than help it.

Finally, the reality is that prosecutors tend to use the EEA sparingly, often only in “egregious and ‘open-and-shut’ cases.”[167] The Correa case likely meets the elements set out by the EEA and would have been a good opportunity for the government to use the EEA in a high-profile case to both publicize the EEA and more concretely broaden trade secret protection in sports.

C. Conventional Methods of Protecting Trade Secrets Are Ineffective

Though teams use many conventional tactics which qualify as reasonable precautions to keep information secret under the EEA, such methods are inadequate to stop the misappropriation of proprietary information on their own. One tactic that teams take is walling off certain information from certain employees.[168] This approach has several pitfalls. First, it does nothing to address what occurs when the General Manager, who is not walled off from any information, moves teams (which, as discussed above, is common practice). Second, creating “information silos” is bad for cooperation and employee morale.[169] It also leads to fewer people making more decisions and increases the likelihood of error.[170] Third, the baseball industry is highly reliant on the use of interns. The sheer number of low level analysts who cycle through an organization makes walling off difficult. As one former Yankees baseball operations intern noted, the number of interns was often so high that there were “more interns than office space.”[171] Further, as fewer (or no) criminal prosecutions are brought, the onus will be on the team to come up with more effective ways to prevent the misappropriation of proprietary information. As a result, teams may wall off more data from certain employees, stifling an organization’s synergy and ability to perform to its full potential.

Alternatively, teams may turn to contract law. The two types of contractual provisions generally used to protect trade secrets—non-disclosure agreements and non-compete agreements—may be inadequate in the context of professional baseball. Non-competes are especially problematic since they receive vastly different treatment from state to state. This could put teams in states which generally prohibit non-competes, such as California (where five teams, or one sixth of the league, are located), at a significant disadvantage.[172] Further, a breach of these agreements would not be adjudicated in the courts. As mentioned above, disputes between teams (for example, an employee disclosing a trade secret in violation of a non-disclosure agreement) are subject to the MLB Constitution’s mandatory arbitration clause. Accordingly, inter-team disputes over non-disclosure agreements would not receive the protections and additional sanctions available through the legal system.

III. On-Field Tactics

One commentator called Correa’s actions a “high-tech version of what’s been going on forever in baseball—stealing signals.”[173] This comment illustrates the potential for complex legal questions to arise if the government more aggressively prosecutes the misappropriation of information in this context. In baseball, the ubiquity of sign-stealing has essentially been baked into the game.[174] In baseball, a sign is when a manager, coach, or player performs a series of physical movements (i.e., touching his hat, nose, or ear) to instruct the player to run a certain play (i.e., stealing a base or putting down a bunt).[175] Though in some situations, stealing signs could technically meet the standard under the EEA or other trade secret statutes, such on-field tactics should not be subject to adjudication in the courts.

A. Non-Verbal Signals Could Meet the Definition of a Trade Secret

The EEA definition specifically provides that the information does not have to be tangible. Though this was likely added to address digital forms of information, hand signals used during a game are a type of intangible business information. Although the signal is displayed in public, the meaning of the signal is not public information nor is the timing as to when the play will be deployed. The secrecy is key to the successful implementation. If a team knows what is coming, it can prepare to counteract that move. Some coaches create decoy signs in which they add a slight variation to the sign so the player knows that play should not actually be implemented. This can help assess the extent to which the signs have been compromised. The timing is also imperative. Even if a player can anticipate what type of pitch will be thrown, the timing of knowing exactly when that pitch will be thrown is where the value of the secret lies. Teams “closely guard . . . the various signals (hand, verbal, or otherwise) used by coaches to relay play calls to players during a game.”[176]

Does a team stealing an opposing team’s signs constitute misappropriation of a trade secret within the meaning of the EEA? If the player notices that a change-up is thrown every time the catcher puts down four fingers and communicates that to the batter while he is standing on second base, did he knowingly steal information? This scenario likely fails to meet the requirement of misappropriation. Rather, it is more akin to reverse engineering. Reverse engineering is when one “start[s] with [a] known product and work[s] backward to divine the process which aided in its development or manufacture.”[177] Here, the player used public information and decoded what the signal meant based on his powers of observation, thereby not acquiring the secret by improper means.[178] Although stealing signs in the manner described is technically “sign-stealing,” it is very common and is not something the criminal law or government should have a hand in.

However, a distinction must be made between signs that are stolen via the naked eye and signs stolen via the aid of other devices. There have been several cases of “sign-stealing” in which teams used more sophisticated means of acquiring the signs than merely observing signals and their outcomes. Teams have used various technologies and devices to decode or intercept signs, such as the use of video cameras or binoculars. In football, where teams communicate plays via electronic headsets, some teams have used electronic means of eavesdropping on these conversations. While non-video sign-stealing is an accepted part of the game, the use of other devices has been treated more seriously. In fact, while there is no official rule against sign-stealing in the MLB Rulebook, MLB issued a memo to clubs in 2001 specifically prohibiting the use of electronic equipment in connection with sign-stealing[179] (and the MLB Commissioner can punish teams for any conduct that is not in the “best interests” of baseball under the Rule Book[180]). This prohibition against the use of other devices in connection with sign-stealing was reiterated by Commissioner Manfred in 2017.[181] Another example comes from professional football, where the New England Patriots videotaped New York Jets coaches sending signals to their players during a game. It was not the stealing of the signs that got the Patriots in trouble but the fact that they did so using a camera.[182]

In 2017, the Yankees filed a claim with the Commissioner alleging that members of the Red Sox staff watching the game in the clubhouse used Apple Watches to communicate with training staff in the dugout about what signs the Yankees were using. Through a series of signals, the Yankees further alleged, the Red Sox training staff in the dugout then communicated this information to their players at the plate. The Red Sox filed a claim in response alleging that the Yankees used its camera from its regional sports network, YES, to steal signs during the game as well. Both teams were fined an “undisclosed amount” by the Commissioner.[183]

B. The Legal System Should Not Be Involved in Adjudicating Disputes over On-Field Misappropriation

On-field tactics like sign-stealing should not be subject to the criminal law, whether it is done with the naked eye or with the help of an electronic device. This is because there is a difference between illegal behavior and “gamesmanship.”[184] Unlike the stealing of sabermetric data or scouting reports, which have a corollary to the broader business world and are akin to the types of material Congress sought to protect when enacting the EEA, policing what is “against the rules” in a sporting event is no place for the judiciary. Sign-stealing is not only a common practice but has also been “lauded as good coaching.”[185] As one law professor argues, “nothing done on the field of play is cheating. What happens on the field, even if it violates the rules of the game, is still the game.”[186]

Questionable on-field tactics—even when done through sophisticated means like cameras or other equipment—are more appropriate for the disciplinary mechanisms built into the league’s arbitration forums. As it relates to on-field play, some level of “cheating” is accepted, and it should be up to those in charge of policing the sport, not judges, to delineate what is proper.[187]

Additionally, the sign must “derive[] independent economic value.”[188] While stealing signs can give teams a meaningful competitive edge[189] and some commentators believe “a sports play can be just as valuable to a sports team as a product, design, formula, or process may be to a manufacturing corporation or product developer,”[190] it would be more difficult to quantify how much a specific play is “worth” to the business. In contrast, the time, money and effort put in to creating analytical databases is easier to calculate and more congruent to what trade secret law was designed to protect.[191] Thus, a line should be drawn between “conduct primarily affect[ing] the integrity of the game” and conduct relating to the business of the enterprise and the information and programs a team creates, which “more closely align with business concerns.”[192] In his note, Andrew Barna puts forth several other policy reasons against adjudicating sign-stealing in the courts, including the fact that signs can be changed easily and at a minimal cost, the customary nature of sign-stealing within the game, and the Commissioner’s ability to impose penalties—such as loss of draft picks—which courts may not impose.[193]

IV. The Future of Sports Data

The data that Correa accessed included several players’ private medical records. Though keeping medical records is nothing new, teams have been pouring more resources into refining and leveraging this type of data. Every team now has in-house sabermetricians,[194] meaning the competitive advantage teams once gained from using sabermetrics has been reduced. As one consultant noted, “by the time someone has taken a statistical method elsewhere, has been able to implement it and is in a position to use that information to influence the decision-making of other teams, we would probably be onto the next thing.”[195] While sabermetric analysis has become the lifeblood of every team, injury avoidance mechanisms have become a greater priority.[196] To that end, teams have turned to biometric data to recapture the competitive edge that was once secured through the early adoption of statistical analysis.

If teams can better harness data to identify the factors that put players at risk for injury, they will have a significant advantage. As injuries derail careers (and cost teams millions of dollars), any informational edge in preventing them is coveted. One focus has been on the jarring increase in tears in the ulnar collateral ligament (“UCL”) of pitchers.[197] UCL tears take on average a period of twelve to sixteen months for recovery, but they can take as many as thirty months.[198] These injuries “keep a tremendous amount of money in the dugout.”[199]

The monitoring systems many teams are beginning to use are extensive and invasive. For example, the Seattle Mariners work with Fatigue Science to monitor player sleeping habits. Players wear wristbands, which were originally developed by the U.S. military to measure fatigue in pilots and soldiers.[200] Teams “speak only in vague terms about their efforts, fearful of publicizing any experiment that could become a competitive advantage,” which shows that teams are taking steps to keep these procedures secret and see some economic value in them.[201] Other examples include the use of harnesses to document “heart rate variability, respiration rate, activity and calories burned”[202] and arm sleeves embedded with 3D sensors to measure the force on the elbow joint of each throw.[203]

The collection and analysis of athletes’ biometric data raises ethical and privacy questions that are outside the scope of this paper.[204] For example, should employers be allowed to keep this kind of information private if it could lead to innovative breakthroughs in preventing injury in the future? If a team discovers a way to minimize or completely avoid the prevalence of a certain kind of injury, should there be a duty to disclose this information so players can protect themselves?[205] What are the ramifications if this information gets stolen? Should the precautions employers take to maintain the secrecy of this data differ from those taken for their normal statistical talent evaluations given the private nature of the data collected?

The collection, disclosure, and storage of biometric data would likely implicate other federal laws such as the Health Insurance Portability and Accountability Act of 1996 (HIPAA)[206] and the Genetic Information Non-Discrimination Act of 2008 (GINA).[207] Further, some states, such as Illinois, have enacted laws relating to employer collection of biometric data. The Illinois Biometric Information Privacy Act (BIPA) requires private entities to “store, transmit, and protect from disclosure all biometric identifiers and biometric information using the reasonable standard of care within the private entity’s industry.”[208] BIPA and similar laws could require teams to implement higher safeguards for the protection of player biometric data than merely protecting their databases with passwords, lest they be subject to liability for inadequately securing biometric data. As long as medical information is housed in the same place as other player data,[209] as was the case with Ground Control, teams should be motivated to strengthen the precautions they take for all their collectively-stored property. As the gathering of this data becomes more widespread and the benefits of its collection become clearer, the law will need to confront novel questions relating to protecting biometric data.


Although Correa was not charged under the EEA, he was ultimately sentenced to a significant amount of time in prison. Nonetheless, the changes over the last two decades in baseball—which have transformed the industry into one obsessed with the collection and analysis of data—show the need for greater legal protection of expensive and labor-intensive proprietary systems, such as Ground Control. Though teams take a somewhat relaxed attitude toward the realities of information sharing when employees switch teams, stronger trade secret protection in baseball is necessary to maintain the public’s confidence in the integrity of the game. The EEA provides one way for the government to stop the misappropriation of this kind of information as personnel move from team to team. The criminal law may not have a place on the baseball field, but it certainly has a place inside the office.


Baseball’s Fluid Talent Pool

* J.D. Candidate, New York University, 2019; B.A., English, University of Pennsylvania, 2013. The author would like to thank Professor Harry First for his expertise and guidance.

[1] Judgement in a Criminal Case at 1-3, United States v. Correa, No. 4:15-CR-00679 (S.D. Tex. July 21, 2016).

[2] Evan Drellich, Astros’ Formula for Success Builds on Its Own Data Bank, Hous. Chron. (Mar. 10, 2014, 9:00 AM),

[3] See generally Leigh Steinberg, Changing the Game: The Rise of Sports Analytics, Forbes (Aug. 18, 2015, 3:08 PM), (describing analytics as the “present and future of professional sports” and that any team not using them is at a “competitive disadvantage”).

[4] Lara Grow & Nathaniel Grow, Protecting Big Data in the Big Leagues: Trade Secrets in Professional Sports, 74 Wash. & Lee L. Rev. 1567, 1575 (2017) [hereinafter Grow & Grow] (“[P]ractically every team in MLB today utilizes sabermetric principles to at least some extent . . . .”).

[5] Id.

[6] Ben Baumer, In a Moneyball World, a Number of Teams Remain Slow to Buy into Sabermetrics, MLB article within The Great Analytics Rankings, ESPN (Feb. 23, 2015),

[7] R. Mark Halligan & Matthew J. Frankel, Nixon Peabody CLE Presentation: Secret Sabermetrics: Trade Secret Protection in the Baseball Analytics Field (Apr. 9, 2012),

[8] Grow & Grow, supra note 4, at 1571-72 (“As early as 1845 . . . newspapers began printing box scores recapping the statistical achievement of players in amateur baseball contests.”).

[9] Id. at 1574, 1575.

[10] Michael Lewis, Moneyball: The Art of Winning an Unfair Game (2003).

[11] Grow & Grow, supra note 4, at 1575.

[12] Id. (“[P]ractically every team in MLB today utilizes sabermetric principles to at least some extent . . . .”).

[13] See United States v. Nosal, 844 F.3d 1024, 1042-43 (9th Cir. 2016) (customer lists), In re Dana Corp., 574 F.3d 129, 152 (2d Cir. 2009) (pricing data), Optic Graphics, Inc. v. Agee, 591 A.2d 578, 586 (Md. Ct. Spec. App. 1991) (marketing strategies).

[14] Statistical analyses include, for example, the calculation of probabilities for defensive positioning, which has led to the proliferation of the infield shift. The infield shift typically involves moving infielders away from their standard positions to better account for a batter’s tendency to put the ball in play on one side of the field. For a brief discussion of the infield shift, see David Waldstein, Who’s on Third? In Baseball’s Shifting Defenses, Maybe Nobody, N.Y. Times (May 12, 2014),

[15] See Grow & Grow, supra note 4, at 1605 (surveying the general counsels of teams across the four professional sports as to what categories of information they deem be trade secrets).

[16] See id.; see also Rich Lederer, An Unfiltered Interview with Nate Silver, Baseball Analysts (Feb. 12, 2007), (referring to the detailed formulas in Nate Silver’s analytics system, PECOTA, as a trade secret); Jenny Vrentas, Mets Statistical Analyst Has Seen Growth and Evolution of Sabermetrics in MLB, Star-Ledger (Apr. 23, 2010), (quoting Ben Baumer saying teams are guarded about the statistical analyses they engage in because “it’s trade secrets”).

[17] Computer Fraud and Abuse Act, 18 U.S.C. § 1030 (2012).

[18] Information at 2, 5, United States v. Correa, No. 4:15-CR-00679 (S.D. Tex. July 21, 2016) (charging Correa with violating 18 U.S.C. §§ 1030(a)(2)(C), 1030(c)(2)(B)(iii)).

[19] 18 U.S.C. § 1030(a)(2)(C) (“Whoever intentionally accesses a computer without authorization . . . and thereby obtains . . . information from any protected computer . . . shall be punished as provided in subsection (c) of this section.”).

[20] Economic Espionage Act of 1996, Pub. L. No. 104-294, 110 Stat. 3488 (codified as amended at 18 U.S.C. §§ 1831-1839).

[21] This will examine only trade secret law in the United States. There is one baseball team in Canada, the Toronto Blue Jays, and thus Canadian law could be implicated. However, for the purposes of this paper, only provisions of U.S. law will be examined. For a brief summary of Canadian trade secret protection in this context, see Grow & Grow, supra note 4, at 1599-1601.

[22] Grow & Grow, supra note 4, at 1605.

[23] Id.

[24] Of the nineteen respondents, two responses came from MLB, seven from the NBA, four from the NFL, and six from the NHL. Each sport has different approaches to the use of data, specifically biometric data. Players in the NHL, NBA, and NFL have been more outspoken with privacy concerns relating to the collection of biometric data and have sought to restrict the use of biometric devices during games. See, e.g., Jeremy Venook, The Upcoming Privacy Battle over Wearables in the NBA, Atlantic (Apr. 10, 2017), When it comes to collecting analytical material in general, sports have relied on analytics at different paces. For example, the NFL has “lagged behind other professional leagues amid an otherwise widespread analytics revolution . . . .” Kevin Clark, NFL’s Brewing Information War, Ringer (June 22, 2016, 1:13 PM),

[25] Grow & Grow, supra note 4, at 1605.

[26] 18 U.S.C. § 1839(3) (2012 & Supp. IV 2017).

[27] See Lederer, supra note 16 (referring to the detailed formulas in Nate Silver’s analytics system, PECOTA, as a trade secret); Vrentas, supra note 16(quoting Ben Baumer saying teams are guarded about the statistical analyses they engage in because “it’s trade secrets”).

[28] Theft of Trade Secrets Clarification Act of 2012, Pub. L. No. 112-236, 126 Stat. 1627 (codified as amended at 18 U.S.C. § 1832(a) (2012)).

[29] Defend Trade Secrets Act of 2016, Pub. L. No. 114-153, 130 Stat. 376 (codified as amended in scattered sections of 18 U.S.C.).

[30] See 18 U.S.C. § 1341, 1343 (2012) (mail and wire fraud).

[31] 18 U.S.C. §§ 2314-2315.

[32] See R. Mark Halligan, Revisited 2015: Protection of U.S. Trade Secret Assets: Critical Amendments to the Economic Espionage Act of 1996, Marshall Rev. Intell. Prop. L. 476, 480 (2015) (“Before the EEA, federal prosecutors relied primarily upon the National Stolen Property Act and the wire and mail fraud statutes to commence criminal prosecutions for trade secret theft. Both statutes were ineffective.” (citation omitted)).

[33] Id. at 480-81.

[34] 18 U.S.C. § 1832(a).

[35] Unif. Trade Secrets Act § 1(4) (amended 1985), 14 U.L.A. 438 (1990).

[36] Trade Secrets Act, Uniform Law Commission, (last visited Jan. 2, 2019).

[37] Defend Trade Secrets Act of 2016, Pub. L. No. 114-153, 130 Stat. 376 (codified as amended in scattered sections of 18 U.S.C.).

[38] See, e.g., United States v. Chung, 659 F.3d 815, 825 (9th Cir. 2011) (“[W]e consider instructive interpretations of state laws that adopted the UTSA without substantial modification.”); see also United States v. Hanjuan Jin, 833 F. Supp. 2d 977, 1007 n.3 (N.D. Ill. 2012) (“Although there are some differences between the definitions of a trade secret found in the UTSA and the EEA, the Court also considers cases that have interpreted the requirements for a trade secret under state law based on the UTSA.”).

[39] 18 U.S.C. § 1839(3).

[40] See Rice Ferrelle, Combatting the Lure of Impropriety in Professional Sports Industries: The Desirability of Treating a Playbook as a Legally Enforceable Trade Secret, 11 J. Intell. Prop. L. 149, 164-65, 168-69 (2003), (listing some of the more obscure types of information that have been considered trade secrets under state law, including “a method of producing unique watercolor paintings,” “techniques for personal spiritual advance,” and a “technique for barbecuing meats”).

[41] 18 U.S.C. § 1839(3).

[42] Nat’l Football Scouting, Inc. v. Rang, 912 F. Supp. 2d 985 (W.D. Wash. 2012).

[43] Grow & Grow, supra note 4, at 1617.

[44] Rang, 912 F. Supp. 2d at 988.

[45] Id. at 995.

[46] Id. at 996.

[47] Id.

[48] Matthew J. Frankel, Hackers Strike Out: Recent Cases of Alleged Sports Analytics IP Theft, 1 J. Sports Analytics 83, 85 (2015).

[49] Alan Siegel, Baseball Scouts Use Numbers, Too, FiveThirtyEight (Aug. 11, 2014, 9:40AM),

[50] See N. Highland, Inc. v. Jefferson Mach. & Tool, Inc., 898 N.W.2d 741, 768. (Wis. 2017) (“Dictionary definitions of ‘information’ suggest that the term encompasses a broad class of knowledge.”).

[51] As of May 2018, twenty of the thirty MLB clubs used Inside Edge’s analytics services. See Jeff Arnold, Remarkable Brings Sports Data to Life, One Stat at a Time, (May 31, 2018),

[52] Alan Schwartz, Score That a Hit (But Was It Well Hit?), N.Y. Times (Oct. 22, 2006),

[53] 18 U.S.C. § 1839(3) (2012).

[54] Id.

[55] Economic Espionage Act of 1996, Pub. L. No. 104-294, § 101(a), 110 Stat. 3488 (codified as amended at 18 U.S.C. § 1839(3)(B)).

[56] Defend Trade Secrets Act of 2016 § 2(b)(1)(A), 18 U.S.C. § 1839(3)(B) (2012 & Supp. IV 2017).

[57] Adam Cohen, Feature: Securing Trade Secrets in the Information Age: Upgrading the Economic Espionage Act After United States v. Aleynikov, 30 Yale J. on Reg. 189, 204 (2013) (“Insiders in a business are considerably more likely to know about particular processes and methods than is the public.”).

[58] 18 U.S.C. § 1839(3)(A).

[59] Grow & Grow, supra note 4, at 1585.

[60] See id. at 1606 (survey finding that 94.74% of teams used computer security methods, 94.74% used non-disclosure agreements, and 78.95% used non-competes); see also Thomas Gorman, Prospectus Q&A: Mark Johnson, Baseball Prospectus (May 11, 2005), (referencing the Cardinals’ Mark Johnson’s non-disclosure agreement); Jon Greenberg, Q&A: New Cubs ‘Saberist’ Tom Tango, ESPN (Jan. 30, 2013), (noting the Chicago Cubs’ Tom Tango’s non-disclosure agreement); Jack Moore, How Wall Street Strangled the Life out of Sabermetrics, Vice Sports (Oct. 22, 2014, 5:30 AM), (discussing how Andrew Friedman’s consultants at the Tampa Bay Rays were “greeted by non-disclosure agreements).

[61] Halligan & Frankel, supra note 7.

[62] 18 U.S.C. § 1839(3)(B).

[63] J.C. Bradbury, Encouraging the Poor to Stay Poor, N.Y. Times (Aug. 28, 2010),

[64] For example, the Los Angeles Dodgers paid Andrew Friedman, their President of Baseball Operations, $35 million. Baumer, supra note 6. A team’s President of Baseball Operations makes all of the final decisions regarding baseball strategy and talent acquisition and helps to shape the analytics department through both hiring personnel and spearheading the development of new analytical tools and programs.

[65] See Ferrelle, supra note 40, at 166-67 (“[T]eam victories . . . in turn lead[] to increased advertising, television, and radio exposure. This exposure often translates into increased merchandise sales or lucrative media contracts. . . . As a team organization garners more victories, it reaps increased financial rewards.”); see also Samuel J. Horovitz, If You Ain’t Cheating You Ain’t Trying: “Spygate” and the Legal Implications of Trying Too Hard, 17 Tex. Intell. Prop. L.J. 305, 312 (2009) (“Profitability correlates to on-field success.”).

[66] See Restatement (Third) of Unfair Competition § 39 cmt. f (Am. Law Inst. 1995) (“[I]t is the secrecy of the claimed trade secret as a whole that is determinative. The fact that some or all of the components of a trade secret are well-known does not preclude protection for a secret combination, compilation, or integration of the individual elements.”); see also United States v. Nosal, 844 F.3d 1024, 1042 (9th Cir. 2016) (“The source lists in question are classic examples of a trade secret that derives from an amalgam of public and proprietary source data. To be sure, some of the data came from public sources . . . . But cumulatively, the Searcher database contained a massive confidential compilation of data . . . .”).

[67] Joshua Green, Extreme Moneyball: The Houston Astros Go All in on Data Analysis, Bloomberg Businessweek (Aug. 28, 2014, 3:00PM),

[68] Restatement (Third) of Unfair Competition § 39 cmt. f (Am. Law. Inst. 1995). (“[I[f acquisition of the information through an examination of a competitor’s product would be difficult, costly, or time-consuming, the trade secret owner retains protection . . . .”).

[69] 18 U.S.C. § 1832(a) (2012) (“Whoever, with intent to convert a trade secret, that is related to a product or service used in or intended for use in interstate or foreign commerce . . .”).

[70] Id.

[71] Economic Espionage Act of 1996, Pub. L. No. 104-294, § 101(a), 110 Stat. 3489.

[72] Theft of Trade Secrets Clarification Act of 2012, Pub. L. No. 112-236, 126 Stat. 1627 (codified as amended at 18 U.S.C. § 1832(a)).

[73] United States v. Aleynikov, 676 F.3d 71 (2d Cir. 2012); 158 Cᴏɴɢ. Rᴇᴄ. H6,848 (daily ed. Dec. 18, 2012) (statement of Rep. Smith).

[74] Aleynikov, 676 F.3d at 79-82.

[75] 158 Cᴏɴɢ. Rᴇᴄ. H6,848 (daily ed. Dec. 18, 2012) (statement of Rep. Smith) (“The Second Circuit’s Aleynikov decision revealed a dangerous loophole that demands our attention. In response, the Senate unanimously passed S. 3642 in November.”).

[76] Fed. Baseball Club of Balt. v. Nat’l League of Prof’l Base Ball Clubs, 259 U.S. 200, 208-09 (1922).

[77] Flood v. Kuhn, 407 U.S. 258 (1972).

[78] Id. at 282. Though it would likely not be difficult to prove that, despite the antitrust language, the business of baseball is connected to interstate commerce, the fact that this question may be less straightforward and that case law like Aleynikov illustrates that this requirement is not something courts are willing to simply look past, prosecutors may be more reluctant to bring charges under the EEA in the context of baseball.

[79] 18 U.S. Code § 1832(a) (“Whoever, with intent to convert a trade secret . . . to the economic benefit of anyone other than the owner thereof, and intending or knowing that the offense will, injure any owner of that trade secret, knowingly” misappropriates a trade secret through an enumerated act shall be subject to punishment).

[80] See, e.g., Congress Asks DOJ to Prove Whether Clemens Lied Under Oath, Associated Press (Feb. 27, 2008),; Del Quentin Wilber & Ann E. Marimow, Roger Clemens Acquitted of All Charges, Wash. Post (June 18, 2012),

[81] Horovitz, supra note 65, at 324 n.101 (“Given the level of Congressional attention Spygate and other sports stories have received recently, the notion of federal prosecution hardly seems farfetched.”).

[82] Jon Werthem, Exclusive: The Evidence that Persuaded U.S. Department of Justice to Investigate MLB Recruitment of Foreign Players, Sports Illustrated (Oct. 2, 2018), https:/

[83] See Sentencing Memo of the United States at 4, United States v. Correa, No. 4:15-CR-00679 (S.D. Tex. July 21, 2016).

[84] Brian McTaggert, Astros Hire Luhnow as General Manager, MLB (Dec. 8, 2011, 12:10 AM),; Brian McTaggert, Analyze This: Astros’ Mejdal Takes on Unique Role, MLB (Jan. 31, 2012, 11:37 AM),

[85] Green, supra note 67.

[86] Sentencing Memo, supra note 83, at 2.

[87] The Astros and Cardinals were both members of the National League Central division before the Astros moved to the American League in 2013.

[88] Drellich, supra note 2.

[89] Plea Agreement at 7, United States v. Correa, No. 4:15-CR-00679 (S.D. Tex. July 21, 2016).

[90] Green, supra note 67.

[91] Plea Agreement, supra note 89, at 8.

[92] Id.

[93] Id.

[94] Sentencing Memo, supra note 83, at 3.

[95] Id. at 4.

[96] Plea Agreement, supra note 89, at 9.

[97] Drellich, supra note 2.

[98] Plea Agreement, supra note 89, at 10.

[99] See Sentencing Memo, supra note 83, at 6 (describing the Deadspin leak).

[100] Id. at 1.

[101] Id.

[102] Ex Parte Motion for Issuance of Subpoena & Prehearing Production of Materials at 1, United States v. Correa, No. 4:15-CR-00679 (S.D. Tex. July 21, 2016) (recounting Correa’s statement made at rearraignment); see also Rearraignment at 9:8-24, United States v. Correa, No. 4:15-CR-00679 (S.D. Tex. July 21, 2016).

[103] Derrick Goold, Correa Gives His Account of Hacking Case, St. Louis Post-Dispatch, Feb. 1, 2017, at B1.

[104] Plea Agreement, supra note 89, at 1.

[105] Judgment in a Criminal Case, supra note 1, at 3, 6.

[106] Tom Verducci, Lax Hack Smack: MLB, Rob Manfred Let Cardinals off Easy in Hacking Scandal, Sports Illustrated (Jan. 30, 2017), See generally Major League Const. art. II, § 3, available at (“In the case of conduct by Major League Clubs, owners, officers, employees or players that is deemed by the Commissioner not to be in the best interests of Baseball, punitive action by the Commissioner for each offense may include . . . a fine, not to exceed $2,000,000 . . . .”).

[107] Sentencing Memo, supra note 83, at 5.

[108] Joshua Green, My Time with the Architect of the Astros’ ‘Ground Control, Bloomberg Businessweek (June 16, 2015, 3:47 PM),

[109] Rearraignment, supra note 102, at 11:8-9.

[110] Responses to Defendant’s PSR Objections at 6, United States v. Correa, No. 4:15-CR-00679 (S.D. Tex. July 21, 2016).

[111] Green, supra note 108.

[112] Plea Agreement, supra note 89, at 4. The prosecution reached the $1.7 million figure by taking the number of players Correa viewed “by 200,” dividing that by the number of players that were eligible to be drafted and multiplying by the scouting budget of the Astros that year. See Rearraignment, supra note 102, at 10:22-11:4. The actual monetary loss incurred by Correa’s victims was established as just over $279,000, and this substantially smaller figure was pertinent to the determination of Correa’s sentence pursuant to the U.S. Sentencing Guidelines.

[113] See Flood v. Kuhn, 407 U.S. 258, 282 (1972) (“Professional baseball is a business and it is engaged in interstate commerce.”).

[114] Goold, supra note 103.

[115] In his guilty plea, Correa conceded that he acted with intent to injure the Astros. See Plea Agreement, supra note 89, at 10 (“The Parties agree that the defendant’s intended loss under the U.S. Sentencing Guidelines definition for all of his intrusions is $1.7 million.”). Conceding that he acted with intent may have been a condition of his guilty plea. However, it does not bear on how Correa would have argued had his case proceeded to trial.

[116] Responses to Defendant’s PSR Objections, supra note 110, at 6.

[117] 18 U.S.C. § 1832(a) (2012).

[118] Responses to Defendant’s PSR Objections, supra note 110, at 7 (“[T]he parties agreed that a more restrained sentence was appropriate, so they agreed on the loss calculations and the sophisticated means enhancement, and to not charge aggravated identity theft.”).

[119] Id.

[120] See Derrick Goold, MLB Commissioner: Teams Need to Protect Intellectual Property, St. Louis Post-Dispatch (Nov. 10, 2015), (“The Cardinals have long since abandoned ‘Red Bird Dog’ for an internal database whose nickname they don’t even want to share.”).

[121] Alex Speier, Red Sox to Retire ‘Carmine, Bos. Globe, Feb. 23, 2017, at D.1.

[122] Alex Kaufman, Moneyball, Before Moneyball Was Cool, SweetSpot (June 7, 2014),

[123] United States v. Nosal, 844 F.3d 1024 (9th Cir. 2016). In Nosal, an employee gave former employees her password so they could continue to access the company’s confidential database. Nosal was convicted under the CFAA, and as Judge Reinhardt noted in his dissent, the application of the CFAA to this scenario had the potential to criminalize broader types of password sharing. Nosal, 844 F.3d at 1048 (Reinhardt, J., dissenting).

[124] See Dean Pelletier, Trade Secrets: Extra Edges on the Diamond, Pelletier L. (Mar. 8, 2015), (calling employee mobility “part of the fabric of all professional sports”).

[125] 2018 Milwaukee Brewers Media Guide 10-12 (Mike Vassallo et al. eds.).

[126] See infra Appendix.

[127] Data was compiled using each team’s 2018 Media Guide. Employees holding the title of “General Manager” were included in this study. The Boston Red Sox’s Dave Dombrowski, the Miami Marlins’ Michael Hill, and the Baltimore Orioles’ Dan Duquette were included in this study, as those three teams do not employ anyone with the title “General Manager.”

[128] Ben Lindbergh, Baseball’s Ever-Expiring Secrets, Ringer (Feb. 6, 2017, 11:49 AM),

[129] Orly Lobel, The New Cognitive Property: Human Capital Law and the Reach of Intellectual Property, 93 Tex. L. Rev. 789, 835 (2015).

[130] Id.; see also Cohen, supra note 57, at 229 (“Diminished labor mobility is costly not only for individual workers, but for the nation as a whole. The economy is at its most efficient when workers are able to take their labor where the market would value it most highly.” (internal citations omitted)).

[131] 18 U.S.C. Code § 1832(a) (2012) (“Whoever, with intent to convert a trade secret . . . to the economic benefit of anyone other than the owner thereof, and intending or knowing that the offense will, injure any owner of that trade secret, knowingly” misappropriates a trade secret through an enumerated act shall be subject to punishment).

[132] Morissette v. United States, 342 U.S. 246, 251 (1952).

[133] Office of the Comm’r of Baseball, The Official Professional Baseball Rules Book, R. 3(k) (2018) [hereinafter MLB Rules Book],

[134] Id.

[135] For an example of an investigation into tampering regarding a team’s manager, see Associated Press, MLB Rules No Tampering Found in Cubs’ Hiring of Joe Maddon, ESPN (Apr. 29, 2015),

[136] Doug Harrison, Jays Amend Employee Policy to Quell Farrell Rumours, CBC Sports (Oct. 25, 2011, 12:39 PM),

[137] Lindbergh, supra note 128.

[138] For example, the Chicago White Sox denied then Assistant General Manager Rick Hahn permission to interview for General Manager of the Seattle Mariners.

[139] Lindbergh, supra note 128.

[140] Drellich, supra note 2.

[141] Responses to Defendant’s PSR Objections, supra note 110, at 4.

[142] Bill Shaikin, Former Cardinals Executive Pleads Guilty, Admits Hacking Astros’ Computers, L.A. Times (Jan. 8, 2016, 6:54 PM),

[143] Drellich, supra note 2.

[144] Lindbergh, supra note 128.

[145] Geraldine Szott Moohr, The Problematic Role of Criminal Law in Regulating Use of Information: The Case of the Economic Espionage Act, 80 N.C. L. Rev. 853, 878 (2002) (“The EEA may be read to protect trade secrets that exist only in the mind of the holders against misappropriation through memorization of another.”).

[146] See, e.g., Allen v. Johar, Inc., 823 S.W.2d 824 (Ark. 1992); see also Ed Nowogroski Ins. v. Rucker, 971 P.2d 936 (Wash. 1999).

[147] Stampede Tool Warehouse, Inc. v. May, 651 N.E.2d 209 (Ill. App. Ct. 1995).

[148] 765 Ill. Comp. Stat. Ann. 1065/1-9 (West 2017).

[149] Stampede Tool Warehouse, Inc., 651 N.E.2d at 217.

[150] Cohen, supra note 57, at 227 (“[M]ost [states] appear to limit criminal liability to cases in which there has been some kind of physical taking and do not require employees to wipe clean the slates of their memories.” (internal quotation marks omitted)).

[151] 18 U.S.C. § 1839(3) (2012) (emphasis added).

[152] Unif. Trade Secrets Act § 1(4) (amended 1985), 14 U.L.A. 438 (1990) (“‘Trade secret’ means information, including a formula, pattern, compilation, program, device, method, technique, or process . . . .”).

[153] 18 U.S.C. 1832(a)(2).

[154] 142 Cong. Rec. S12,213 (daily ed. Oct. 2, 1996) (Managers’ Statement for H.R. 3723, The Economic Espionage Bill).

[155] Id. at S12,212 (daily ed. Oct. 2, 1996) (statement of Sen. Kohl) (“[W]e do not want this law used to stifle the free flow of information or of people from job to job.”).

[156] Id.

[157] Id. at S12,213 (Managers’ Statement for H.R. 3723, The Economic Espionage Bill).

[158] Lindbergh, supra note 128.

[159] Id. (“Most of the time in offices that are more inclusive by nature you will be exposed to the development and actual usage of the systems you develop and as such when you leave you take that with you. In fact, in most cases that is part of the reason you are being hired to begin with.” (quoting the former General Manager of the Colorado Rockies, Dan O’Dowd)).

[160] Id.

[161] 2017–2021 Collective Bargaining Agreement Between Thirty Major League Clubs and the Major League Baseball Players Association art. XIII (Dec. 21, 2016),

[162] Major League Const. art. VI, § 1, available at

[163] Id. art. II, § 3.

[164] Michael McCann, Breaking Down Chris Correa’s Prison Sentence For Hacking Astros, Sports Illustrated (July 18, 2016), (“The record of team punishments is fairly barren.”).

[165] See, e.g., Matt Snyder, MLB Rules on Red Sox-Yankees Sign Stealing and Fines Both Teams, CBS Sports (Sep. 15, 2017), (discussing fines of an “undisclosed amount” levied on the Red Sox and Yankees in a recent dispute over sign-stealing).

[166] 142 Cong. Rec. S740 (daily ed. Feb. 1, 1996) (statement of Sen. Kohl).

[167] Halligan, supra note 32, at 499.

[168] Drellich, supra note 2 (“There are ways to protect yourself by making sure that people have access to the data that they only need to make the decisions in the area.” (quoting Luhnow)).

[169] Lindbergh, supra note 128 (“It creates real morale issues in the staff if they are walled off from things, particularly once you get into director and higher levels. Everyone doesn’t need to know every piece of information, but if you start excluding department heads from certain things in the fear that they might leave, you are sort of inviting them to leave for somewhere else where they will be more involved and more trusted.” (quoting an unnamed executive)).

[170] Id. (“A walled-off employee can’t make as many direct contributions, and the smaller the pool of potential peer reviewers, the more likely it is that mistakes will survive.”).

[171] Id.

[172] Grow & Grow, supra note 4, at 1618.

[173] Tyler Kepner, Former Cardinals Executive Pleads Guilty to Hacking Astros, N.Y. Times (Jan. 8, 2016),

[174] See, e.g., Tim Kurkjian, Sign-Language Hidden Cameras, Phony Signals, Double-Dealing Espionage. No This Isn’t the CIA—We’re Talking About the Game Within the Game of Baseball, Sports Illustrated (July 28, 1997), (quoting former Minnesota Twins Manager Tom Kelly saying that “stealing signs is part of the job”); Scott Lauber, Dustin Pedroia Downplays Scandal: ‘Don’t Think This Should Be News, ESPN (Sept. 6, 2017), (quoting the Boston Red Sox’s Dustin Pedroia calling sign-stealing “part of the game”).

[175] For a more thorough explanation of the history and different variations of signs in baseball, as well as how each element of the UTSA and EEA may be applied to sign-stealing, see Andrew G. Barna, Note, Stealing Signs: Could Baseball’s Common Practice Lead to Liability for Corporate Espionage?, 8 Berkeley J. Ent. & Sports L. (forthcoming 2019).

[176] See Grow & Grow, supra note 4, at 1579; see also, Barna, supra note 175, at 19-20 (“Per industry practice, MLB teams take many measures to protect their signs. They use false signs, change signs throughout the game, change signs after players get traded, ensure the pitcher is not ‘tipping’ his signs, and speed up the pitcher’s delivery.”).

[177] Kewanee Oil Co. v. Bicron Corp., 416 U.S. 470, 476 (1974).

[178] Delineating the line between misappropriation and superior knowledge or an educated guess is a common difficulty in the criminal law, especially in the insider trading context. See, e.g., SEC v. Steffes, 805 F. Supp. 2d 601 (N.D. Ill. 2011).

[179] Ken Rosenthal, Red Sox Crossed a Line and Baseball’s Response Must Be Firm, Athletic (Sept. 5, 2017),

[180] MLB Rules Book, supra note 133, at R. 21(f).

[181] Robert D. Manfred, Jr., Commissioner’s Statement Regarding Red Sox-Yankees Violations, MLB (Sept. 15, 2017),

[182] See generally Horovitz, supra note 65.

[183] Scott Lauber, Red Sox, Yankees Fined Separate as Part of MLB Investigation Into Sign-Stealing, ESPN (Sept. 15, 2017), (discussing the Commissioner’s determination that there was “insufficient evidence” to back the Red Sox claim against the Yankees, but nonetheless fined the Yankees after uncovering evidence that the Yankees had engaged in improper conduct in connection with the use of a dugout phone in a previous season).

[184] Horovitz, supra note 65, at 327 (“The blurring of the cheating-gamesmanship line is of paramount legal importance—the former is intuitively misappropriation, the latter proper.”).

[185] Id. at 318 (internal quotation marks omitted).

[186] Id.

[187] Id. at 328-29 (“It would be difficult for courts to accurately determine what is proper or improper in a world governed by unwritten laws that are hardly unanimous.”).

[188] 18 U.S.C. § 1839(3)(B) (2012).

[189] Barna, supra note 175, at 5.

[190] Ferrelle, supra note 40, at 167.

[191] Horovitz, supra note 65, at 329 (“[T]he core focus of trade secret law is still the business world.”).

[192] Id. at 330. (“[T]he more conduct is directly related to business (that is, the more it is removed from pure athletic competition), it not only more closely aligns itself with the core justifications for trade secret protection but it also becomes easier and more natural for courts to classify as proper or improper.”).

[193] Barna, supra note 175, at 22.

[194] Ben Lindbergh & Rob Arthur, Statheads Are the Best Free Agent Bargains in Baseball, FiveThirtyEight (Apr. 26, 2016, 11:04 AM),

[195] Lindbergh, supra note 128 (quoting director of analytics Jesse Smith).

[196] Associated Press, Putting Data Science on the Pitcher’s Sleeve, N.Y. Times (Apr. 2, 2016), (quoting Glenn Fleisig calling biometric data collection “the next sabermetrics”).

[197] Jonah Keri, The Tommy John Epidemic: What’s Behind the Rapid Increase of Pitchers Undergoing Elbow Surgery, Grantland (March 10, 2015), (twenty-five percent of major league pitchers and fifteen percent of minor league pitchers in 2015 had Tommy John Surgery to repair the ulnar collateral ligament, and more pitchers had the surgery in 2014 than all of the 1990s).

[198] Tommy John FAQ, MLB: Pitch Smart, (last visited Dec. 20, 2017).

[199] Joe Greenberg, Q&A: New Cubs ‘Saberist’ Tom Tango, ESPN (Jan. 30, 2013),

[200] Brian Costa, Baseball’s Fight with Fatigue, Wall St. J. (Feb. 23, 2015, 12:45 PM),

[201] Id.

[202]The Sports Industry’s New Power Play: Athlete Biometric Data Domination, SportTechie (March 3, 2017),

[203] Grow & Grow, supra note 4, at 1578.

[204] For a discussion of the ethical and privacy issues surrounding the collection of athletes’ biometric data, see id. at 1619-20.

[205] See id. at 1620.

[206] Health Insurance Portability and Accountability Act of 1996, Pub. L. No. 104-191, 110 Stat. 1936 (codified in scattered sections of 26, 29, and 42 U.S.C.).

[207] Genetic Information Nondiscrimination Act of 2008, Pub. L. No. 110- 233, 122 Stat. 881 (codified in scattered sections of 26, 29, and 42 U.S.C.).

[208] 740 Ill. Comp. Stat. Ann. 14/15 (West 2017).

[209] For example, Correa viewed medical pages that were housed in Ground Control for 1B/DH/LF Conrad Gregor and 1B Chase McDonald. See Responses to Defendant’s PSR Objections, supra note 110, at 1.

[210] Thank you to Mike Passanisi for helping design this image.

JIPEL Vol. 7, No. 2 – Spring 2017

The NYU Journal of Intellectual Property and Entertainment Law is proud to present Volume 7 Issue 2 of the Journal. While PDFs of the individual articles may be found accompanying their respective posts, you may view and download a PDF of the complete issue here

A Judicial ‘Supplement’ to Advertising Law: The Fourth Circuit’s GNC Decision and Policy Implications for the Dietary Supplement Industry

A Judicial ‘Supplement’ to Advertising Law: The Fourth  Circuit’s GNC Decision and Policy Implications for the Dietary Supplement Industry
By Gia Wakil*

Download a PDF version of this article here



Every home has a story, as the old adage goes, and the story in this Note is set in the modern-day vitamin cabinet. Few items are as ubiquitous in American homes as the pill bottle. From vitamins like A and C, or minerals like iron and zinc, the supplements found in a single household typically run the full alphabet. The majority of adults in the United States take one or more dietary supplements occasionally, or even every day,[1] with sales to American consumers exceeding $35 billion per year.[2] Americans rely on supplements to ameliorate nutritional deficiencies or to maintain their health. And yet, despite the prevalence of supplements, concerns abound about their efficacy and the veracity of claims supporting the use of supplements. Oftentimes, the prospect of shorter colds, stronger nails, or improved general health justifies the price of the “gamble” in the minds of many consumers.

It was in similar, health-conscious hopes that consumers in the In re GNC Corp. [3] case had reached for bottles of glucosamine-chondroitin, two common ingredients in joint health specialty products.[4] The labels touted that the mixtures “promote[] joint mobility and flexibility” and provide “[c]linical strength for daily long-term use.” GNC allegedly had, and on some labels cited to, a reasonable scientific basis for these claims, though its legitimacy was tarnished in the face of mounting scientific evidence indicating otherwise.

After years of medical controversy examining whether glucosamine-chondroitin is effective, lawsuits against a variety of its manufacturers and sellers popped up across the United States.[5] Perhaps the consumers were tired of “gambling” on glucosamine and thought that the products did not live up to their billing. More likely, plaintiffs’ lawyers had been following the debate, and they believed there were issues to be asserted in the form of consumer class actions.[6] Multiple false advertising lawsuits against GNC commenced between March and December 2013, with allegations of consumer deception across California, Florida, Ohio, New York, Illinois, Pennsylvania, and New Jersey.[7]

The controversy surrounding truthfulness in supplement advertising has only grown louder in recent years, and for good reason. Over the last fifteen years, sales of vitamins, minerals, and nutritional and herbal supplements — which, together, comprise the dietary supplement industry — have surged. In December 2013, McKinsey & Company estimated the global value of the supplement market to be $82 billion, with a high concentration of that value in the U.S.[8] High levels of sales appear to correlate with high consumer confidence. The majority of U.S. adults — 68 percent — take supplements, and nearly 85 percent of those consumers express “overall confidence in the safety, quality and effectiveness of dietary supplements.”[9] Competition in the supplement industry is fierce and no one company accounts for more than a five percent share.[10]

Unbeknownst to many consumers, the laws governing supplement labeling and advertising are notoriously tricky to navigate. The Food and Drug Administration (FDA) issues rules and regulations regarding supplement labeling, marketing, and safety,[11] but there are significant limitations to FDA oversight. For example, the manufacturer or seller does not need to prove that a claim is accurate to the FDA’s satisfaction before it appears on the product,[12] and the agency reviews substantiation for claims periodically, as resources permit.[13] The Federal Trade Commission (FTC) polices health and safety claims made in advertising for dietary supplements,[14] but its oversight is similarly limited by the availability of resources.[15]

High sales, patchy government supervision, and dubious marketing practices have created a “perfect storm” on the litigation frontier. False advertising in the consumer class action context is a rapidly growing area of law.[16] GNC is a rich case study for exploring issues related to truth in advertising, particularly when dealing with scientific controversy. In GNC, the Fourth Circuit held that the existence of a single expert in agreement with a claim bars the statement from being found literally false.[17] Since Plaintiffs did not challenge the validity of GNC’s study that supported the claims found on bottles — even after an opportunity to amend their complaint — they could not prove that glucosamine-chondroitin was not beneficial to joint health. In other words, at least one reasonable expert or study is sufficient to carry the claim on the bottle, even if the vast weight of scientific evidence suggests otherwise. Furthermore, a “battle of the experts” did not necessitate a jury trial, nor did it forestall judgment in the supplement manufacturer’s favor.

This Note argues that the Fourth Circuit’s holding in GNC is the preferred solution to the truth-in-advertising dilemma that is rampant in the supplement industry, particularly in periods of scientific controversy. Scientific evidence is neither static nor consistent; there can be two pools of reasonable evidence on either side of a medical controversy, and the minority position may well be proven right. Reserving judgment for juries beyond the pleadings stage would overestimate juries’ abilities to resolve highly technical scientific controversies. Both consumers and retailers benefit from consistent interpretation of consumer protection measures, and the Fourth Circuit has articulated a workable standard for these various state law claims.

The argument unfolds in three parts. Part I surveys the landscape of advertising laws, with heightened focus on the Lanham Act and state consumer deception statutes. Supplement advertising should be treated as a carve-out amidst this patchwork of laws, and I will argue that the Fourth Circuit’s ruling, if appropriately limited, does no damage to the existing doctrines. Part II offers a detailed description of the GNC case, both at the district and appellate court levels. This background serves to illustrate why the unanimous GNC judgment is an interesting solution to a difficult truth-in-advertising question. Part III assesses the policy implications of the Fourth Circuit’s decision. It begins by addressing criticism of the GNC ruling, which was presented in an Amicus Brief submitted by a prominent group of law professors. It then responds to the criticism provided in the Amicus Brief and counters with the merits of the Fourth Circuit’s holding. The Conclusion reviews the argument and demonstrates why the analysis presented is persuasive.

Part I: The Legal Landscape of the Supplement Industry

A. Sources of Truth-in-Advertising Law

There is a patchwork of federal, state, regulatory, and common law sources of truth-in-advertising law, in addition to industry-specific trade associations that advise dietary supplement manufacturers and retailers. Both the FTC and FDA[18] can initiate their own investigations into false advertising and labeling of dietary supplements. Their overlapping jurisdiction and shared enforcement responsibilities are explained further in Section I.B., infra. However, since actual oversight by these bodies is notoriously constrained, consumers and business competitors have often resorted to alternative forms of legal redress. Companies may challenge claims made by their competitors under the federal trademark law, Lanham Act § 43(a), discussed in Section I.C., infra.[19] While the federal cause of action is not available to consumers, the Lanham Act remains instructive in understanding state deception laws that are available to consumers.[20] The relationship between the Lanham Act and state consumer protection laws is explored further in Section I.D., infra. Lastly, the dietary supplement industry finds guidance from self-regulatory associations: the Council for Responsible Nutrition (CRN) is its primary trade association,[21] and the Better Business Bureau’s National Advertising Division (NAD) offers alternative forms of dispute resolution. [22]

B. The FTC-FDA Regulatory Regime

Congress has entrusted matters of food, drug, and supplement policing to both the FTC and FDA,[23] which have overlapping jurisdiction and work together to ensure consistency in consumer products.[24] “In 1971, the agencies issued a memorandum of understanding under which the FTC assumed primary responsibility for advertising and the FDA assumed primary responsibility for labeling of food, medical devices, and cosmetics.”[25] Advertising in the FTC’s domain includes print and broadcast ads, infomercials, catalogs, and similar direct marketing materials; labeling refers to the information panels on product itself, such as the nutritional content and manufacturer address.[26]

There is an enormous disparity between what the law permits and what the agencies enforce. Dietary supplements have posed a unique challenge to the joint FTC-FDA working relationship. Neither food nor drug, dietary supplements occupied a “liminal” regulatory category for much of the 20th century.[27] By the 1990s, however, Congress overrode attempts by the FDA to regulate supplements more extensively. The resulting legislation paved the way for the modern explosion of supplement products on the American market.[28]

1. FTC Regulation of Supplement Advertising

The FTC is the federal consumer protection agency charged with safeguarding consumers against “unfair and deceptive trade practices,” under the authority granted to it in Section 5(a) of the Federal Trade Commission Act. It uses two investigative methods to challenge advertising it finds to be false: either compelling information from a potential defendant (for example, in the form of a subpoena), or requesting voluntary cooperation (in the form of an access letter). Notably, consumers do not enjoy the same authority in compelling disclosure of a corporate advertiser’s studies.

The FTC has articulated its own standard for actionable false advertising. The 1972 Pfizer doctrine[29] is an unsubstantiated theory of liability: no dissemination (of an ad claim) without prior substantiation. For an advertisement to be substantiated, the advertiser must have had a “reasonable basis” for its advertising claims before they are disseminated. For health or safety claims, which include representations made in connection with dietary supplements, the Commission has typically required a relatively high level of substantiation. In such cases, the “reasonable basis” must be “competent and reliable scientific evidence,” typically defined as “tests, analyses, research, studies, or other evidence based upon the expertise of professionals in the relevant area, that has been conducted and evaluated in an objective manner by persons qualified to do so, using procedures generally accepted in the profession to yield accurate and reliable results.”[30] The FTC reaffirmed this standard in its Policy Statement Regarding Advertising Substantiation, and there are numerous representative cases applying the doctrine.[31]

2. FDA Regulation of Supplement Labeling

Unlike drugs, dietary supplements do not need FDA approval before being legally marketed in the United States. Under the Dietary Supplement Health and Education Act (“DSHEA”) of 1994,[32] the FDA’s role in regulating the dietary supplement industry is constrained. Under DSHEA, a supplement company is responsible for determining that its products are safe and that any representations made are substantiated by adequate evidence of their truthfulness. A company does not have to provide the FDA with the evidence it relies on to substantiate safety and effectiveness unless specifically requested. Instead, the FDA has relied on a disclosure theory of consumer protection, requiring firms to identify the product as a dietary supplement and include a “Supplement Facts” panel that identifies each ingredient contained in the product.[33]

C. The Lanham Act

Congress enacted the Lanham Act (codified at 15 U.S.C. §§ 1051-1127 (2012)) in 1946. It serves both as the federal trademark law and as a primary source of truth-in-advertising regulation. Section 43(a) of the Lanham Act provides that “any person who believes that he or she is or is likely to be damaged” by a false or misleading description or representation of fact may sue.[34] The Trademark Revision Act of 1988 substantially broadened the scope of section 43(a) to cover a company’s false representations about itself and others.[35]

Section 43(a) is the core legal protection for those injured by false advertising. The Supreme Court has interpreted standing under section 43(a) to be limited to business competitors; consumers are excluded from filing suits under the Lanham Act. In Lexmark International, Inc. v. Static Components, Inc., the Supreme Court explained that standing to sue for false advertising under the Lanham Act requires pleading “injury to a commercial interest in sales or business reputation,” and that injury must be “proximately caused by the defendant’s misrepresentations.”[36] The injury cognized by Lexmark is “unfair competition” through false or misleading advertising, rather than consumer deception or confusion. Accordingly, consumers’ interests do not fall within the “zone of interests” protected by section 43(a)(1)(B).[37] As a result, plaintiffs in these lawsuits are typically commercial competitors of the alleged false advertiser.

Since remedy under the Lanham Act is unavailable to non-competitors, consumers often rely on state consumer protection statutes to bring cases against advertisers. Since these state statutes effectively police the same mischief, federal common law remains instructive in understanding the state law cause of actions.[38]

Section 43(a)(1)(B) establishes liability for two modes of advertising: advertising that is false and advertising that is misleading. Actionable representations include statements made about one’s own goods or services, as well as the commercial disparagement of others. These two categories of advertising, as well as the specific violations that they encompass, are next considered.

1. Two Modes of False Advertising

Courts are tasked with determining whether representations are either false or misleading. Both incur liability under section 43(a), but there are important doctrinal and pleading distinctions. The analysis begins with five basic elements: (i) whether the alleged misrepresentation is false or misleading, as opposed to non-actionable puffery,[39] (ii) whether a plaintiff can demonstrate actual deception or capacity for deception as a result of the falsehood, (iii) whether the falsehood or misleading information is material to a consumer’s decision to buy,[40] (iv) whether a plaintiff was harmed, either by direct diversion of its sales or by a lessening of the goodwill associated with its products, and (v) whether the falsehood or deception occurs in interstate commerce.[41] These elements are conjunctive, and a plaintiff must meet all five to trigger liability.

i. Literal Falsity

Literally false representations communicate one unambiguous message, either verbally or visually, that is untrue or unsupported. These representations violate the Lanham Act without proof of consumer deception; instead, courts presume that the buying public has received the false message.[42] Consequently, plaintiffs in these cases can avoid the time and expense of preparing consumer surveys regarding the ad’s message. The burden of proof in falsity cases is on the plaintiff, and falsity is assessed on the basis of scientific testing or related types of extrinsic evidence.[43]

A classic example of a literally false representation comes from Coca-Cola.[44] In 1982, Tropicana featured athlete Bruce Jenner in a television commercial for orange juice, squeezing juice out of an orange directly into a Tropicana carton while saying, “[i]t’s pure, pasteurized juice as it comes from the orange.” In fact, the juice was heated and, in some cases, frozen before packaging. Further, the juice did not in fact come “pasteurized” straight from the orange. The Second Circuit granted preliminary injunctive relief against the “blatantly false” statement. The plaintiff did not have to make a showing that the advertisement would mislead the consuming public.

ii. Literal Falsity by Necessary Implication

A court may evaluate an advertisement for falsity even if the representation is implied by context, rather than stated directly. This category of false advertising is referred to as false by necessary implication. Actionable advertisements of this type convey one unambiguous message. While the message is conveyed implicitly, the meaning of the message is clear and unequivocal. Ultimately, if the impression left on the viewer conflicts with reality, it is effectively treated as if that false impression was directly stated.

The Second Circuit embraced the false-by-necessary-implication doctrine for the first time in Time Warner Cable, Inc. v. DIRECTV, Inc.[45]— a case which provides an excellent illustration of this type of actionable falsity. A series of DIRECTV commercials featured celebrities touting the merits of the satellite service provider’s 1080i high definition (HD) transmission. A commercial from the “Source Matters” campaign concluded with William Shatner stating that “settling for cable would be illogical.” This statement was made in the context of surrounding text (“amazing picture quality of […] DIRECTV HD”) and images (a very clear DIRECTV picture and a far inferior picture from an anonymous second provider, though cable was obviously targeted). Without extrinsic evidence, the district court determined that Mr. Shatner’s assertion (“settling for cable would be illogical”) “could only be understood as making the literally false claim that DIRECTV HD is superior to cable HD in picture quality.”[46] In reality, however, DIRECTV and cable HD’s picture quality were equivalent. The Second Circuit upheld this ruling for Time Warner, concluding that the impression left on the viewer conflicted with reality.[47]

2. Misleading Advertising

The second category of actionable false advertising claims are, while not literally false, deemed misleading to consumers. Such ads contain representations of fact about a product or service that are ambiguous or tend to deceive consumers. The statements may be literally true, and yet have the tendency to mislead. The language of section 43(a) deems merely misleading representations equally as objectionable as those that are literally false, explicitly rendering unlawful a “false or misleading description of fact, or false or misleading representation of fact.”

There are a number of important differences between false and misleading advertising, even though both types of representations violate section 43(a). Although the categories are doctrinally distinct, the evidentiary differences are more significant in practice. Literal falsehoods — those that are false on their face — are actionable without proof of consumer deception, as in the Coca-Cola and DIRECTV cases previously considered. In a case of misleading advertising, plaintiffs must present extrinsic evidence to establish consumers’ perception of the implied falsehood.[48] Such proof typically includes consumer surveys, direct consumer testimony, or consumer comments received in the ordinary course of business.

In Vidal Sassoon, Inc. v. Bristol-Myers Co.,[49] the Second Circuit addressed literally false as well as misleading claims. This case provides a clear example of the difference between literal falsity and misleadingness. In a television commercial, a spokesperson states that “in shampoo tests with over 900 women, Body on Tap got higher ratings than Prell for body. Higher than Flex for conditioning. Higher than Sassoon for strong, healthy looking hair.” The literally false claim is that the tests involved over 900 women; in fact, only two-thirds of testers were actually adult women. The claims regarding higher satisfaction were deemed misleading based on a consumer survey assessing the message of the ad. Consumers surveyed about the commercial said that they thought each tester had tried at least two of the named products in order to compare their quality. Since this was not the actual testing procedure, the court held that consumers were deceived by the ad’s claims, and that the claims were actionable as misleading.

D. State Consumer Protection Statutes

Every state has a consumer protection law that prohibits deceptive practices.[50] Many of these statutes take the form of a “little” FTC Act, incorporating the language of Section 5 of the statute for which they are named.[51] These statutes provide the basic protections for consumers engaging in thousands of transactions across the United States, prohibiting “unfair methods of competition or unfair deceptive trade practices” as well as “all forms of fraudulent, deceptive, and unfair acts.” While many statutes track the federal guideline, they nevertheless may vary in form and strength from state to state.[52] State attorneys general and consumers may bring suits pursuant to these acts.

As mentioned infra, the state consumer deception statutes at issue in GNC were interpreted in accordance with the great body of federal common law surrounding false advertising and the Lanham Act. There remains a significant difference in the standing requirements: in a Lanham Act lawsuit, the plaintiff is typically a competitor; in putative consumer class action lawsuits, representatives must allege that they were personally deceived to have standing.[53] Despite these differences in standing, the laws have been interpreted relatively congruously. The mischief to be corrected, in both cases, is false advertising and unfair competition. However, there is no requirement that each state’s consumer deception law track the Lanham Act or federal case law.

Part II: The GNC Case and the Glucosamine-Chondroitin Problem

A. Factual Background

General Nutrition Corporation (GNC), a national nutritional products retailer, has manufactured and sold a line of joint health supplements[54] for years. These products, which list glucosamine, chondroitin, and various other compounds as the primary active ingredients, are marketed collectively under the “TriFlex” brand. Although the products differ slightly in their total combination of ingredients, they advertise similar claims[55]; essentially, the TriFlex brand improves the health, comfort, and function of joints. The label for one product, TriFlex Fast-Acting, included a ”[c]linically studied” establishment claim: a “12-week multi-center, randomized, double-blind, placebo controlled study of 60 adults [. . .] taking 250 mg/day of the GNC TriFlex Fast-Acting Blend” proved that the product was “shown to improve joint comfort and function,” in addition to promising 20% improvement in joint function and 25-30% improvement in joint flexibility.[56] GNC produced a similar line of products for Rite Aid, which claimed to “promote joint health” and “help[. . .] rebuild cartilage and lubricate joints.” In compliance with FDA guidelines, all of the TriFlex and Rite Aid products included the disclaimer: “This statement has not been evaluated by the Food and Drug Administration. This product is not intended to diagnose, treat, cure, or prevent any disease.”

Beginning in March 2013, class action lawsuits against GNC and Rite Aid popped up across the United States. Plaintiffs in California, Illinois, Florida, New York, New Jersey, Pennsylvania, and Ohio alleged violations of an array of state consumer protection, deceptive practices and/or express warranty statutes in regard to the glucosamine-chondroitin products.[57] In December 2013, the pending lawsuits were consolidated by the Judicial Panel on Multidistrict Litigation and, pursuant to 28 U.S.C. § 1407, transferred to the U.S. District Court for the District of Maryland.[58]

B. The Complaint

In the Consolidated Amended Complaint, Plaintiffs alleged that the “overwhelming weight of high quality, credible and reliable studies demonstrate that glucosamine and chondroitin…do not provide joint health benefits” (emphasis added).[59] They stated that the inefficacy of these supplements was “generally recognize[ed]” by the scientific community.[60] In support of their allegations, Plaintiffs cited to thirteen studies released between 2004 and 2013.[61] In the studies, researchers concluded that (1) glucosamine and chondroitin did not reduce symptoms for osteoarthritic users or chronic joint pain sufferers (who, Plaintiffs alleged, were an appropriate proxy for non-arthritic users), and (2) MSM, another compound found in the TriFlex products, did not provide pain or joint symptom relief for osteoarthritic consumers. Notably, the Plaintiffs did not provide any testing of GNC’s particular products or combination of ingredients. Instead, they relied on the “vast weight of competent and reliable scientific evidence” (i.e., the cited studies) to infer that the “ingredients in GNC’s TriFlex Products do not work as represented” and that the “representations [were] false.”[62]

Defendants GNC and Rite Aid filed a Motion to Dismiss,[63] arguing that the studies did not test the products (or specific combination of ingredients) at issue,[64] that no representations were made with regard to improving osteoarthritis symptoms, and that osteoarthritis patients were not an appropriate proxy for non-arthritic users, among other deficiencies. Regarding the osteoarthritis issue, Defendants specifically cited to the FDA disclaimer on all of the labels, which state that the products are “not intended to diagnose, treat, cure, or prevent any disease.”[65]

C. Decision Granting Motion to Dismiss

In spite of Plaintiffs’ allegations, District Court Judge J. Frederick Motz dismissed the Complaint with leave to amend. He found that Plaintiffs’ allegations regarding the “vast weight of scientific evidence” — in light of GNC’s study to the contrary — could not support the single conclusion that the claims are false.[66] Under the Twombly/Iqbal plausibility standard, there was a “fatal flaw” in the allegations of the CAC: it did not allege that “experts in the field” were prepared to testify that, on the basis of the existing scientific evidence, any reasonable expert would conclude from the cited studies that glucosamine and chondroitin are ineffective in non-arthritic consumers.[67] The “mere existence of a ‘battle of the experts’” was not proof of falsity, but was rather to be “expected.”[68] The following excerpt from Judge Motz’s Memorandum Decision provides an important glimpse into the Court’s reasoning:

“Disagreements between experts, even under the ‘reasonable degree of scientific certainty’ standard, are to be expected. In my judgment, however, the fact that one set of experts may disagree with the opinions expressed by other qualified experts does not ipso facto establish any violation of the applicable consumer protection laws. If there are experts who support what defendants say in their advertisements, the advertisements are not false and misleading . . . unless the clinical trial relied upon by defendants was itself false and/or deceptive.”[69]

In this passage, the Court referenced the “reasonable basis” standard that marketers must adhere to when preparing advertising claims (discussed later in Section III, supra). Stated briefly, if scientific evidence points in different directions, the reasonable basis standard will allow for inconsistent messaging. In a footnote to this excerpt, the Court addressed whether this is a burden of proof issue and concluded that it is not, based on the nature of the advertising claims in the case.[70] The Court contrasted the example to a product liability case, in which plaintiff must establish that a product is defective. There, Judge Motz said, it would be “entirely appropriate for a jury to decide the defect issue” based on expert testimony that, to a reasonable degree of scientific certainty, a product is defective.[71] This distinction is enormously significant for procedural and substantive reasons, reserving for juries their fact-finder role beyond false advertising cases.

By granting the Plaintiffs leave to amend their complaint, Judge Motz offered Plaintiffs an opportunity to revive their claim for literal falsity. He specifically instructed that Plaintiffs must plead that the clinical trial relied on by GNC (1) did not exist at all, (2) exists but did not support any of GNC’s representations about TriFlex, or (3) exists and supports the assertions on TriFlex Fast-Acting’s bottle, but was not conducted in an appropriately scientific manner.[72] Only under such a pleading could the Court infer, on the face of the complaint, that the products are ineffective as to non-arthritic users. Otherwise, if at least one expert supported what GNC and Rite Aid said in their ads, the advertisements could not be false.[73] Plaintiffs could only file an amended complaint if they could do so in accordance with Fed. R. Civ. P. 11, which requires sufficient due diligence to avoid sanctions.[74] Absent such a pleading, Plaintiffs’ claim of falsity would not be facially plausible.

Six days after the Motion to Dismiss was granted, GNC’s counsel sent plaintiffs a letter contending that “qualified experts” support the TriFlex and Rite Aid products’ claims.[75] Plaintiffs rejected the opportunity to amend, and instead filed a motion for reconsideration on the basis of the existing complaint.[76] Judge Motz denied the motion and reiterated the rationale presented in the Memorandum Decision. Regarding policy, he added one additional reason to his earlier holding: It would be unfair to consumers who wish to gamble on glucosamine and chondroitin if lay juries could effectively ban the sale (or artificially raise its pricing) simply because evidence of their effectiveness is inconclusive.[77] Plaintiffs appealed to the Fourth Circuit.

D. Plaintiffs’ Appeal

After oral argument before a three-judge panel, the Fourth Circuit unanimously affirmed the District Court’s decision in favor of GNC.[78] The Fourth Circuit inexplicably held that “plaintiffs must allege that all reasonable experts in the field agree that the representations are false.”[79] In other words, a single expert in disagreement bars a statement from being literally false. In the Court’s words:

“[I]n order to state a false advertising claim on a theory that representations have been proven false, plaintiffs must allege that all reasonable experts in the field agree that the representations are false. If plaintiffs cannot do so because the scientific evidence is equivocal, they have failed to plead that the representations based on this disputed scientific evidence are false.”[80]

Under the Fourth Circuit’s test, “the question of falsity hinges on the existence (or not) of scientific consensus.”[81] Scientific experts may — and often do — disagree about the truthfulness of a statement, but equivocalness is not falsity.[82] Regarding Plaintiffs’ argument that a “battle of the experts” could not be resolved on the pleadings, the Court retorted:

“When litigants concede that some reasonable and duly qualified scientific experts agree with a scientific proposition, they cannot also argue that the proposition is ‘literally false.’ Either the experts supporting the Companies are unreasonable and unqualified (in which case, there is no real battle of the experts to begin with) or they reflect a reasonable difference of scientific opinion (in which case the challenged representations cannot be said to be literally false).”[83]

Indeed, by characterizing the dispute as a “battle of the experts,” the Court held that Plaintiffs (inadvertently) conceded that “a reasonable difference of scientific opinion exists as to whether glucosamine and chondroitin can provide the advertised joint health benefits.”[84] The Fourth Circuit also responded to Plaintiffs’ concern that manufacturers might hide behind so-called experts in proffering dubious marketing claims, relying on the Federal Rules of Evidence to ensure relevant and reliable scientific testimony.[85]

Part III: Legal and Policy Considerations

A. Law Professors’ Amicus Brief, with Rebuttal

After the Circuit Court handed down its GNC decision, sixteen prominent law professors submitted an Amicus Brief for the Plaintiffs in support of rehearing.[86] The Amici describe their interest in the case as “promoting truth in advertising, which protects consumers and promotes fair competition.”[87] In the eight-page brief, the law professors[88] supported Plaintiffs’ petition for rehearing and criticized the Fourth Circuit’s decision. The Amici argued that the Fourth Circuit “erred when it disregarded binding precedent” to produce a ruling with adverse procedural and substantive consequences. They further argued that questions of falsity cannot be resolved on the pleadings; instead, the fact finder (presumably a jury) should evaluate competing experts and pools of scientific evidence to determine the truth of a claim.

The brief is problematic in numerous respects. Specifically, the Amici (i) misstate the legal standards in falsity cases, (ii) misstate the nature of “binding” precedent, (iii) mischaracterize the nature of scientific evidence, and (iv) fail to appropriately consider the impact of the suggested alternative on jury instructions. This Note argues that the Amicus Brief, as well as the alternative outcome that Amici support, is a less persuasive answer to the truth-in-advertising dilemma presented by the GNC case. Instead, the Amici’s criticism offers insight into the utility and strengths of the Fourth Circuit’s logic. This section presents the Amici’s arguments and responds to points addressed in their brief.

1. Inaccurate Legal Standards

The Amicus Brief begins with a “Summary of Argument” that describes the purported standard in literal falsity cases: “Literal falsity is about how an advertisement is received by consumers. The adjectives ‘literal,’ ‘explicit,’ and ‘implicit’ (and falsity ‘by necessary implication’) describe consumer reaction to a message, which is either proved by evidence such as surveys or presumed as a matter of law.”[89] The italicized portions of this excerpt are problematic and, pun intended, literally false. First, literal and explicit claims are not assessed on the basis of consumer perception; rather they are accepted as literal messages. Second, literal and explicit claims are evaluated based on underlying substantiation, such as scientific testing, regardless of the message received by consumers. These important doctrinal and evidentiary distinctions are reviewed in Section I and, briefly, below.

As mentioned in Part I, infra, two modes of advertising are liable under Section 43(a): advertising that is false, and advertising that is misleading. Within the umbrella of false advertising, courts have generally recognized two types of falsity: (i) advertising that is literally false (based on a clear and explicit misrepresentation of fact), and (ii) advertising that is false by necessary implication (based on a claim that, considered in context, necessarily and unambiguously implies a message). Neither case requires surveys or other proof of consumer deception. Amici’s emphasis on “consumer reaction to a message” and “surveys” is misplaced, as both of those issues are irrelevant to a finding of literal falsity.

Amici then describe an example of a potentially misleading representation: “Vitamin B7 can remedy hair loss.” Since implied messages can convey multiple messages to the buying public, evidence of consumer confusion is required to determine which message consumers are receiving. While a discussion of misleading advertising can be helpful for comparison purposes, it is not relevant in the GNC case. First, the TriFlex claims at issue are literal and explicit: for example, “promotes joint mobility and flexibility” and “protects joints from wear and tear of exercise.” Second, there was no survey evidence of the messaging received by consumers, which would be required to sustain a cause of action for misleading representations.[90]

There are subsequent discussions regarding advertising that is false by necessary implication.[91] While actionable, this mode of deception is also irrelevant to the GNC case. The claims at issue on the TriFlex label are clear and explicit. Advertisements that are false by necessary implication do not involve an explicit statement of fact, but rather an unambiguous assumption that follows from a claim. Consider a comparison to the claims at issue in DIRECTV v. Time Warner Cable, discussed in Part I.C., infra. In that case, the court considered other words and images within the context of the statement that “settling for cable would be illogical;” the necessary implication was that settling for cable would be illogical because DIRECTV offers superior picture quality (it did not). The clear and explicit TriFlex claims simply do not fit into this category of consumer deception.

2. Misstatement of Precedent

The Amicus Brief continues with a citation to “binding precedent,” namely C.B. Fleet Co. v. SmithKline Beecham Consumer Healthcare, L.P., which the Fourth Circuit courts allegedly “disregarded” in dismissing the GNC case. In the opening paragraph of the Amicus Brief’s “Argument,” Amici state: “This Court has repeatedly held that both inquiries involve questions of fact. C.B. Fleet Co. v. SmithKline Beecham Consumer Healthcare, L.P., 131 F.3d 430, 434 (4th Cir. 1997) (“Whether an advertisement is literally false is an issue of fact.”).” This is both a mischaracterization of the holding as well as the case’s precedential value.

Amici appropriately classify C.B. Fleet as a literal falsity case.[92] At issue were “improved design” and “comparative superiority” claims for SmithKline’s new model of a feminine hygiene product. The Amici presumably provide this quotation in support of the alternative outcome they propose: factual disputes, such as a battle of the experts, should be reserved for juries. However, this proposition is conceptually distinct from the “issue of fact” identified by the C.B. Fleet Court. In C.B. Fleet, the Fourth Circuit was reviewing the district court’s determinations of falsity on appeal; the quote provided by Amici stands for the proposition that district courts, in their fact-finding role, are given deference in reaching such conclusions.[93] Furthermore, there is a citation following the C.B. Fleet reference — omitted without acknowledgment in the Amicus Brief — to L&F Prods. v. Procter & Gamble Co., 45 F.3d 709, 712 (2d Cir. 1995). That underlying authority from L&F Products states as follows: “The district court’s determination with respect to facial falsity is a finding of fact which we review for clear error.”[94] In both C.B. Fleet and L&F Products, the “issue of fact” is raised with respect to the deference due to district courts, rather than juries’ roles in resolving factual disputes.

In fact, the definition of falsity in C.B. Fleet does not conflict with the determination made in GNC. However, there remains a crucial factual difference between the cases. As the GNC Court noted, plaintiffs were given leave to amend the Consolidated Amended Complaint to plead that GNC’s study (1) did not exist at all, (2) existed but did not support any of GNC’s representations about TriFlex, or (3) existed and supported the assertions on TriFlex Fast-Acting’s bottle, but was not conducted in an appropriately scientific manner. The Court specifically instructed that plaintiffs could not rely on the “vast weight of the evidence” to prove falsity. In C.B. Fleet, plaintiff met this burden, challenging the scientific reliability of SmithKline’s testing procedures. In GNC, by contrast, plaintiffs twice relied on the “weight of the evidence” without attacking GNC’s testing – even after a second bite at the apple.

Although the Amici state that the Fourth Circuit has “repeatedly held that [the falsity inquiry] involve[s] questions of fact,” only C.B. Fleet is offered as support. One might reasonably expect a string cite supporting such an assertion. While the Fourth Circuit cemented a new rule for the meaning of literal falsity in supplement cases, its definition is hardly inconsistent with “binding” and “disregarded” precedent. In fact, the issue presented in GNC was a matter of first impression before the Court. While Amici’s proposal is a permissible solution to the GNC question, it is neither required by precedent nor, as this Note argues, able to achieve a better outcome on policy grounds. The Fourth Circuit’s ruling is a novel but befitting solution to truth-in-advertising questions in the dietary supplement industry.

3. The Nature of Scientific Evidence

The most persuasive portion of the Amicus Brief comes in the form of a hypothetical regarding the medical cause of ulcers. Amici present the following scenario: In 1910, a doctor would have said that stress and diet caused stomach ulcers, and that bacteria did not. We now know that this is untrue; the bacterium H. pylori — not stress and diet — causes many stomach ulcers. [95] The Amici use this illustration to argue that, despite expert support for a claim, it may be false. The fear is that corporate defendants may insulate themselves from lawsuits by hiding behind one unreliable study. This illustration lends support to Amici’s claim that experts are not infallible, and perhaps that juries serve as a valuable check on expert elitism.

Understood another way, the ulcer hypothesis could be marshaled in support of the Fourth Circuit’s ruling. Indeed, scientific knowledge is always a work in progress, and our understanding of medical issues is constantly evolving in the face of new research and methodology. Consider an alternative analysis of the ulcer illustration. In 1910, the majority of doctors, based on our understanding of ulcers at the time, would have said that the claim that bacteria cause ulcers is literally false. In the decades that followed, medical explanation for ulcers changed in light of evolving research. By 1950, the weight of scientific evidence would indicate that bacteria do, in fact, cause ulcers. Today, it is scientific truth that H. pylori is the culprit behind stomach ulcers — a proposition that would have been literally false in 1910 if the “vast weight of scientific evidence” set the standard in falsity cases.

The point is that the minority opinion in a scientific debate may turn out to be correct. The GNC holding corrects for the mistaken understanding that a minority position is wrong simply because it is not supported by a majority of scientists at that time. In a 1910 jury trial, the jury would likely — and incorrectly — have discounted the minority position because it would not meet the preponderance of the evidence standard. Then, the question becomes, why should we let a jury decide such matters? In the ulcer case, how and why would a jury have reached a better result?

The Fourth Circuit recognized a key difference between claims that, in light of the existing scientific evidence, are inherently false, as opposed to claims that are reasonably debatable. The GNC standard is most receptive to the changing nature of scientific evidence and techniques. Until consensus among duly qualified and reasonable experts emerges around the veracity of a claim, it would simply be premature to rule out the minority opinion as false. The ulcer anecdote, perhaps inadvertently, stands for this proposition.

Amici would likely retort that the GNC rule too easily insulates corporate advertisers, hiding behind a single study, in literal falsity cases. However, as the GNC Court noted, plaintiffs remain protected due to the rules of evidence and by alternative causes of action. As Judge Floyd explained, “Plaintiffs remain protected from dubious experts by the Federal Rules of Evidence, which ‘ensure that any and all scientific testimony…is not only relevant, but reliable.’” The importance of relevant and reliable scientific testing cannot be overstated, since plaintiff must prove that no reasonable scientific expert could find merit in the advertiser’s claims. The Fourth Circuit is careful to rule out quackery as a source of substantiation. If no reasonable study exists in support of a proposition, then the plaintiff meets their burden of proof in literal falsity cases. Additionally, Judge Floyd noted that “[p]laintiffs who cannot meet the burden of proving literal falsity may avail themselves of a claim regarding misleading representations.” This is helpful to plaintiffs who cannot or do not challenge the scientific legitimacy of a claim, but consider the messaging to consumers to be deceptive. Plaintiffs may then rely on survey evidence to establish consumer deception or confusion regarding an ad claim.

4. Jury Instructions

Finally, a crucial issue is not addressed in the Amicus Brief, but is relevant to the alternative outcome that Amici support: if the GNC case were assigned to a jury, what would the jury instructions look like? Juries undeniably serve an important fact-finding role in courtrooms across the United States, but play no part in the GNC story. Amici aptly draw attention to this procedural predicament, but serious questions remain if a panel of jurors would have reached a better outcome in the glucosamine case.

GNC is procedurally significant because the Court ruled, on a motion to dismiss, that a literal falsity case could be resolved on the pleadings. The Fourth Circuit articulated a very specific pleading standard with respect to plausibility in falsity cases. To recall, future plaintiffs must allege that no reasonable study (or scientist) exists or supports the challenged advertising claims; reliance on the so-called “vast weight of evidence” is insufficient to survive a motion to dismiss. In so ruling, the Fourth Circuit held that a jury would not resolve conflicting disagreements among experts.

One could say, as Plaintiffs argued, that a battle of opinions among qualified and competent experts creates a genuine issue of material fact, and that juries can handle highly technical scientific evidence. The involvement of juries seems even more urgent when considering the resource limitations of the FTC and FDA. GNC’s resolution at the plausibility stage usurped the jury of its central province,[96] and corporate defendants can more easily evade liability from the FTC, FDA, and courts alike.

While plaintiffs’ argument prevails in principle, it has less purchase in practice. When there is reasonable evidence on both sides of a scientific controversy, why should six jurors decide if a product cannot be sold or advertised? Reserving judgment for juries beyond the pleadings stage would overestimate juries’ abilities to resolve highly technical scientific controversies. However, if plaintiffs plead issues related to the credibility or reliability of the defendant’s studies (which the GNC plaintiffs did not), a jury’s intervention seems far more befitting. In such case, if the evidence on one side is unreliable, there is no “battle of the experts” to begin with, and we can trust juries to side with the (only) party bearing substantiation. GNC preserves this crucial distinction beyond the pleadings stage.

Lastly, the GNC holding should be cabined to the dietary supplement area. Judges, academics (including Amici), and others would likely agree with this assertion. The GNC ruling does not rob the jury of its fact-finder role in cases involving product liability, wrongful death, and related matters. The District Court specifically addressed this concern, distinguishing the question in GNC from other burden of proof issues. In the words of Judge Motz:

“I have considered whether the issue is one of burden of proof, and I have concluded that it is not. Rather, the basis of my holding lies in the nature of the claims asserted by plaintiffs that rely upon the falsity, deceptiveness, or unfairness of defendants’ advertisements. In contrast, for example, in a product liability case in which a plaintiff must establish that a product is defective, it would be entirely appropriate for the jury to decide the defect on the basis of expert testimony that, to a reasonable degree of scientific certainty, a product is defective.”[97]

As is evident in the Court’s language, the decision is meant to be narrowly construed, and this message appears to have been received by other courts. As of February 2018, the GNC holding has not been applied outside of advertising cases.

B. Additional Merits of the Fourth Circuit’s Decision

There is useful criticism of the GNC ruling from Amici and others.[98] On the one hand, the GNC ruling seems like a massive victory for corporate defendants — and a massive loss for consumer plaintiffs — on both substantive and procedural grounds. GNC arguably imposes a high hurdle on consumer plaintiffs in literal falsity cases; without alleging perfect scientific consensus or attacking defendant’s study, plaintiff’s case can no longer survive a motion to dismiss. But this is only part of the story.

Prior to the GNC decision, the question of what constituted false advertising during a period of scientific controversy was unanswered by the case law. Despite the prevailing criticism, the Fourth Circuit’s position is a persuasive solution to a previously unresolved legal dilemma. The arguments that follow address the policy merits of the GNC standard for literal falsity in supplement cases. Specifically, this Section argues that (i) the Fourth Circuit’s standard fits neatly into the patchwork of pre-existing consumer protection measures, namely those of the FTC and FDA, (ii) the decision promotes fair competition and preserves consumer choice, and (iii) the res judicata effect in these cases is limited, preserving opportunities for future plaintiffs to take a bite at the proverbial apple as scientific knowledge evolves.

1. FTC, FDA, and Fourth Circuit Parity

The falsity standard articulated in GNC is consistent with other consumer protection measures, namely, the FTC and FDA’s approaches to truth-in-advertising. The Fourth Circuit’s holding reflects the policy rationales underlying the FTC’s prior substantiation doctrine and the FDA’s disclosure regime. Both agencies, which are most directly involved in regulating consumer communications about supplements, would not take issue with the TriFlex ads.

Under the FTC’s advertising standards, the TriFlex claims are truthfully advertised. The FTC’s prior substantiation rule, discussed in Section I.B., states that an advertiser must have a “reasonable basis” for making objective claims and must “possess substantiation, prior to running the ad, for affirmative product claims.”[99] GNC cited to a particular study in its possession on the bottle of one of the TriFlex products. Plaintiffs nowhere alleged that GNC did not have a reasonable basis for making the joint health improvement claims; they failed to allege this deficiency even after the District Court granted them leave to amend their complaint. While the FTC has investigated other glucosamine-chondroitin products, the author has no information indicating that the FTC acted (or would need to act) with respect to the TriFlex claims.

GNC was also compliant with the FDA’s supplement labeling requirements, since each TriFlex label included the ingredients in each bottle, the standard disclaimer, and other required information. In broader perspective, the Fourth Circuit’s ruling is consistent with the FDA’s position on supplement labeling and advertising. Consider the evolution of the FDA’s approach to regulating the supplement industry. The 1906 Federal Food and Drugs Act used the label as a tool to empower consumers to have a much better conception of what they were putting into their bodies. In the 1930s, the FDA confronted the question of allowable representations in the face of scientific uncertainty, further extending the “informative labeling” approach: representations could be made, even if not reflective of the consensus of scientific opinion, as long as there was reliable scientific basis behind them. In such cases, the statements would have to be qualified with a disclaimer that reliable scientific opinion differed on such claims.[100] The goal was to preserve freedom of consumer choice, with the agency’s primary concern being the safety of the products. The Fourth Circuit’s ruling follows the hands-off approach that the FDA has taken towards safe and properly labeled dietary supplements.

Sources of truth-in-advertising law are too numerous to recount here, but the Fourth Circuit’s position achieves congruity across axes. Both the District Court and the Fourth Circuit interpreted the state consumer protection laws similarly; the fact that the laws come from different states is a distinction without a difference.[101] The overall objective of these statutes remains the same: protecting consumers from false and misleading advertising. It seems reasonable that a federal court seeking guidance on false advertising issues would look to the Lanham Act on regulating advertising and consumer communications. In view of these considerations, an ad that is false under the Lanham Act should likewise be deemed false under a state consumer deception statute. Furthermore, both industry and consumers benefit from consistent interpretations of state consumer protection statutes, which provide adequate notice to both parties about pleading requirements and possible defenses.

2. Fair Competition and Consumer Choice

Over the last decade, dozens of false advertising lawsuits have been brought against purveyors of glucosamine-chondroitin. The claims typically reflect the conventional wisdom about improving joint health and flexibility, just as GNC’s TriFlex did. These cases across the United States are undoubtedly expensive for advertisers to defend. Ultimately, the price is borne by the consumer who continues to “gamble” on glucosamine because, in her subjective opinion, it works. Both GNC decisions consider this economic dimension, limiting litigation beyond the motion to dismiss stage for cases in which there are two legitimate medical understandings of a supplement’s efficacy. Furthermore, from a policy perspective, it seems unadvisable to “put out of business” an approach for which there is a sound medical view.

As Judge Motz aptly noted in his Memorandum Decision, “It is unfair to consumers who wish to gamble that glucosamine and chondroitin may be effective if lay juries can effectively ban the sale of glucosamine and chondroitin simply because the evidence of their effectiveness is inconclusive.” Satisfied consumers are free to purchase the products, and dissatisfied consumers are, of course, free to seek alternatives. Unlike drugs, which operate under completely different advertising standards, consumers are not required to take supplements. The perspective of jurors adds no value to this highly individual decision. The Court continued: “After all, damage awards and even the cost of defending against high stakes litigation has the effect of increasing the cost of glucosamine/chondroitin pills or, potentially, driving the pills from the market. Should those who choose to purchase the pills have to pay more for them (or be deprived of the opportunity to purchase them at all) when the science is uncertain merely because juries disagree with their own judgment about the pills’ efficacy?”[102]

In addition to preserving consumer choice, the GNC ruling respects that whether the supplements work is an enormously subjective inquiry. In another case involving glucosamine-chondroitin products, the court neatly summarized the dilemma:

“The health and comfort of joints is probably influenced by a number of variables. Did [plaintiff] keep all of them constant, adjust for ones that can’t be kept constant (like aging), and then somehow have her cartilage and joints examined? Did she keep precise records of how much [glucosamine-chondroitin] she took, why she took it, and just how long she took it for? Can she document what her physical condition was before and after she took [glucosamine-chondroitin]? Probably not. What’s more likely is that she took [glucosamine-chondroitin] casually and just didn’t feel much better, but that makes her own claims just as speculative as she alleges [the supplement’s] benefits are.”[103]

We could, as Amici argue, send these cases to juries and let them decide if the minority view does not hold water. But should we? The Court asked and answered this very question, stating: juries serve as “a proper check on expert elitism. However, in this case the question is not whether the views of jurors should prevail over the views of asserted experts and judges. Rather the question is whether the views of jurors should prevail over the views of those who choose to purchase glucosamine/chondroitin pills. What is ‘democratic’ in one instance may be tyrannical in another. . . .”[104]

Finally, there is a hidden narrative beneath these lawsuits. At first glance, it may seem that consumers are fired up about wasting money on useless products. However, a closer look at the court dockets suggests an alternative story: many of the glucosamine-chondroitin lawsuits seem to be brought by the same lawyers, not consumers. Bonnett, Fairbourn and Denlea & Carton have been heavily involved in various glucosamine-chondroitin product lawsuits. The results have been lucrative for law firms. In the Plaintiffs’ Memorandum supporting its Motion for Reconsideration, counsel noted “at least three nationwide settlements against manufacturers of similar joint relief supplements are in various stages of review.”[105] Should law firms be able to line their pockets from consumers’ purses? The FDA, FTC, state attorneys general, and corporate competitors are better safeguards of truth-in-advertising than plaintiffs’ firms would care to admit.

3. Res Judicata Effects

Critics of the GNC ruling may rest assured that claim preclusion in false advertising cases is limited to the named plaintiff(s). The res judicata effect does not bind anyone to the result aside from the person or persons who brought the suit. If a party sued GNC for its TriFlex claims in Texas today, nothing is settled, and the court is well within its discretion to hear the case. It is particularly important to preserve these opportunities for litigation as scientific knowledge evolves and the conventional wisdom shifts.

On the other hand, the limits of the res judicata effects in these cases reveal a troubling pattern. As mentioned previously, there have been dozens of lawsuits across the United States involving glucosamine-chondroitin supplements, many of which were brought within the last ten years. These cases, which have not utilized the GNC pleading standard, have resulted in disparate verdicts and settlements. For example, the glucosamine-chondroitin cases in California have come out all over the place: it is entirely possible that a Los Angeles jury would find the claims to be truthful, and a jury in San Francisco would deem them falsely advertised. As a result, glucosamine products may bear dissimilar advertising claims depending solely on the location where the product is purchased. This result seems arbitrary and premature, given the conflicting scientific evidence on the claims at issue.


During periods of scientific controversy, how should advertisers, regulators, and courts address truth-in-advertising dilemmas? As the GNC case has shown, this is a complex and interesting issue that is worth examining. The Fourth Circuit’s answer to this question is a fair and reasonable standard, relying on expert consensus to establish literal falsity. While respected academics and courts have greeted the Fourth Circuit’s ruling with caution, advertisers and consumers may be more satisfied with its outcome. Corporate defendants are once again reminded of the prior substantiation requirement in articulating advertising claims and can, with assurance, hang their hats on reasonable scientific testing. Consumers enjoy the continued availability of supplements that, in their subjective opinions, provide symptom relief – without paying for the extra costs associated with consumer class actions. Consumer plaintiffs are also given clear guidance on surviving a motion to dismiss, as the Fourth Circuit neatly articulated the pleading requirements in alleging literal falsity.

The novelty and significance of the GNC decision are only beginning to be understood, but it offers compelling policy advantages to proposed alternatives. The Fourth Circuit’s ruling fits into the preexisting patchwork of consumer protection measures, announcing similar principles to the FTC’s substantiation theory and the FDA’s disclosure theory of liability. The solution is well suited to the nature of scientific evidence, which is continually changing in response to new research and technology. It is hard to argue that, if qualified scientists disagree about the efficacy of a dietary supplement, lay juries would achieve a better outcome in determining truth from falsity. Elevating the opinions of six jurors over experts and consumers does not provide any more certainty in arriving at the truthful result.

Questions remain about what impact, if any, the GNC decision should have on false advertising law. The author hopes that this unanimous decision by a panel of respected appellate judges will receive thoughtful consideration. In broader perspective, the standard does no damage to the existing falsity doctrines, and can be carefully confined to the dietary supplement space. The new standard articulated in GNC may seem like a big pill to swallow, but it is an effective remedy to the truth-in-advertising question that plagued the dietary supplementary industry.

*J.D. Candidate, New York University School of Law, 2018; B.A., Political Science, Columbia University, 2011. The author would like to thank Kenneth A. Plevan for his guidance and support, as well as the invitation to explore this research topic. She thanks Professor Barton Beebe, her faculty advisor, for his edits and insights, and her fellow JIPEL Notes Program participants: Julian Pymento, Ryan Jin, Vincent Honrubia, and Neil Yap.

[1] National Institutes of Health, Dietary Supplements: What You Need To Know, U.S. Dep’t of Health & Hum. Servs. (June 17, 2011),

[2] See National Institutes of Health, Multivitamin/mineral Supplements: Fact Sheet for Health Professionals, U.S. Dep’t of Health & Hum. Servs. (July 8, 2015), (“In 2014, sales of all dietary supplements in the United States totaled an estimated $36.7 billion.”).

[3] See Brown v. GNC Corp. (In re GNC Corp.), 789 F.3d 505 (4th Cir. 2015), reh’g denied Ct. Order Den. Mot. for Reh’g and Reh’g En Banc (July 27, 2015), ECF No. 47.

[4] There is a well-publicized association between glucosamine-chondroitin and joint health. An article on WebMD states that the natural glucosamine in our bodies “helps keep up the health of your cartilage – the rubbery tissue that cushions bones at your joints.” As we age, levels of the natural compound begin to drop, which leads to the gradual breakdown of the joint. WebMD reports that there is “some evidence” that glucosamine sulfate supplements help counteract this problem, though experts “aren’t sure how.” There are a plethora of glucosamine supplements advertising joint health benefits, such as Osteo Bi-flex and Nature’s Bounty glucosamine-chondroitin compound. The widespread use of glucosamine-chondroitin has even reached our pets, with supermarket giant Trader Joe’s and others proffering a line of the supplement for dogs. Is Glucosamine Good for Joint Pain?, WebMD (Jan. 17, 2018),

[5] The glucosamine-chondroitin lawsuits are too numerous to list here, but are discussed later in Section III.B. See, e.g., Lerma v. Schiff Nutrition Int’l, Inc., No. 11-cv-1056 (S.D. Cal. filed May 13, 2011); Padilla v. Costco Wholesale Corp., No. 11-C-7686 (N.D. Ill. filed Oct. 28, 2011); Quinn v. Walgreen Co., No. 12-cv-8187 (S.D.N.Y. filed Nov. 9, 2012). According to, at least nine class action lawsuits had been filed by October 2013 claiming that companies were falsely marketing the health benefits of glucosamine supplements. Consumers Claim This Joint Is Not Jumping, Truth in Advertising Organization (Oct. 22, 2013), Some of the lawsuits have resulted in settlements. See, e.g., McCrary v. The Elations Company, LLC, No. 13-cv-00242 (C.D. Cal. filed Feb. 07, 2013); Pearson v. NBTY, Inc., No. 1:11-cv-07972 (N.D. Ill. filed Nov. 09, 2011). Additionally, there is no indication that these lawsuits will stop being filed. As of December 2017, glucosamine-chondroitin is still a hot topic in false advertising. According to a consumer class action blog, plaintiffs’ firms have commenced investigations into Osteo Bi-flex, Schiff Move Free and Glucosamine, Walmart’s Spring Valley Glucosamine, and Nature Made Triple Flex, among others. Tracy Colman, Different Brands of Glucosamine Chondroitin May Be Falsely Advertised, Top Class Actions (Dec. 21, 2017),

[6] It is the author’s suggestion that the facts of these cases offer support for the latter. Section III.B. offers commentary on this aspect of the glucosamine lawsuits.

[7] Procedurally, these cases are as complex and interesting as the ruling itself. Ultimately, according to the Fourth Circuit’s order, there were five GNC plaintiffs (Howard, Toback, Lerma, Calvert, and Gaatz) and three Rite Aid plaintiffs (Flowers, George, and Gross). GNC, 789 F.3d at 509 n.1. The underlying lawsuits were consolidated in the U.S. District Court for the District of Maryland. The consolidated cases are docketed as: Howard v. GNC Corp., No. 14-cv-00002 (D. Md. filed Jan. 3, 2014); Toback v. GNC Holdings, Inc., No. 14-cv-00122 (D. Md. filed Jan. 14, 2014); Lerma v. GNC Corp., No. 14-cv-00120 (D. Md. filed Apr. 18, 2013); Calvert v. GNC Corp., No. 14-cv-00123 (D. Md. filed Jan. 16, 2014). The Rite Aid product case was Flowers v. Rite Aid, No. 14-cv-00465 (D. Md. filed Feb. 18, 2014). Flowers was effectively a lawsuit against GNC, as the products at issue were manufactured for Rite Aid by GNC, and GNC was contractually obligated to indemnify Rite Aid for the claims at issue in the litigation. Consolidated Amended Complaint at ∂ 2, In re GNC Corp. Triflex Prods. Mktg. & Sales Practices Litig. (No. II), No. 14-2491, 2014 U.S. Dist. LEXIS 84184 (D. Md. June 20, 2014), ECF No.20. To the best of the author’s knowledge, GNC did not contest this assertion. There are multiple related lawsuits that were resolved separately, such as Galvin v. GNC, No. 14-cv-00810 (D. Md. filed Mar. 21, 2014). Although Brown v. GNC Corp., No. 13-05890 (N.D. Cal. filed Dec. 19, 2013), was also transferred to the district court by the Multidistrict Litigation Panel, the Consolidated Amended Complaint does not include Yvonne Brown (the plaintiff in that action) among the named plaintiffs. GNC, 789 F.3d at 509 n.2. Galvin was voluntarily dismissed in October 2015, and therefore did not proceed with the other consolidated cases.

[8] Warren Teichner & Megan Lesko, Cashing in on the booming market for dietary supplements, McKinsey & Company Marketing & Sales Insights (Dec. 2013),

[9] Council for Responsible Nutrition, The Dietary Supplement Consumer: 2015 CRN Consumer Survey on Dietary Supplements, CRN USA (2015),

[10] Country Report: Vitamins in the US, Euromonitor International (Nov. 2017), This insight into the brutally competitive landscape offers some explanation for GNC’s insistence on keeping its joint health tag line. Marketing and advertising claims, such as “promotes joint health,” are valued shortcuts for consumers as they navigate supplement shelves. The alternative is a sea of supplements that are labeled and identified solely by their active ingredient(s), like a bottle that simply reads “Vitamin D” or “Glucosamine-Chondroitin” on the front, without further description of health benefits. A consumer who is unfamiliar with glucosamine-chondroitin may not buy the product, were it not for the “promotes joint health” byline. The explanation provides quick information and incentive to buy.

[11] See National Institutes of Health, supra note 2.

[12] FDA 101: Dietary Supplements, U.S. Food & Drug Admin. (Nov. 6, 2017),

[13] Id.

[14] See National Institutes of Health, supra 2.

[15] For more information about the FDA and FTC’s roles in policing supplement labeling and advertising, see Section I.B., infra.

[16] Kenneth A. Plevan, Recent Trends in the Use of Surveys in Advertising and Consumer Deception Disputes, 15 Chi.-Kent J. Intell. Prop. 49, 61 (2016) (commenting on the “recent explosion of consumer deception lawsuits brought as putative class actions, filed by private plaintiffs under state consumer protection laws . . .”); see also Theodore V.H. Mayer, Recent Developments and Current Trends in United States Class Action Law, 826 PLI/Lit 313, 325 (May 24, 2010) (citing Federal Judicial Center, The Impact of the Class Action Fairness Act of 2005 on the Federal Courts 4 (4th interim report, 2008)) (“Among the most remarkable trends from the period between 2001 and 2007 [was] . . . the continuing growth of consumer-protection or consumer-fraud class actions, which increased by more than 150 percent and now account for 20 percent of all federal class actions.”) Private plaintiffs have increasingly availed themselves of consumer protection statutes. According to one study of over 17,000 reported federal district and state appellate decisions, “[b]etween 2000 and 2007 the number of [consumer protection act] decisions reported in federal district and state appellate courts increased by 119%. This large increase in CPA litigation far exceeds increases in tort litigation as well as overall litigation during the same period.” See Searle Civil Justice Institute, State Consumer Protection Acts: An Empirical Investigation of Private Litigation Preliminary Report (Dec. 2009). An additional concern is the “double litigation” frontier: “There is nothing to prevent a private litigant from filing suit against a consumer product advertiser or manufacturer after a federal regulatory agency, such as the Federal Trade Commission, takes action against the same company and obtains full monetary redress for consumers.” Dana Rosenfeld & Daniel Blynn, The “Prior Substantiation” Doctrine: An Important Check On the Piggyback Class Action, 26 Antitrust 1, 68 (Fall 2011). Rosenfeld and Blynn state that there is an emerging trend of plaintiffs filing class action complaints that are “virtually identical to or rely heavily upon” FTC complaints or FDA warning letters. Id.

[17] Brown v. GNC Corp. (In re GNC Corp.), 789 F.3d 505, 515 (4th Cir. 2015).

[18] Plaintiffs may seek redress from the Federal Trade Commission under the FTC Act, 15 U.S.C. §§ 41-58. Barton Beebe, Trademark Law: An Open-Source Casebook, Part IV, 2 (Jul. 20, 2017),

[19] Lanham Act § 43(a)(1)(B) is codified at 15 U.S.C. § 1125(a)(1)(B).

[20] The state consumer deception statutes are also known as “little” or “baby” FTC Acts. See Beebe, supra note 18.

[21] Denise E. DeLorme, Jisu Huh, Leonard N. Reid & Soontae An, Dietary supplement advertising in the US: A review and research agenda, 31 Int’l J. of Advert. 547, 555 (2012).

[22] See Beebe, supra note 18. See also Joshua M. Dalton & Jared A. Craft, What You Should Know About NAD False Advertising Claims, Law360 (Jan. 4, 2013),

[23] Note that the FDA and FTC are agencies of different stature. Unlike the FTC, the FDA is not an independent entity of the U.S. government. The FDA is nestled under the Department of Health and Human Services and funded by the Department of Agriculture.

[24] Advertising Dietary Supplements, Consumer Healthcare Products Association, (last visited Mar. 23, 2017).

[25] Rebecca Tushnet, Advertising and Marketing Law 1289 (3rd ed. 2014).

[26] U.S. Food & Drug Admin. Guidance for Industry: A Food Labeling Guide, (revised Jan. 2013),

[27] Mark Nichter & J.J. Thompson, For my wellness, not just my illness: North Americans’ use of dietary supplements, 30 Culture, Med. & Psychiatry 175, 176 (2006). Unlike drugs, supplements can be readily purchased without a prescription in a wide variety of brick-and-mortar stores and online.

[28] 15 U.S.C. § 45 (2012).

[29] Pfizer, Inc., 81 F.T.C. 23 (1972).

[30] Lesley Fair, Federal Trade Commission Advertising Enforcement, Fed. Trade Comm’n (revised Mar. 1, 2008),; see, e.g., Brake Guard Products, Inc., 125 F.T.C. 138 (1998); ABS Tech Sciences, Inc., 126 F.T.C. 229 (1998).

[31] See, e.g., Removatron International Corp., 111 F.T.C. 206 (1988), aff’d 884 F.2d 1489 (1st Cir. 1989) (“adequate and well-controlled…clinical testing” is required to substantiate claims for hair removal product); Schering Corp., 118 F.T.C. 1030 (1994) (consent order) (tests and studies relied upon as “reasonable basis” must employ appropriate methodology and address specific claims made in the advertisement).

[32] Dietary Supplement Health and Education Act of 1994, Pub. L. No. 103-417, 108 Stat. 4325 (codified in scattered sections of 21 U.S.C.).

[33] Questions and Answers on Dietary Supplements, U.S. Food & Drug Admin., (last updated Nov. 29, 2017).

[34] In its current form, section 43(a)(1)(B) states: (1) Any person who, on or in connection with any goods or services, or any container for goods, uses in commerce any word, term, name, symbol, or device, or any combination thereof, or any false designation of origin, false or misleading description of fact, or false or misleading representation of fact, which — . . . (B) in commercial advertising or promotion, misrepresents the nature, characteristics, qualities, or geographic origin of his or her or another person’s goods, services, or commercial activities, shall be liable in a civil action by any person who believes that he or she is or is likely to be damaged by such act. 15 U.S.C. § 1125(a)(1)(B) (2018).

[35] See Beebe, supra note 18, at Part IV, 2.

[36] 134 S. Ct. 1377, 1395 (2014).

[37] Id. at 1389; see also J. Thomas McCarthy, McCarthy on Trademarks and Unfair Competition § 27.30 (5th ed. 2017).

[38] See Brown v. GNC Corp. (In re GNC Corp.), 789 F.3d 505, 514 (4th Cir. 2015) (noting that, although consumers cannot invoke the protections of the Lanham Act, the “considerable body of federal common law construing the Act is instructive in construing the state laws at issue here”).

[39] Puffery is a safe harbor for advertisers who proffer exaggerated and unsupported (or, perhaps, unsupportable) claims. It is a non-actionable carve-out from the false advertising provisions of the Lanham Act, since the statements are typically so exaggerated that no reasonable consumer could be deemed to rely on them. To fall out of the test for false advertising, statements of puffery are not considered to be statements of fact; consequently, a plaintiff would fail to meet the first element of the test. Courts have different definitions of puffery. In Time Warner Cable, Inc. v. DIRECTV, Inc., the Second Circuit described one party’s internet ads as “inaccurate descriptions” of the television service, but “so grossly distorted and exaggerated that no reasonable buyer would take them to be accurate depictions.” 497 F.3d 144, 159 (2d Cir. 2007). In United Industries Corp. v. Clorox Co., “[p]uffery is ‘exaggerated advertising, blustering, and boasting upon which no reasonable buyer would rely and is not actionable under § 43(a).’” 140 F.3d 1175, 1180 (8th Cir.1998) (quoting Southland Sod Farms v. Stover Seed Co., 108 F.3d 1134, 1145 (9th Cir. 1997)). In other iterations, the puffery doctrine has protected purveyors of “The Best Beer in America,” In re Bos. Beer Co., 198 F.3d 1370 (Fed. Cir. 1999), “The Most Advanced Home Gaming System in the Universe,” Atari Corp. v. 3D0 Co., No. C 94-20298 RMW (EAI), 1994 U.S. Dist. LEXIS 8677 (N.D. Cal. May 16, 1994), and “Better Ingredients. Better Pizza,” Pizza Hut, Inc. v. Papa John’s Int’l, Inc., 227 F.3d 489 (5th Cir. 2005).

[40] Under National Basketball Ass’n v. Motorola, Inc., the test for materiality is whether the statement “misrepresent[s] an inherent quality or characteristic of a product.” 105 F.3d 841, 855 (2d Cir. 1997).

[41] Schick Mfg., Inc. v. Gillette Co., 372 F.Supp.2d 273, 276 (D. Conn. 2005).

[42] See Tushnet, supra note 25, at 259; see also Coca-Cola Co. v. Tropicana Prods., Inc., 690 F.2d 312, 317 (2d Cir. 1982) (“[T]he Court may grant relief without reference to the advertisement’s [actual] impact on the buying public.”), abrogated on other grounds by Fed. R. Civ. P. 52(a)(6), as recognized in Johnson & Johnson v. GAC Int’l, Inc., 862 F.2d 975, 979 (2d Cir. 1988).

[43] Consumer surveys are not valid supporting evidence in a falsity case, discussed further in Section I.B.2. The GNC case fits neatly within the question of literally false advertising, as the TriFlex ads and packaging presented explicit, unambiguous statements, and no evidence of consumer deception was presented. See generally GNC, 789 F.3d 505.

[44] See generally Coca-Cola, 690 F.2d 312.

[45] See Time Warner Cable, Inc. v. DIRECTV, Inc., 497 F.3d 144, 158 (2d Cir. 2007).

[46] Id. at 152.

[47] See id. at 158. Note that the Second Circuit ruled on numerous commercials and Internet ads in this case, and some of the district court’s opinion was reversed.

[48] Where an advertisement is literally true but misleading, the advertisement “has left an impression on the listener that conflicts with reality[;]” with proof of consumer confusion, the representations are considered misleading. See id. at 153.

[49] 661 F.2d 272 (2d Cir. 1981). See also 2-7 Gilson on Trademarks § 7.02 (2017).

[50] Carolyn L. Carter, Consumer Protection in the States: A 50-State Report on UDAP Statutes, 5 (Feb. 2009),

[51] 2-7 Gilson on Trademarks § 7.02 (2017).

[52] Carter, supra note 50, at 5.

[53] See Kenneth A. Plevan & Angela Colt, Consumer Surveys: Certification, Bloomberg BNA Prod. Safety & Liab. Rep. 4 (Sept. 12, 2016).

[54] See Consolidated Amended Complaint at ∂ 26-35, In re GNC Corp. Triflex Prods. Mktg. & Sales Practices Litig. (No. II), No. 14-2491, 2014 U.S. Dist. LEXIS 84184 (D. Md. June 20, 2014) ECF No.20. For reference, the four GNC products at issue, according to the Plaintiffs, were TriFlex, TriFlex Sport, TriFlex Fast-Acting, and TriFlex Complete Vitapak. The six Rite Aid products at issue are Rite Aid Glucosamine/Chondroitin, Rite Aid Natural Glucosamine/Chondroitin, Rite Aid Glucosamine Chondroitin Advanced Complex, Rite Aid Glucosamine Chondroitin, Triple Strength + MSM, Rite Aid Glucosamine Chondroitin + MSM, and Rite Aid Glucosamine Chondroitin Advanced Complex with HA. The court does not distinguish between the GNC and Rite Aid brands, nor does it distinguish between the individual products at issue.

[55] See Consolidated Amended Complaint at ∂ 26-37, In re GNC Corp. Triflex Prods. Mktg. & Sales Practices Litig. (No. II), No. 14-2491, 2014 U.S. Dist. LEXIS 84184 (D. Md. Jun. 20, 2014) ECF No.20; see also In re GNC Corp. Triflex Prods. Mktg. & Sales Practices Litig. (No. II), No. 14-2491, 2014 U.S. Dist. LEXIS 84184 (D. Md. June 20, 2014). The claims on the labels, as presented by Plaintiffs and GNC, are as follows. The GNC TriFlex Dietary Supplement label contains “Maximum strength now with hyaluronic acid” and “Promotes joint mobility and flexibility.” The GNC TriFlex Sport label contains: “Protects joints from wear and tear of exercise,” “Maximum strength joint comfort for active individuals,” and “Clinical strength for daily long-term use.” TriFlex Fast-Acting label contains: “Now with a joint cushioning blend including hyaluronic acid and vitamin C,” “Maximum strength, fast-acting support – works in days,” and “Clinical strength for daily long-term use.” Lastly, the TriFlex Complete Vitapak label contains: “Maximum strength, fast-acting joint comfort – works in days,” “Rebuilds cartilage and lubricates joints,” and “Supports natural anti-inflammatory response.” The Rite Aid Glucosamine/Chondroitin Dietary Supplement label contains “helps rebuild cartilage and lubricate joints.” Each label contains an FDA disclaimer: “This statement has” or “these statements have” “not been evaluated by the Food and Drug Administration. This product is not intended to diagnose, treat, cure, or prevent any disease.” It appears that the only product to include a “here’s proof” claim is TriFlex Fast-Acting. There is a reference to a scientific study that supports GNC’s representations: “Clinically studied doses of glucosamine and chondroitin combined with MSM and a proprietary herbal blend, which is shown to improve joint comfort and function. In a 12-week multi-center, randomized, double blind, placebo controlled study of 60 adults, subjects taking 250 mg/day of the GNC TriFlexTM Fast-Acting Blend showed statistically significant improvements in measures of joint function and joint flexibility within 30 days compared to subjects on placebo.”

[56] See Defendant’s Memorandum in Support of its Motion to Dismiss Class Action Complaint, In re GNC Corp. Triflex Prods. Mktg. & Sales Practices Litig. (No. II), No. 14-2491, 2014 U.S. Dist. LEXIS 84184 (D. Md. June 20, 2014) ECF No.38; see also Rebecca Tushnet, Fourth Circuit Destroys Literal Falsity, Rebecca Tushnet’s 43(B)log (June 30, 2015), The GNC study was not published or otherwise publicly available, and there is currently no law requiring such disclosure.

[57] Some of the state consumer protection laws at issue include the California False Advertising Law, § 17500, et seq. (“FAL”), the California Unfair Competition Law, Cal. Bus. & Prof. Code § 17200, et seq. (“UCL”), the Florida Deceptive and Unfair Trade Practices Act, Fla. Stat. §§ 501.201, et seq. (“FDUTPA”), Illinois Consumer Fraud and Deceptive Business Practice Act, 815 Ill. Comp. Stat. 502/1, et seq. (“ICFA”), the New York Consumer Protection From Deceptive Acts and Practices Law, N.Y. Gen. Bus. Law § 349, et seq. (“NYGBL”), Ohio Rev. Code Ann. § 1302.26, the New Jersey Consumer Fraud Act, N.J.S.A. 56:8-1, et seq. (“NJCFA”), Pennsylvania Unfair Trade Practices and Consumer Protection Law, 73 Pa. Stat. Ann. §§ 201-1, et seq. (“PUTPCPL”).

[58] In re GNC Corp. TriFlex Prods. Mktg. & Sales Practices Litig. (No. II), No. 14-120, 2014 U.S. Dist. LEXIS 84184 at *1 (D. Md. June 20, 2014).

[59] Consolidated Amended Complaint at ∂ 38, In re GNC Corp. Triflex Prods. Mktg. & Sales Practices Litig. (No. II), No. 14-2491, 2014 U.S. Dist. LEXIS 84184 (D. Md. June 20, 2014) ECF No.20.

[60] Id.

[61] The studies can be found in the Consolidated Amended Complaint at ∂∂ 39-48, In re GNC Corp. Triflex Prods. Mktg. & Sales Practices Litig. (No. II), No. 14-2491, 2014 U.S. Dist. LEXIS 84184 (D. Md. June 20, 2014) ECF No.20.

[62] Consolidated Amended Complaint at ∂ 32, In re GNC Corp. Triflex Prods. Mktg. & Sales Practices Litig. (No. II), No. 14-2491, 2014 U.S. Dist. LEXIS 84184 (D. Md. June 20, 2014) ECF No.20.

[63] Since this case is considered procedurally significant, it is important to say a word about the legal standard on a motion to dismiss. A Rule 12(b)(6) motion to dismiss challenges the legal sufficiency of a complaint. Fed. R. Civ. P. 12(b)(6); see also Presley v. City of Charlottesville, 464 F.3d 480, 483 (4th Cir. 2006). The court has drawn a line between the “mere possibility” and “plausibility” that a defendant has acted unlawfully; plaintiff must meet the latter to survive a motion to dismiss. Ashcroft v. Iqbal, 556 U.S. 662, 678 (2009). The issue is whether plaintiff has stated enough factual matter, rising above the speculative level, to warrant a claim for relief. To defeat a 12(b)(6) motion, a complaint’s factual allegations must be sufficient to “raise a right to relief above the speculative level.” Bell Atl. Corp. v. Twombly, 550 U.S. 544, 555 (2007). The complaint must contain “sufficient factual matter, accepted as true, to state a claim for relief that is plausible on its face.” Ashcroft, 556 U.S. at 678. The “mere possibility of misconduct” is insufficient to avoid dismissal. Id. at 679. Likewise, “[l]abels and conclusions” and “naked assertion devoid of further factual enhancement” will fail to show that plaintiff is entitled to relief. Id. at 678. Ultimately, Plaintiffs’ bare assertions were insufficient to establish that the representations are false.

[64] Notably, similar and even predicate cases challenging the efficacy of glucosamine-chondroitin had been dismissed on similar grounds. See, e.g., Toback v. GNC Holdings, Inc., No. 13-80526, 2013 U.S. Dist. LEXIS 131135, at *16 (S.D. Fla. Sept. 13, 2013) (“Plaintiff’s allegations regarding the inefficacy of glucosamine and chondroitin simply fail to address the efficacy of the TriFlex Vitapak’s multifarious composition in promoting joint health . . .”); Eckler v. Wal-Mart Stores, Inc., No. 12-727, 2012 WL 5382218, at *6 (S.D. Cal. 2012) (plaintiff’s studies did not assess the “overall formulation that’s behind the representations at issue,” and so “the Court would be left with no facts from which to infer that [defendant] is liable for false advertising.”).

[65] Defendants’ Memorandum in Support of their Motion to Dismiss Consolidated Amended Complaint at 5, In re GNC Corp. Triflex Prods. Mktg. & Sales Practices Litig. (No. II), No. 14-2491, 2014 U.S. Dist. LEXIS 84184 (D. Md. June 20, 2014), ECF No.25.

[66] The Consolidated Amended Complaint repeatedly described the TriFlex advertising as “false, misleading, and reasonably likely to deceive the public.” Consolidated Amended Complaint, In re GNC Corp. Triflex Prods. Mktg. & Sales Practices Litig. (No. II), No. 14-2491, 2014 U.S. Dist. LEXIS 84184 (D. Md. June 20, 2014), ECF No.20. However, to plead that an advertisement is misleading, the allegation must be supported by evidence of consumer confusion. Since Plaintiffs did not provide any evidence of consumer confusion, the Court appropriately limited its analysis to a claim of literal falsity.

[67] Memorandum at 7, In re GNC Corp. Triflex Prods. Mktg. & Sales Practices Litig. (No. II), No. 14-2491, 2014 U.S. Dist. LEXIS 84184 (D. Md. June 20, 2014), ECF No. 38.

[68] Id.

[69] Id. (emphasis added).

[70] Id. at n.2.

[71] See id.; see also Memorandum at 4 n.4, In re GNC Corp. Triflex Prods. Mktg. & Sales Practices Litig. (No. II), No. 14-2491, 2014 U.S. Dist. LEXIS 84184 (D. Md. June 20, 2014), ECF No. 51. The significance of this carve-out for advertising cases is discussed further in Part III, supra.

[72] Memorandum at 7, In re GNC Corp. Triflex Prods. Mktg. & Sales Practices Litig. (No. II), No. 14-2491, 2014 U.S. Dist. LEXIS 84184 (D. Md. June 20, 2014), ECF No. 38. An obvious issue, then, becomes whether Plaintiffs could so allege without appropriate discovery. Judge Motz addressed this question in his second Memorandum Decision, denying the motion for reconsideration: “[I]f plaintiffs can specify discovery requests that would aid them in alleging the above facts, they should file a motion setting forth the discovery that they request. Presumably, however, if plaintiffs’ experts are of the view that no reasonable expert would reach the conclusion reached by the expert upon whom defendant relies, they are already, by virtue of their asserted expertise, in possession of the relevant factual information.” Memorandum at 5, In re GNC Corp. Triflex Prods. Mktg. & Sales Practices Litig. (No. II), No. 14-2491, 2014 U.S. Dist. LEXIS 84184 (D. Md. June 20, 2014), ECF No. 51.

[73] Memorandum at 4, In re GNC Corp. Triflex Prods. Mktg. & Sales Practices Litig. (No. II), No. 14-2491, 2014 U.S. Dist. LEXIS 84184 (D. Md. June 20, 2014), ECF No. 51.

[74] Memorandum at 8, In re GNC Corp. Triflex Prods. Mktg. & Sales Practices Litig. (No. II), No. 14-2491, 2014 U.S. Dist. LEXIS 84184 (D. Md. June 20, 2014), ECF No. 38.

[75] Exhibit 1 to Plaintiffs’ Memorandum of Law in Support of Motion to Correct Mistake of Law Pursuant to F.R.C.P. 60, In re GNC Corp. Triflex Prods. Mktg. & Sales Practices Litig. (No. II), No. 14-2491, 2014 U.S. Dist. LEXIS 84184 (D. Md. June 20, 2014), ECF No. 44-1.

[76] See Plaintiffs’ Motion to Correct Mistake of Law Pursuant to F.R.C.P. Rule 60, In re GNC Corp. Triflex Prods. Mktg. & Sales Practices Litig. (No. II), No. 14-2491, 2014 U.S. Dist. LEXIS 84184 (D. Md. June 20, 2014), ECF No. 43; Plaintiffs’ Memorandum of Law in Support of Motion to Correct Mistake of Law Pursuant to F.R.C.P. 60, In re GNC Corp. Triflex Prods. Mktg. & Sales Practices Litig. (No. II), No. 14-2491, 2014 U.S. Dist. LEXIS 84184 (D. Md. June 20, 2014), ECF No. 44.

[77] Memorandum at 4, In re GNC Corp. Triflex Prods. Mktg. & Sales Practices Litig. (No. II), No. 14-2491, 2014 U.S. Dist. LEXIS 84184 (D. Md. June 20, 2014), ECF No. 51; see also Brown v. GNC Corp. (In re GNC Corp.), 789 F.3d 505, 512 (4th Cir. 2015).

[78] There are some important differences between the District Court’s Memorandum Decision and the Fourth Circuit’s Order. First, the Court included alternative grounds for affirming the District Court, noting that Plaintiffs “failed to allege that all of the purportedly active ingredients in each product are ineffective at promoting joint comfort, health, and flexibility.” Second, the Fourth Circuit disagreed with Judge Motz that specific formulations of the GNC and Rite Aid products needed to be tested to assess the truth of the labels’ representations; instead, at the motion to dismiss stage, the scientific studies in the CAC could render the claim facially plausible. Lastly, the court found that the factual dispute regarding the effectiveness of glucosamine-chondroitin for non-arthritic users was not appropriate for resolution on a motion to dismiss. The Fourth Circuit declined to adopt the latter grounds for affirming the District Court’s Order. See Brown v. GNC Corp. (In re GNC Corp.), 789 F.3d 505 (4th Cir. 2015).

[79] See GNC, 789 F.3d at 516.

[80] Id.

[81] Id. at 514 n.7.

[82] See generally id. at 515-16.

[83] Id. at 515.

[84] Id.

[85] Id.

[86] See Brief of Law Professors as Amici Curiae in Support of Plaintiffs-Appellants’ Petition for Rehearing and for Rehearing En Banc, and in the Alternative, for Modification of Opinion and Judgment, Brown v. GNC Corp. (In re GNC Corp.), 789 F.3d 505 (4th Cir. 2015) (No. 14-1724) ECF No.45.

[87] Id. at 1.

[88] Professor Rebecca Tushnet, whose advertising textbook is extensively cited in this Note, appears to have been the lead academic on the Amicus Brief. It is co-signed with Brian Wolfman, then a Visiting Professor at Stanford Law School. Other Amici include Mark Lemley of Stanford Law School, Jessica Litman of the University of Michigan Law School, and Barton Beebe of New York University School of Law, who supervised this Note’s completion. Notably, Professor Eric Goldman of Santa Clara University School of Law is not a signatory to the Amicus Brief, although he co-authored the seminal textbook on advertising law, Advertising & Marketing Law, with Professor Tushnet.

[89] Brief of Law Professors as Amici Curiae in Support of Plaintiffs-Appellants’ Petition for Rehearing and for Rehearing En Banc, and in the Alternative, for Modification of Opinion and Judgment at 1-2, Brown v. GNC Corp. (In re GNC Corp.), 789 F.3d 505 (4th Cir. 2015) (No. 14-1724) ECF No.45. The brief is inexplicably wrought with references to consumer messaging in literal falsity cases. As Amici later state, “To determine if an ad makes a false claim, a court must determine what message consumers will perceive . . . .”

[90] Perhaps there is some suggestion that Plaintiffs’ should have argued that the joint health claims were misleading, rather than literally false. The argument would appear to be that glucosamine-chondroitin is particularly attractive to osteoarthritis patients, who seek to alleviate the pain of their condition. Plaintiffs offered numerous studies that allegedly prove that glucosamine-chondroitin is not beneficial to osteoarthritis sufferers. On that basis, and with an allegation of consumer confusion, the claims could have been deemed misleading to a segment of consumers. However, a claim of misleading advertising would, too, seem to fail, as the District Court noted that “the TriFlex labels expressly disclaim any ability to ‘diagnose, treat, cure, or prevent any disease.” (citations omitted). In re GNC Corp. TriFlex Prods. Mktg. & Sales Practices Litig. (No. II), No. 14-2491, 2014 U.S. Dist. LEXIS 84184 at *3-4 (D. Md. June 20, 2014).

[91] Brief of Law Professors as Amici Curiae in Support of Plaintiffs-Appellants’ Petition for Rehearing and for Rehearing En Banc, and in the Alternative, for Modification of Opinion and Judgment at 5, Brown v. GNC Corp. (In re GNC Corp.), 789 F.3d 505 (4th Cir. 2015) (No. 14-1724) ECF No.45.

[92] Id at 2.

[93] In the discussion that follows that quote, the C.B. Fleet Court states: “Fleet challenges the district court’s determinations of no literal falsity on the grounds that the court imposed upon Fleet a wrong– overly stringent– ‘burden of proof’ on the issue, and, relatedly, that both of the ultimate fact findings were clearly erroneous[;]” “[W]e think that whether an advertising claim implicitly, though not expressly, asserts that it is test-validated must be considered a question of fact whose resolution is subject to clearly erroneous review[;]” “Our standard of review of district court fact findings is greatly deferential under controlling authority[]” (emphasis added). C.B. Fleet Co. v. SmithKline Beecham Consumer Healthcare, L.P., 131 F.3d 430, 434-436 (4th Cir. 1997).

[94] L&F Prods. v. Procter & Gamble Co., 45 F.3d 709, 712 (2d Cir. 1995).

[95] Brief of Law Professors as Amici Curiae in Support of Plaintiffs-Appellants’ Petition for Rehearing and for Rehearing En Banc, and in the Alternative, for Modification of Opinion and Judgment at 3, Brown v. GNC Corp. (In re GNC Corp.), 789 F.3d 505 (4th Cir. 2015) (No. 14-1724) ECF No.45.

[96] Plaintiffs’ Reply Memorandum of Law in Further Support of Motion to Correct Mistake of Law Pursuant to F.R.C.P. 60 at 4, In re GNC Corp. TriFlex Prods. Mktg. & Sales Practices Litig. (No. II), No. 14-2491, 2014 U.S. Dist. LEXIS 84184 (D. Md. June 20, 2014) ECF No.50.

[97] Defendant’s Memorandum in Support of its Motion to Dismiss Class Action Complaint at n.2, In re GNC Corp. Triflex Prods. Mktg. & Sales Practices Litig. (No. II), No. 14-2491, 2014 U.S. Dist. LEXIS 84184 (D. Md. June 20, 2014) ECF No.38.

[98] See generally Brief of Law Professors as Amici Curiae in Support of Plaintiffs-Appellants’ Petition for Rehearing and for Rehearing En Banc, and in the Alternative, for Modification of Opinion and Judgment, Brown v. GNC Corp. (In re GNC Corp.), 789 F.3d 505 (4th Cir. 2015) (No. 14-1724) ECF No.45; Olamide Orebamjo, Comment, Brown v. GNC Corp.: The Fourth Circuit’s New Standard for Literal Falsity, 12 J. Bus. & Tech. L. Proxy 1 (2017).

[99] Pfizer, 81 F.T.C. at 29 (1972).

[100] John P. Swann, The history of efforts to regulate dietary supplements in the USA, 8 Drug Testing & Analysis 271, 275 (2016).

[101] Brown v. GNC Corp. (In re GNC Corp.), 789 F.3d 505, 514 (4th Cir. 2015) (“In construing the diverse state statutes at issue here, we apply this broadly shared understanding of the difference between false and misleading representations.”).

[102] See Memorandum at 4 n.4, In re GNC Corp. Triflex Prods. Mktg. & Sales Practices Litig. (No. II), No. 14-2491, 2014 U.S. Dist. LEXIS 84184 (D. Md. June 20, 2014), ECF No. 51.

[103] Eckler v. Wal-Mart Stores, Inc., No. 12-727, 2012 WL 5382218, at *8 n.2 (S.D. Cal. 2012.

[104] Memorandum at 4 n.4, In re GNC Corp. Triflex Prods. Mktg. & Sales Practices Litig. (No. II), No. 14-2491, 2014 U.S. Dist. LEXIS 84184 (D. Md. June 20, 2014), ECF No. 51.

[105] Plaintiffs’ Memorandum of Law in Support of Motion to Correct Mistake of Law Pursuant to F.R.C.P. 60, In re GNC Corp. Triflex Prods. Mktg. & Sales Practices Litig. (No. II), No. 14-2491, 2014 U.S. Dist. LEXIS 84184 (D. Md. June 20, 2014), ECF No. 44.

Think Big! The Need for Patent Rights in the Era of Big Data and Machine Learning

Think Big! The Need for Patent Rights in the Era of Big Data and Machine Learning
By: Hyunjong Ryan Jin*   Download a PDF version of this article here  


With AlphaGo’s triumph over the 9-dan Go professional Lee Sedol in March 2016, Google’s DeepMind team conquered the last remaining milestone in board game artificial intelligence.[1] Just nineteen years after IBM Deep Blue’s victory over the Russian chess grandmaster Garry Kasparov,[2] Google’s success exceeded expert predictions by decades.[3] AlphaGo demonstrated how machine learning algorithms could enable processing of vast amounts of data. Played out on a 19 by 19 grid, the number of possible configurations on a Go board is astronomical.[4] With near-infinite number of potential moves, conventional brute-force comparison of all possible outcomes is not feasible.[5] To compete with professional level human Go players, the gaming artificial intelligence requires a more sophisticated approach than the algorithms employed for chess — machine learning. The underlying science and implementation of machine learning was described in a Nature article two months prior to AlphaGo’s match with Lee. In the article, the Google team described how a method called “deep neural networks” decides between the insurmountable number of possible moves in Go.[6] The AlphaGo model was built by reinforcement learning from a database consisting of over thirty million moves of world-class Go players.[7] This allowed the algorithm to optimize the search space of potential moves, therefore reducing the required calculations to determine the next move.[8] In other words, the algorithm mimics human intuition based on the “experience” it gained from the database “fed” into the algorithm, which drastically increases computational efficiency by eliminating moves not worth subsequent consideration. This allows the algorithm to devote computational resources towards the outcomes of “worthwhile” moves. The advent of such powerful analytical tools, capable of mimicking human intuition alongside massive computation power, opens endless possibilities—early stage cancer detection[9], accurate weather forecasting,[10] prediction of corporate bankruptcies,[11] natural event detection,[12] and even prediction of elections.[13] For information technology (“IT”) corporations, investment in such technology is no longer an option, but a necessity. The question that this Note addresses is whether the current state of intellectual property law is adequate to harness the societal benefits that we hope to enjoy through the advances in machine learning. In particular, are patents necessary in the age of big data? And if they are, how should we apply patent protection in the field of big data and machine learning? Part I of this Note examines the need for intellectual property rights in machine learning and identifies the methods by which such protection may be achieved. The differences between trade secret, copyright and patent protection in software are discussed, followed by the scope of protection offered by each means. This background provides the basis to discuss the effectiveness of each method in the context of machine learning and big data innovations. Part II discusses the basics of the underlying engineering principle of machine learning and demonstrates how the different types of intellectual property protection may apply. Innovators may protect their contributions in machine learning by defending three areas—(1) the vast amount of data required to train the machine learning algorithm, (2) innovations in the algorithms itself including advanced mathematical models and faster computational methods, and (3) the resulting machine learning model and the output data sets. Likewise, there are three distinct methods of protecting these intellectual properties: patents, copyright, and secrecy.[14] This Note discusses the effectiveness of each method of intellectual property protection with three principles of machine learning innovation in mind: facilitating data sharing, avoiding barriers to entry from data network effects, and providing incentives to address the key technological challenges of machine learning. This Notes proposes that patents on computational methods adequately balance the concern of patent monopoly and promoting innovation, hence should be the primary means of intellectual property protection in machine learning. Part III then visits the legal doctrine of patentable subject matter starting with the United States Supreme Court’s Alice decision. While Alice imposed a high bar for software patents, the post-Alice Federal Circuit decisions such as Enfish, Bascom, and McRO suggest that certain types of software inventions are still patentable. Specifically, this section will discuss the modern framework pertinent to subject matter analysis: (1) inventions that are directed to improvements of computer functionality rather than an abstract idea, (2) inventions that contain an inventive concept, and (3) inventions that do not improperly preempt other solutions. The Note will apply this framework to innovations in machine learning. The Note proposes that patents for computational methods balance the need for intellectual property protection while permitting data sharing, paving the pathway for promoting innovation in machine learning. The Note further argues that machine learning algorithms are within patentable subject matter under 35 U.S.C. §101.

I. Need for Intellectual Property Rights in Machine Learning

He who receives an idea from me, receives instruction himself without lessening mine; as he who lights his taper at mine, receives light without darkening me.” – Thomas Jefferson “I’m going to destroy Android, because it’s a stolen product. I’m willing to go thermonuclear war on this. They are scared to death, because they know they are guilty.” – Steve Jobs The two quotes above demonstrate the conflicting views on protecting intangible ideas with intellectual property law. Thomas Jefferson implied that the free circulation of inventive ideas and thoughts would not dampen the progress of innovation nor disadvantage innovators. On the other hand, Steve Jobs exhibited fury over the similarity between the iOS and the Android OS. Why? Was it because his company was worse off due to the similarity between the two products? Would Apple have refrained from inventing the iPhone had it known others would enter the smartphone market? This section discusses the motives behind the grant of intellectual property rights and whether such protection should be extended to machine learning innovations. Basics of patent law, copyright law, and trade secret are introduced to provide the analytical tools for subsequent discussion on which type of intellectual property protection best promotes the socially-beneficial effects of machine learning.

A. Do We Need Intellectual Property Rights for Machine Learning?

The primary objectives of intellectual property rights are to encourage innovation and to provide the public with the benefits of those innovations.[15] In the context of machine learning, it is not clear whether we need any additional incentives to promote participation in this field. Machine learning is already a “hot field,” with countless actors in industry and academia in active pursuit to keep pace.[16] Hence investment incentivizing may not be a valid justification for granting intellectual property rights in machine learning. Rather, such protection is crucial to promote competition and enhance public benefits. The quality of inferences that may be drawn from a given data set increases exponentially as the aggregation diversifies, which is why cross-industry data aggregation will greatly enhance the societal impact of machine learning.[17] Companies will need to identify new data access points outside of their own fields to gain access to other data sets to further diversity their data. Yet the incentive structures of behemoth corporations may not be well-suited to identify and grow niche markets.[18] It would be up to the smaller, specialized entities to find the gaps that the larger corporations overlooked and provide specialized services addressing the needs of that market. Protective measures that assist newcomers to compete against resource-rich corporations may provide the essential tools for startups to enter such markets. Sufficient intellectual property protection may serve as leverage that startups may use to gain access to data sets in the hands of the Googles and Apples of the world, thus broadening the range of social benefits from machine learning.

B. The Basics of Patent Law

To promote the progress of science and useful arts, by securing for limited times to authors and inventors the exclusive right to their respective writings and discoveries” – United States Constitution, Article I, § 8 The United States Constitution explicitly authorizes Congress to promote useful arts by granting inventors the exclusive rights of their discoveries. Such constitutional rights stems from two distinct bases — (1) a quid pro quo where the government issues a grant of monopoly in exchange for disclosure to society, and (2) property rights of the inventor. The purpose for such rights is explicitly stated in the Constitution—to promote new inventions. The goal is to prevent second arrivers who have not invested in the creation of the initial invention from producing competing products and services at a lower price, undercutting the innovator whose costs are higher for having invested to create the invention. As an incentive for innovators willing to invest in new, useful arts, the patent system provides the innovator rights to exclude others from practicing the invention. Another purpose of such rights is the concept of “ “mining rights.” Akin to the grant of mining rights to the owner in efforts to suppress aggressive mining, the inventor should have the right to define and develop a given field by excluding other people from the frontiers of that knowledge. Considering the importance of industry standards in modern electronics, such a purpose acknowledges the importance of early stage decisions that may define the trajectory of new technological advances.

C. The Thin Protection on Software Under Copyright Law

The Copyright Act defines a “computer program” as “a set of statements or instructions to be used directly or indirectly in a computer to bring about a certain result.”[19] Though it may be counterintuitive to grant copyright protection for “useful arts” covered by patents, Congress has explicitly mandated copyright protection for software.[20] However, as will be discussed below, copyright protection of software has been significantly limited due to case law. Copyright protects against literal infringement of the text of the program. Source code, code lines that the programmers “author” via computer languages such as C++ and Python, is protected under copyright as literary work.[21] In Apple v. Franklin Corp., the Third Circuit Court of Appeals held that object code, which is the product of compiling the source code, is also considered a literary work.[22] Given that compiled code is a “translation” of the source code, this ruling seems to be an obvious extension of copyright protection. Removing the copyright distinction between source code and object code better reflects the nature of computer languages such as Perl, where the source code is not translated into object code but rather is directly fed into the computer for execution. However, the scope of protection on either type of code is very narrow. The copyright system protects the author against literal copying of code lines. This leaves open the opportunity for competitors to avoid infringement by implementing the same algorithm using different text. Fortunately, in addition to protection against literal copying of code, copyright law may provide some protection of the structure and logical flow of a program. Equivalent to protecting the “plot” of a novel, the Second Circuit Court of Appeals ruled that certain elements of programming structure are considered an expression (copyrightable) rather than idea (not copyrightable), extending copyright protection to non-literal copying.[23] The Computer Associates International v. Altai court applied a three-step test to determine whether a computer program infringes other programs—(1) map levels of abstraction of the program; (2) filter out protectable expression from non-protectable ideas; and (3) compare which parts of the protected expression are also in the infringing program.[24] The merger doctrine is applied to step two of the Altai test to limit what may be protected under copyright law. Under the merger doctrine, code implemented for efficiency reasons is considered as merged with the underlying idea, hence not copyrightable.[25] Since most algorithms are developed and implemented for efficiency concerns, the Altai framework may prevent significant aspects of software algorithms from receiving copyright protection. This means that for algorithms related to computational efficiency, patents may provide significantly more meaningful protection than copyright. The Federal Circuit, in the 2016 case McRO Inc. v. Bandai Namco Games America Inc., ruled that patent claims with “focus on a specific means or method that improves the relevant technology” may still be patentable.[26] Although preemption concerns may impede patentability, exemption of patent right by preemption is narrow compared to that of copyright by the merger doctrine. Scène à faire doctrine establishes yet another limitation on copyright for computer programs. Aspects of the programs that have been dictated by external concerns such as memory limits, industry standards and other requirements are deemed as non-protectable elements.[27] For mobile application software, it is difficult to imagine programs that are not restricted by form factors such as mobile AP computation power, battery concerns, screen size, and RAM limitations. As for machine learning software, the algorithms determine the “worthiness” of computation paths based on conserving computational resources. The external factors that define the very nature and purpose of such machine learning algorithms may exempt them from copyright protection.

D. Comparing Trade Secret and Non-disclosures with Patents

The crucial distinction between trade secret and patent law is secrecy. While patent applicants are required to disclose novel ideas to the public in exchange for a government granted monopoly, trade secret requires owners to keep the information secret. Though trade secret protection prevents outsiders from acquiring the information by improper means, it does not protect the trade secret against independent development or even reverse engineering of the protected information. In trade secret doctrine, the existence of prior disclosed art is only relevant for discerning whether the know-how is generally known, a different and simpler analysis than the issue of novelty in patent law.[28] The United States Supreme Court has specified in Kewanee Oil that all matters may be protected under trade secret law, regardless whether it may or may not be patented.[29] The Kewanee Oil court predicted that inventors would not resort to trade secret when offered a presumptively stronger protection by patent law:
“The possibility that an inventor who believes his invention meets the standards of patentability will sit back, rely on trade secret law, and after one year of use forfeit any right to patent protection, 35 U.S.C. § 102(b), is remote indeed.”[30]
Trade secret is an adequate form of protection for innovators that are concerned with the limits of what may be patentable. The secrecy requirement of trade secret inherently provides protection that may potentially outlive any patent rights, provided a third party does not independently acquire the secret. This coincides with an interesting aspect of machine learning and big data—the need for massive amounts of data. Developers need data to “train” the algorithm, and increase the accuracy of the machine learning models. Companies that have already acquired massive amounts of data may opt to keep their data secret, treating the aggregated data as a trade secret. In addition to the amount of amassed data, companies have all the more reason to keep their data secret if they have access to meaningful, normalized data. Even if a company amasses an enormous amount of data, the data sets may not be compatible with each other. Data gathered from one source may have different reference points or methodologies that are not immediately compatible with data from another source. This raises the concern of “cleaning” massive amounts of data.[31] Such concerns of data compatibility mean that parties with access to a single, homogenous source of high quality data enjoy a significant advantage over parties that need to pull data from multiple sources. However, data secrecy may not be a suitable strategy for companies that are aiming for cross-industry data aggregation. Institutions such as Global Alliance for Genomics and Health are promoting data sharing between research participants. The Chinese e-commerce giant Alibaba announced a data sharing alliance with companies such as Louis Vuitton and Samsung to fight off counterfeit goods.[32] To facilitate the development of technology and to mitigate risks, various companies and research institutions across diverse fields are engaging in joint development efforts and alliances. Seeking protection under trade secret runs against this trend of engaging in effective cross-industry collaboration. Yet there are countervailing arguments that trade secret promotes disclosure by providing legal remedies that can replace the protection of secrets.[33] Parties can sidestep the limitations of trade secrets by sharing proprietary information under the protection of contract law. While data sharing practices may void trade secret protection, the nature of continued accumulation of data and carefully drafted contractual provisions may provide sufficient protection for the data owners.

II. Placing Machine Learning within Intellectual Property Law

Learning is any process by which a system improves performance from experience.” – Herbert Simon, Nobel Prize in Economics 1978. The concept of machine learning relates to computer programs that have the capability to improve performance based on experience, with limited intervention of the programmer.[34] Machine learning models have the capability to automatically adapt and customize for individual users, discover new patterns and correlations from large databases, and automate tasks that require some intelligence by mimicking human intuition.[35] This section dissects the mechanics of machine learning to identify the aspects of machine learning innovations that are at issue as intellectual property.

A. Machine Learning Basics

Machine learning methods are divided into two different approaches—supervised machine learning and unsupervised machine learning. For supervised machine learning, models are typically established by applying “labeled” sets of data to a learning algorithm. Labeled data refers to data sets that have both relevant features and the target results that the programmer is interested in. For example, we may be interested in developing a machine learning model that classifies images with dogs in them. The data sets for supervised machine learning would indicate whether a given images has dogs or not. The learning process begins with the algorithm fitting trends found in the training data set into different types of models. The algorithm compares the prediction errors of the models by inputting the validation set data into each model, measuring their accuracy. This allows the algorithm to decide which of the various models is best suited as the resulting machine learning model. Finally, the machine learning model is then evaluated by assessing the accuracy of the predictive power of the model. The developed model is then applied to data without a correct answer to test the validity of the model. In unsupervised machine learning, the data sets are “unlabeled” data, which may not contain the result that the programmer is interested in. Returning to our dog image classification example, data sets for unsupervised machine learning will have pictures of various animals that are not labeled—the computer does not know which pictures are associated with dogs. The unsupervised machine learning algorithm develops a model that extracts common elements from the picture, teaching itself the set of features that makes the subject of the picture a dog. In essence, unsupervised machine learning uses data sets that do not have specific labels fed into the algorithm for the purpose of identifying common trends embedded in that data set. The objective of developing such machine learning models varies. Sometimes the goal is to develop a prediction model that can forecast a variable from a data set. Classification, which assigns records to a predefined group, is also a key application of the algorithm. Clustering refers to splitting records into distinct groups based on the similarity within such group. Association learning identifies the relationship between features.         Figure 1. Overview of Machine Learning Model Development   Figure 1 illustrates the overall process of machine learning model development. The learning process of machine learning algorithms begins with aggregation of data. The data originates from an array of diverse sources ranging from user input, sensor measurement, or monitoring of user behavior.[36] The data sets are then preprocessed. The quality of data presents a challenge in improving machine learning models—any data that has been manually entered contains the possibility of error and bias.[37] Even if the data is collected through automatic means, such as health monitoring systems or direct tracking of user actions, the data sets require preprocessing to account for systematic errors associated with the recording device or method.[38] This includes data skews due to difference between individual sensors, errors in the recording or transmission of data, and incorrect metadata about the sensor.[39] Simply put, the data sets may have differing reference points, embedded biases, or differing formats. The “cleaning” process accommodates for the data skews. The objective of machine learning models is to identify and quantify “features” from a given data set. The term “feature” refers to individually measurable property of an observed variable.[40] From the outset, there may be an extensive list of features that are present in a set of data. It would be computationally expensive to define and quantify each feature, and then to identify the inter-feature relationships, from massive amounts of data. Due to the high demand for the computational power required for processing massive amounts of data, dedication of computational resources to features that are outside the scope of the designer’s interest would be a waste of such limited computational capacity.[41] The machine learning algorithm reduces waste of computational resources by applying dimensionality reduction to the pre-processed data sets.[42] The algorithm can identify an optimal subset of features by reducing the dimension and the noise of the data sets.[43] Dimensionality reduction allows the machine learning model to achieve higher level of predictive accuracy, increased speed of learning, and improves the simplicity and comprehensibility of the results.[44] However, the reduction process has limitations—reducing dimensionality inevitably imposes a limit on the amount of insights and information that may be extracted from the data sets. If the machine learning algorithm discerns a certain feature, the model would not be able to draw inferences related to said feature. Following dimensionality reduction, the machine learning algorithm attempts to fit the data sets into preset models. Typically, three different types of data are fed into the machine learning model—training set, validation set, and test set.[45] The machine learning algorithm “trains” the model by fitting the training set data into various models to evaluate the accuracy of each selection. Then the validation set is used to estimate error rates of each model when applied to data outside the training set that was used to develop each model. Through this process, the machine learning algorithm selects the model that best describes the characteristics and trends of the target features from the test and validation sets.[46] The test set is then used to calculate the generalized prediction error, which is reported to the end user for proper assessment of the predictive power of the model.[47] Simply put, the training test and validation set is used to develop and select a model that reflects the trends of the given data set, and the test set is used to generate a report on the accuracy of the selected model. The crucial elements in developing a machine learning model are (1) training data, (2) inventions related to the machine learning algorithm such as the method of preprocessing the training data, the method of dimensional reduction, feature extraction, and the method of model learning/testing, and (3) the machine learning model and output data.[48] An ancillary element associated with the three elements above is the human talent that is required to implement such innovation.[49] Innovators in the field of machine learning may protect their investments by protecting one or more of the elements listed above. The difference between training data and output data, as well as the difference between the machine learning algorithm and the machine learning model, are best illustrated with an example. Let us assume a credit card company wants to use machine learning to determine whether the company should grant a premium credit card to a customer. Let us further assume that the company would prefer to grant this card to customers that would be profitable to the company while filtering out applicants that are likely to file for bankruptcy. Data sets about prior applicant information would correspond to training data. The company would apply a mathematical method of extracting insight about the correlation between features and the criteria that the company wants to evaluate (e.g., profitable for the firm or likely to file bankruptcy). The mathematical methods are referred as machine learning algorithms. The resulting mechanism, such as a scoring system, that determines the eligibility of card membership is the machine learning model. The credit card applicant’s personal data would be the input data for the machine learning model, and the output data would include information such as expected profitability of this applicant and likelihood of bankruptcy for this applicant.

B. Industry Trends in Machine Learning

Discussing incentive structures and trends behind the machine learning industry is essential in identifying adequate methods of intellectual property rights. The current trends in the world of machine learning will predict what intellectual property regime is most useful to companies to protect their work. The United States has chronically struggled to maintain adequate supply of talent in the high-tech industry, a deficit of talent that continues in the field of machine learning.[50] From a report by the McKinsey Global Institute, the United States’ demand for talent in deep learning “could be 50 to 60 percent greater than its projected supply by 2018.”[51] Coupled with the dearth of machine learning specialists, the short employment tenure of software companies further complicates the search for talent. Software engineers from companies such as Amazon and Google have reported an average employment tenure of one year.[52] While some parts of the high attrition rate may be attributed to cultural aspects of the so-called “Gen Y” employees, the “hot” demand for programming talent has significant impact on the short employee tenure.[53] Job mobility within the software industry is likely to increase as the “talent war” for data scientist intensifies. Employee mobility and California’s prohibition against “covenants not to compete” have been accredited as a key factor behind the success of Silicon Valley.[54] Another trend in the field is the rapid advances in machine learning methods. Due to the fast-paced development of the field, data scientists and practitioners have every reason to work with companies that would allow them to work at the cutting edge of machine learning, using the best data sets. This may influence the attrition rates and recruiting practices of the software industry mentioned above.[55] Eagerness of employees to publish scientific articles and contribute to the general machine learning committee may be another factor of concern. To accelerate innovation by repurposing big data for uses different from the original purpose, and to form common standards for machine learning, more industries are joining alliances and collaborations.[56] Cross-industry collaborations may enable endless possibilities. Imagine the inferences that may be drawn by applying machine learning methods to dietary data from home appliances, biometric data, and data on the weather patterns around the user. Putting privacy nightmares aside, machine learning with diverse data sets may unlock applications that were not previously possible. More companies are attempting to capitalize on commercial possibilities that data sharing may unlock.[57]

C. Machine Learning Innovators – Protect the Data or Inventions?

Though it may seem intuitive that patent protection may be the best option, innovations in machine learning may not need patent protection. Trade secret protection on the data sets may be sufficient to protect the interests of practicing entities while avoiding disclosure of their inventions during the patent prosecution process. Furthermore, numerous software patents have been challenged as unpatentable abstract subject matter under 35 U.S.C. §101 since the Alice decision in 2014.[58] Though subsequent decisions provided guidelines for types of software patents that would survive the Alice decision, it is not clear how the judiciary will view future machine learning patents. Such issues raise the question about the patentability of machine learning – should we, and can we, resort to patents to protect machine learning inventions? Following the discussion on the building blocks of machine learning and recent emerging trends in the field, this section discusses the mode and scope of protection that current legal system provides for each element pertinent to innovation in machine learning. The possible options for protecting innovations are (1) non-disclosure agreements and trade secret law, (2) patent law, and (3) copyright. The three options for protection may be applied to the three primary areas of innovation—(1) training data, (2) inventions related to computation, data processing, and machine learning algorithms, and (3) machine learning models and output data. This discussion will provide context about the methods of protection for innovations in machine learning by examining the costs and benefits of the various approaches.
1. Protecting the Training Data—Secrecy Works Best
Access to massive amounts of training data is a prime asset for companies in the realm of machine learning. The big data phenomenon, which triggered the surge of interest in machine learning, is predicated on the need for practices to analyze large data resources and the potential advantages from such analysis.[59] Lack of access to a critical mass of training data prevents innovators from making effective use of machine learning algorithms. Previous studies suggest that companies resent sharing data with each other.[60] Michael Mattioli discusses the hurdles against sharing data and considerations involved with reuse of data in his article Disclosing Big Data.[61] Indeed, there may be practical issues that prevent recipients of data from engaging in data sharing. Technical challenges in comparing data from different sources, or inherent biases embedded in data sets may be reasons that complicate receiving outside data.[62] Mattioli also questions the adequacy of the current patent and copyright system to promote data sharing and data reuse—information providers may prefer not to disclose any parts of their data due to the rather thin legal protection for databases.[63] Perhaps this is why secrecy seems to be the primary method of protecting data.[64] The difficulty of reverse engineering to uncover the underlying data sets promotes the reliance on non-disclosure.[65] Compared to the affirmative steps required to maintain trade secret protection if the data is disclosed, complete non-disclosure may be a cost effective method of protecting data.[66] Companies that must share data with external entities may exhibit higher reliance on contract law rather than trade secret law. In absence of contract provisions, it would be a challenge to prove that the trade secret has been acquired by misappropriation of the recipient party. The “talent war” for data scientists may also motivate companies to keep the training data sets secret. With a shortage of talent to implement machine learning practices and rapid developments in the field, retaining talent is another motivation for protecting against unrestricted access to massive amounts of data. Companies may prefer exclusivity to the data sets that programmers can work with — top talents in machine learning are lured to companies with promises of exclusive opportunities to work with massive amounts of data.[67] The rapid pace of development in this field encourages practitioners to seek opportunities that provide the best resources to develop their skill sets. This approach is effective since a key limitation against exploring new techniques in this field is the lack of access to high quality big data. Overall, secrecy over training data fits well with corporate recruiting strategies to retain the best talents in machine learning. Non-disclosure and trade secret protection seems to be the best mode of protection. First, despite the additional legal requirements necessary to qualify as trade secrets, trade secret protection fits very well with non-disclosure strategy. On the other hand, patent law is at odds with the principle of non-disclosure. While trade secret law provides companies protection without disclosing information, patent law requires disclosure in exchange for monopolistic rights. Furthermore, neither patent nor copyright provide adequate protection for underlying data. Patent law rewards creative concepts and inventions, not compiled facts themselves. Copyright may protect labeling or distinct ways of compiling information, but does not protect underlying facts. Also, as a practical matter, the difficulty of reverse engineering of machine learning models does not lend well to detecting infringement. Analysis of whether two parties used identical training data would not only be time consuming and costly, but may be fundamentally impossible. If companies were to seek protection of training data, it would be best to opt for secrecy by non-disclosure. This would mean companies would opt out of the cross-industry collaborations that were illustrated above. This may be less of a concern for innovation, as companies may still exchange output data as means of facilitating cross-industry collaboration.
2. Protecting the Inventions—Patent Rights Prevail
Adequate protection over inventive approaches in processing data is becoming increasingly important as various industries begin to adopt a collaborative alliance approach in machine learning. Cross-industry collaboration requires implementation of methods such as preprocessing diverse data sets for compatibility. As the sheer amount of data increases, more processing power is required. The machine learning algorithm needs to maintain a high degree of dimensionality to accurately identify the correlations between a high number of relevant features. The need for more innovative ideas to address such technological roadblocks will only intensify as we seek more complex applications for machine learning. The three primary areas where novel ideas would facilitate innovations in machine learning are pre-training data processing, dimensional reduction, and the machine learning algorithm. Access to massive amounts of data alone is not sufficient to sustain innovation in machine learning. The raw data sets may not be compatible with each other, requiring additional “cleaning” of data prior to machine learning training.[68] The data provided to the machine learning algorithm dictates the result of the machine learning model, hence innovations in methods to merge data with diverse formats is essential to enhancing the accuracy of the models. As cross-industry data analysis becomes more prominent, methods of merging data will have more significant impact on advancing the field of machine learning than mere collection of large data sets. Cross-industry data sharing would be useless unless such data sets are merged in a comparable manner.[69] Companies can opt to protect their inventive methods by resorting to trade secret law. The difficulty of reverse engineering machine learning inventions, coupled with the difficulty of patenting software methods provides incentives for innovators to keep such inventions secret from the public. However, two factors would render reliance on non-disclosure and trade secret ineffective—frequent turnover of software engineers and rapid speed of development in the field. Rapid dissemination of information from employment mobility may endanger intellectual property protection based on secrecy. Furthermore, while the law will not protect former employees that reveal trade secrets to their new employers, the aforementioned fluid job market coupled with general dissemination of information make it difficult to distinguish between trade secrets from former employment and general knowledge learned through practice. The difficulties of reverse engineering machine learning models work against the trade secret owner as well in identifying trade secret misappropriation—how do you know others are using your secret invention? The desire for software communities to discuss and share recent developments in the field does not align well with the use of secrecy against innovations in machine learning. Secrecy practices disincentivize young data scientists from joining due to the limits against gaining recognition.[70] The rapid development of machine learning technology also presents challenges against reliance on trade secret law. Secret methods may be independently developed by other parties. Neither trade secret law nor non-disclosure agreements protect against independent development of the same underlying invention.[71] Unlike training data, machine learning models, or the output data, there are no practical limitations that impedes competitors from independently inventing new computational methods of machine learning algorithms. With such a fluid employment market, high degree of dissemination of expertise, and rapid pace of development, patent protection may provide the assurance of intellectual property protection for companies developing inventive methods in machine learning. Discussions on overcoming the barriers of patenting software will be presented in later sections.[72]
3. Protecting the Machine Learning Models and Results—Secrecy Again
The two primary products from applying the machine learning algorithms to the training data are the machine learning model and the accumulation of results produced by inputting data into the machine learning model. The “input data” in this context may refer to individual data that is analyzed by the insights gained from the machine learning model. In a recent article, Brenda Simon and Ted Sichelman discuss the concerns of granting patent protection for “data-generating patents,” which refers to inventions that generate valuable information in their operation or use.[73] Exclusivity based on patent protection may be extended further by trade secret protection over the data that has been generated by the patented invention.[74] Simon and Sichelman argue that the extended monopoly over data may potentially overcompensate inventors since the “additional protection was not contemplated by the patent system[.]”[75] Such expansive rights will cause excessive negative impact on downstream innovation and impose exorbitant deadweight losses.[76] The added protection over the resulting data derails the policy rationale behind the quid pro quo exchange between the patent holder and the public by excluding the patented information from public domain beyond the patent expiration date.[77] The concerns addressed in data-generating patents also apply to machine learning models and output data. Corporations may obtain patent protection over the machine learning models. Akin to a preference for secrecy for training data, non-disclosure would be the preferred mode of protection for the output data. The combined effect of the two may lead to data network effects where users have strong incentives to continue the use of a given service.[78] The companies that have exclusive rights over the machine learning model and output data gather more training data, increasing the accuracy of their machine learning products. The reinforcement by monopoly over the means of generating data allows few companies to have disproportionately strong dominance over their competitors.[79] Market dominance by data-generating patents becomes particularly disturbing when the patent on a machine learning model preempts other methods in the application of interest. Trade secret law does not provide protection against independent development. However, if there is only one specific method to obtain the best output data, no other party would be able to create the output data independently. The exclusive rights over the only methods of producing data provides means for the patent holder to monopolize both the patent and the output data.[80] From a policy perspective, the excessive protection does seem troubling. Yet such draconian combinations are less feasible after the recent rulings on patentable subject matter of software, which will be discussed below.[81] Mathematical equations or concepts are likely directed to an “abstract concept,” thus will be deemed directed to a patent ineligible subject matter.[82] Furthermore, though recent cases in the Federal Circuit have found precedents where software patents passed the patentable subject matter requirement, those cases expressed limitations against granting patents that would improperly preempt all solutions to a particular problem.[83] The rapid pace of innovation in the field of machine learning compared to the rather lengthy period required to obtain patents may also dissuade companies from seeking patents. Overall, companies have compelling incentives to rely on non-disclosure and trade secrets to protect their machine learning models instead of seeking patents. The secrecy concerns regarding training data applies to machine learning models and the output data as well. Non-disclosure would be the preferred route of obtaining protection over the two categories. However, use of non-disclosure or trade secrets to protect machine learning models and output data presents challenges that are not present in the protection of training data. The use of secrecy to protect machine learning models or output data conflicts with recruiting strategies to hire and retain top talent in the machine learning field. The non-disclosure agreements limit the employee’s opportunity to gain recognition in the greater machine learning community. In a rapidly developing field where companies are having difficulty hiring talent, potential employees would not look fondly on corporate practices that limit avenues of building a reputation within the industry.[84] Companies have additional incentives to employ a rather lenient secrecy policy for machine learning models and the output data. They have incentives to try to build coalitions with other companies to monetize on the results. Such cross-industry collaboration may be additional source of income for those companies. The data and know-how that Twitter has about fraudulent accounts within their network may aid financial institutions such as Chase with novel means of preventing wire fraud. The reuse of insights harvested from the large amount of raw training data can become a core product the companies would want to commercialize. Data reuse may have an incredible impact even for applications ancillary to the primary business of the company. Interesting aspects of disclosing machine learning models and output data are the difficulty of reverse engineering and consistent updates. If the company already has sufficient protection over the training data and/or the computational innovations, competitors will not be able to reverse engineer the machine learning model from the output data. Even with the machine learning model, competitors will not be able to provide updates or refinements to the model without the computational techniques and the sufficient data for training the machine learning algorithm. In certain cases, the result data becomes training data for different applications, which raises concerns of competitors using the result data to compete with the innovator. Yet the output data would contain less features and insights compared to the raw training data that the innovator possesses, and therefore would inherently be at a disadvantage when competing in fields that the innovator has already amassed sufficient training data. Grant of patents on machine learning models may incentivize companies to build an excessive data network while preempting competitors from entering competition. This may not be feasible in the future, as technological preemption is becoming a factor of consideration in the patentable subject matter doctrine. Companies may use secrecy as an alternative, yet may have less incentives to keep secrecy compared to the protection of training data.

D. Need of Patent Rights for Machine Learning Inventions in the Era of Big Data

The current system, on its surface, does not provide adequate encouragement for data sharing. If anything, companies have strong incentives to avoid disclosure of their training data, machine learning model, and output data. Despite these concerns, data reuse may enable social impacts and advances that would not be otherwise possible. Previous studies have pointed out that one of the major barriers preventing advances in machine learning is the lack of data sharing between institutions and industries.[85] Data scientists have demonstrated that they were able to predict flu trends with data extracted from Twitter.[86] Foursquare’s location database provides Uber with the requisite data to pinpoint the location of users based on venue names instead of addresses.[87] Information about fraudulent Twitter accounts may enable early detection of financial frauds.[88] The possibilities that cross-industry data sharing may bring are endless. To encourage free sharing of data, companies should have a reliable method of protecting their investments in machine learning. At the same time, protection based on non-disclosure of data would defeat of purpose of promoting data sharing. Hence protection over computation methods involved with machine learning maintains the delicate balance between promoting data sharing and protecting innovation. Protection over inventions in the machine learning algorithm provides one additional merit other than allowing data sharing and avoiding the sort of excessive protection that leads to a competitor-free road and data network effects. It incentivizes innovators to focus on the core technological blocks to the advancement of technology, and encourages disclosure of such know-how to the machine learning community. Then what are the key obstacles in obtaining patents in machine learning inventions? While there are arguments that the definiteness requirement of patent law is the primary hurdle against patent protection of machine learning models due to reliance on subjective judgment, there is no evidence that the underlying inventions driving big data faces the same challenge.[89] Definiteness may be overcome by providing reasonable certainty for those skilled in the art of defining what the scope of the invention is at the time of filing.[90] There is no inherent reason why specific solutions for data cleaning, enhancement of computation efficiency, and similar inventions would be deemed indefinite by nature. Since the United States Supreme Court invalidated a patent on computer implemented financial transaction methods in the 2014 Alice decision, the validity of numerous software and business method patents were challenged under 35 U.S.C. §101.[91] As of June 8th, 2016, federal district courts invalidated 163 of the 247 patents that were considered under patentable subject matter—striking down 66% of challenged patents.[92] The U.S. Court of Appeals for the Federal Circuit invalidated 38 of the 40 cases it heard.[93] Arguably, the public benefits more from such high rates of post-issuance invalidity. The public still has access to the disclosures from the patents and patent applications. In reliance on granted patents, companies may have already invested in growing related businesses, catering to the need of consumers. At the same time, the patent holder’s monopolistic rights have been shortened as the result of litigation. Effectively, the price that the public pays to inventors in exchange for the benefits of disclosure is reduced. Yet the high degree of invalidity raises several concerns for the software industry. Smaller entities, lacking market influence and capital, have difficulty competing against established corporations without the monopolistic rights granted through the patent system. Investors become hesitant to infuse capital into startups for fear that invalidity decreases the worth of patents. Reliance on trade secret has its own limitations due to the disclosure dilemma—the inventor needs to disclose the secret to lure inventors, but risks losing secrecy in the process. Copyright law does not provide appropriate protection. The restrictions imposed by the merger doctrine and scène à faire doctrine constrain copyright protection of software. Though copyright provides an alternative method of protecting literal copying of code, it does little to protect the underlying software algorithms and innovation. Ultimately, the increase of alliances and collaboration provides incentives for parties to obtain patent rights. Reliance on trade secret or copyright are not suitable methods of protecting their intellectual property. Furthermore, market power or network effects alone cannot sufficiently mitigate the risks involved with operating a business. Patents become even more important for startups since patents provide investors with assurance that in the worst case, the patents may still serve as potential collateral.        

III. Patentability of Machine Learning Innovations in the Era of Big Data

  Patentable subject matter continues to be a barrier for patenting innovations in software. Additional doctrines such as enablement, written description, and obviousness are also serious obstacles against obtaining patents, yet such requirements are specific to each claimed invention and the draftsmanship of claims. Subject matter is considered a broader, categorical exclusion of patent rights. This section explores the current landscape of the patentable subject matter doctrine in the software context.

A. Alice: The Legal Framework of Patentable Subject Matter in Software

The complexity involved with software, coupled with the relatively broad scope of software patents, has presented challenges in identifying the boundaries of the claims.[94] Many members of the software community detest imposing restrictions on open source material and attest that many key innovations in algorithms are rather abstract.[95] Such hostility against patenting software has raised the question of whether patent rights should be the proper method of protecting innovations in software. Alice was a case that embodied such opposition to the grant of software patents. The case involved patents on computerized methods for financial trading systems that reduce “settlement risk” when only one party to financial exchange agreement satisfies its obligation.[96] The method proposed the use of a computer system as a third-party intermediary to facilitate the financial obligations between parties.[97] The United States Supreme Court ruled that the two-step test established from Mayo governed all patentable subject matter questions.[98] In particular, for the abstract idea context, the Supreme Court established the following two-step framework for patentable subject matter of software inventions:  
1. Step one: “[D]etermine whether the claims at issue are directed to a patent-ineligible concept. If so, the Court then asks whether the claim’s [additional] elements, considered both individually and ‘as an ordered combination,’ ‘transform the nature of the claim’ into a patent-eligible application.”[99]
2. Step two: “[E]xamine the elements of the claim to determine whether it contains an ‘inventive concept’ sufficient to ‘transform’ the claimed abstract idea into a patent-eligible application. A claim that recites an abstract idea must include ‘additional features’ to ensure that the [claim] is more than a drafting effort designed to monopolized the [abstract idea]” which requires “more than simply stat[ing] the [abstract idea] while adding the words ‘apply it.’”[100]
  The Alice Court found that the patent on financial transaction was “directed to a patent-ineligible concept: the abstract idea of intermediated settlement,” and therefore failed step one.[101] Furthermore, the Court ruled that the claims did “no more than simply instruct the practitioner to implement the abstract idea of intermediated settlement on a generic computer” and did not provide an inventive concept that was sufficient to pass step two.[102]

B. The post-Alice cases from the Federal Circuit

The Alice framework was considered as a huge setback for the application of patentable subject matter doctrine to software. It was a broad, categorical exclusion of certain inventions that were deemed “directed to” an abstract idea, natural phenomenon, or law of nature. The biggest misfortune was the lack of guidance in the Alice decision on the threshold for such categorical exclusion—we were left without any suggestions on the type of software patents that would be deemed as patentable subject matter. The recent line of cases in the Federal Circuit provides the software industry with the much-needed clarification on the standards that govern patentability of software inventions. Enfish v. Microsoft, decided on March 2016, involved a “model of data for a computer database explaining how the various elements of information are related to one another” for computer databases.[103] In June 2016, the Federal Circuit decided another case on the abstract idea category for patentable subject matter. Bascom Global v. AT&T Mobility is on a patent disclosing an internet content filtering system located on a remote internet service provider (ISP) server.[104] Shortly after Bascom, the Federal Circuit decided McRO v. Bandai Namco Games in September 2016.[105] The case ruled that an automated 3D animation algorithm that renders graphics in between two target facial expressions is patentable subject matter.[106] The rulings from the Federal Circuit on the aforementioned three cases provide guidelines along the two-step Alice test of patentable subject matter. The software patents in Enfish and McRO were deemed “directed to” a patent eligible subject matter, informing the public of what may pass the first set of the Alice test. Bascom failed the first step.[107] Yet the court ruled that those patents had inventive concepts sufficient to transform a patent ineligible subject matter into a patent eligible application. Combined together, the three cases give more certainty in what may pass the 35 U.S.C. §101 patentable subject matter inquiry. Reiterating the Alice test, whether an invention is a patentable subject matter is determined by a two-step process—(1) is the invention directed to, rather than an application of, an abstract idea, natural phenomenon, or law of nature, and even if so, (2) do the elements of the claim, both individually and combined, contain an inventive concept that transforms this invention into a patent-eligible application? The Federal Circuit fills in the gaps that were left unexplained from the Alice ruling.
1. The Federal Circuit’s Standard for Alice Step One
The Enfish court discussed what constitutes an abstract idea at the first step of the Alice inquiry. Judge Hughes instructs us to look at whether the claims are directed to a specific improvement rather than an abstract idea. In this case, the patent provides the public with a solution to an existing problem by a specific, non-generic improvement to computer functionality. The Enfish court ruled that such invention is patent eligible subject matter.[108] McRO also ruled that the facial graphic rendering for 3D animation was not an abstract concept. Here, the Federal Circuit again emphasized that a patent may pass step one of the Alice test if the claims of the patent “focus on a specific means or method that improves the relevant technology.”[109] The McRO court also noted that preemption concerns may be an important factor for the 35 U.S.C. §101 subject matter inquiry—that improper monopolization of “the basic tools of scientific and technological work” is a reason why such categorical carve outs against granting patents on abstract ideas exist.[110] Bascom provides the standards on what would fail step one of the Alice patentable subject matter inquiry. If the patent covers a conventional, well-known method in the field of interest, then the invention would be considered abstract. This is akin to the inventive concept considerations conducted at the second phase of the 35 U.S.C. §101 subject matter inquiry. The main takeaway from Enfish and McRO is that in the first step of the Alice test, a patent application is not directed to an abstract idea if (1) the invention addresses an existing problem by specific improvements rather than by conventional, well-known methods and (2) the claims do not raise preemption concerns. This encourages practitioners to define the problem as broadly as possible, while defining the scope of improvement in definite terms.
2. The Federal Circuit’s Standard for Alice Step Two, and the Overlap with Step One
The second step of the Alice test is an inquiry of whether the patent application, which is directed to a patent ineligible subject, still contains a patent-worthy inventive concept. Bascom ruled in favor of granting the patent following the second step of the Alice test.[111] While the patent at hand was considered directed to patent ineligible subject matter, the Bascom court found that the content filter system invention still had an inventive concept worthy of a patent.[112] Even if elements of a claim are separately known in prior art, an inventive concept can be found in the non-conventional and non-generic arrangement of known, conventional pieces. This inquiry seems like a lenient standard compared to the 35 U.S.C. §103 obviousness inquiry; hence, it is not clear if this step has an independent utility for invalidating or rejecting a patent. Nonetheless, the court found that merely showing that all elements of a claim were already disclosed in prior art was not sufficient reason to make an invention patent ineligible.[113] While it is possible to infer sufficient reasons of ruling out inventive concepts from the Bascom case, it is still unclear what would warrant an invention to pass the second step of the Alice test. Cases such as DDR Holdings v. have suggested that the second step of Alice is satisfied since it involved a solution to a specific technological problem that “is necessarily rooted in computer technology in order to overcome a problem specifically arising in the realm of computer networks.”[114] This interpretation of inventive concept becomes perplexing when comparing the two steps of Alice—both steps look to whether the proposed solution addresses problems that are specific to a given field of interest. While we would need additional cases to gain insight on whether the two steps have truly distinct functions, at the very least the Federal Circuit provided essential guidelines on what may be deemed as patentable software.

C. Applying Patentable Subject Matter to Machine Learning Inventions

As the Bascom court has taught, the first step in the Alice inquiry is to ask whether an invention (1) provides a solution to an existing problem by (2) a specific, non-generic improvement that (3) does not preempt other methods of solving the existing problem. Applying this test to inventions in machine learning, mathematical improvements and computational improvements would be treated differently. As mentioned before, a key aspect of machine learning is the “noise” associated with the data sets.[115] Another concern is the fitting of a given algorithm to a certain model. Methods that facilitate the computations of the training process may be deemed as a specific improvement. However, machine learning algorithms themselves, including the base models that the algorithm fits the training into would not be pertinent to just a specific improvement. Hence, generic mathematical methods applicable to various problems are directed to an abstract idea. For example, an invention that addresses the issue of normalizing data from different sources would be a computational issue and hence would pass the Alice test given that it did not preempt other solutions to the problem of data normalization. On the other hand, a specific mathematical equation that serves as a starting model for the machine learning algorithm would be mathematical and hence directed to an abstract idea. Even if the mathematical starting model is only good for a specific application, the model is not a specific improvement pertinent to that application. Although the model may not necessarily be a good starting model for other applications, it is nonetheless a generic solution that applies to other applications as well.


While highly restrictive, the guidelines from the Federal Circuit still allow the grant of patent rights for the computational aspects of machine learning algorithms. The guidelines also would prevent highly preemptive mathematical innovations, including data-generating patents such as machine learning models. The narrow range of patentability makes a patent regime appealing for computational methods. The recent emphasis on preemption concerns acts in favor of preventing data network effects based on data-generating patents. While not discussed in this paper, other patentability requirements such as obviousness or definiteness would further constraint the grant of overly broad data-generating patents. Such an approach strikes the appropriate balance between promoting innovation and encouraging data reuse for societal benefits. Compared to other approaches of providing protection over innovations in machine learning, the narrowly tailored approach for patent rights for computational inventions fits best with the policy goal of promoting innovation through data reuse. The industry trends in collaboration and recruiting also matches the proposed focus on patent law protection.
* J.D. Candidate, New York University School of Law, 2018; Ph.D. in Electrical and Computer Engineering, University of Illinois at Urbana-Champaign, 2012. The author would like to thank Professor Katherine Jo Strandburg for her guidance; fellow JIPEL Notes Program participants Julian Pymento, Gia Wakil, Neil Yap, and Vincent Honrubia for their comments and feedback; and Dr. Sung Jin Park for her support throughout the process.
[1] Sang-Hun Choe & John Markoff, Master of Go Board Game Is Walloped by Google Computer Program, N.Y. Times (March 9, 2016), (reporting the shocking defeat of Go Master Lee Se-dol to Google DeepMind’s AlphaGo).
[2] Laurence Zuckerman, Chess Triumph Gives IBM a Shot in the Arm, N.Y. Times (May 12, 1997), (detailing IBM’s highly publicized win through Deep Blue’s victory over world chess champion Garry Kasparov).
[3] See Choe & Markoff, supra note 1.
[4] David Silver et al., Mastering the game of Go with deep neural networks and tree search, 529 Nature 484, 484 (2016).
[5] Id.
[6] Id.
[7] Id. at 485.
[8] Id.
[9] See Andre Esteva et al., Dermatologist-level classification of skin cancer with deep neural networks, 542 Nature 115 (2017).
[10] See Sue Ellen Haupt & Branko Kosovic, Big Data and Machine Learning for Applied Weather Forecasts, (2015).
[11] See Wei-Yang Lin et al., Machine Learning in Financial Crisis Prediction: A Survey, 42 IEEE Transactions on Systems, Man, and Cybernetics 421 (2012).
[12] See Farzindar Atefeh & Wael Khreich, A Survey of Techniques for Event Detection in Twitter, 31 Computational Intelligence 132 (February 2015).
[13] See Corey Blumenthal, ECE Illinois Students Accurately Predicted Trump’s Victory, ECE Illinois (Nov. 18, 2016),
[14] For the purpose of this Note, secrecy refers to the use of trade secret and contract based non-disclosure agreements.
[15] Mark A. Lemley, The Surprising Virtues of Treating Trade Secrets As IP Rights, 61 Stan. L. Rev. 311, 332 (2008) (“Patent and copyright law do not exist solely to encourage invention, however. A second purpose — some argue the main one — is to ensure that the public receives the benefit of those inventions.”).
[16] Andrew Ng et al., How Artificial Intelligence Will Change Everything, Wall Street Journal (March 7, 2017),
[17] Limor Peer, Mind the Gap in Data Reuse: Sharing Data Is Necessary But Not Sufficient for Future Reuse, London Sch. Econ. & Poli. Sci. (Mar. 28, 2014) (“The idea that the data will be used by unspecified people, in unspecified ways, at unspecified times . . . is thought to have broad benefits”).
[18] See Saeed Ahmadiani & Shekoufeh Nikfar, Challenges of Access to Medicine and The Responsibility of Pharmaceutical Companies: A Legal Perspective, 24 DARU Journal of Pharmaceutical Sciences 13 (2016) (discussing how “pharmaceutical companies find no incentive to invest on research and development of new medicine specified for a limited population . . .”).
[19] 17 U.S.C. §101 (2012).
[20] Id.
[21] 17 U.S.C. §102(a) (Copyright exists “in original works of authorship fixed in any tangible medium of expression . . .”).
[22] Apple Comput., Inc. v. Franklin Comput. Corp., 714 F.2d 1240 (3d Cir. 1983).
[23] Comput. Assocs. Int’l v. Altai, 982 F.2d 693 (2d Cir. 1992).
[24] Id.
[25] See id. at 707-09.
[26] 837 F.3d 1299, 1314 (Fed. Cir. 2016).
[27] Altai, 982 F.2d at 698.
[28] See Dionne v. Se. Foam Converting & Packaging, Inc., 240 Va. 297 (1990).
[29] Kewanee Oil v. Bicron Corp., 416 U.S. 470 (1974).
[30] Id. at 490.
[31] Nikolay Golova & Lars Rönnbäck, Big Data Normalization For Massively Parallel Processing Databases, 54 Computer Standards & Interfaces 86, 87 (2017).
[32] Jon Russell, Alibaba Teams Up with Samsung, Louis Vuitton and Other Brands to Fight Counterfeit Goods, TechCrunch (Jan. 16, 2017)
[33] Lemley, supra note 15, at 33
[34] See Lior Rokach, Introduction to Machine Learning, Slideshare 3 (July 30, 2012),
[35] Id. at 4.
[36] Id. at 10.
[37] See Lars Marius Garshol, Introduction to Machine Learning, Slideshare 26 (May 15, 2012)
[38] Id.
[39] Id.
[40] See Lei Yu et al., Dimensionality Reduction for Data Mining – Techniques, Applications and Trends, Binghamton University Computer Science 11, (last visited Feb. 23, 2018).
[41] Id.
[42] See Rokach, supra note 34, at 10.
[43] Yu et al., supra note 40.
[44] Laurens van der Maaten et al., Dimensionality Reduction: A Comparative Review, Tilburg Centre for Creative Computing, TiCC TR 2009-005, Oct. 26, 2009, at 1 (“In order to handle such real-world data adequately, its dimensionality needs to be reduced. Dimensionality reduction is the transformation of high-dimensional data into a meaningful representation of reduced dimensionality. Ideally, the reduced representation should have a dimensionality that corresponds to the intrinsic dimensionality of the data. The intrinsic dimensionality of data is the minimum number of parameters needed to account for the observed properties of the data”).
[45] Andrew Ng, Nuts and Bolts of Applying Deep Learning (Andrew Ng), YouTube (Sept. 27, 2016),
[46] Andrew Ng, Model Selection and Train/Validation/Test Sets, Machine Learning, (last visited Feb. 23, 2018).
[47] Id.
[48] See Rokach, supra note 34, at 10.
[49] Alex Rampell & Vijay Pande, a16z Podcast: Data Network Effects, Andreesen Horowitz (Mar. 8, 2016),
[50] James Manyika et. al., Big Data: The Next Frontier for Innovation, Competition, and Productivity, McKinsey Global Inst., May 2011, at 11, available at
[51] Id.
[52] Leonid Bershidsky, Why Are Google Employees So Disloyal?, Bloomberg (July 13, 2013, 11:41 AM),
[53] Id.
[54] Rob Valletta, On the Move: California Employment Law and High-Tech Development, Federal Reserve Bank of S.F. (Aug. 16, 2002),
[55] Id.
[56] See Quentin Hardy, IBM, G.E. and Others Create Big Data Alliance, N.Y. Times (Feb. 15, 2015),
[57] See, e.g., Finicity and Wells Fargo Ink Data Exchange Deal, Wells Fargo (Apr. 4, 2017),
[58] Alice Corp. Pty. Ltd. v. CLS Bank Int’l, 134 S. Ct. 2347 (2014).
[59] Karen E.C. Levy, Relational Big Data, 66 Stan. L. Rev. Online 73, 73 n.3 (2013), (explaining that the big data phenomenon is due to the need of practices to analyze data resources).
[60] Christine L. Borgman, The Conundrum of Sharing Research Data, 63 J. Am. Soc’y for Info. Sci. & Tech. 1059, 1059-60 (2012) (discussing the lack of data sharing across various industries).
[61] See Michael Mattioli, Disclosing Big Data, 99 Minn. L. Rev. 535 (2014).
[62] See id. at 545-46 (discussing the technical challenges in merging data from different sources, and issue of subjective judgments that may be infused in the data sets).
[63] See id. at 552 (discussing how institutions with industrial secrets may rely on secrecy to protect the big data they have accumulated).
[64] See id. at 570 (“[T]he fact that these practices are not self-disclosing (i.e., they cannot be easily reverse-engineered) lends them well to trade secret status, or to mere nondisclosure”).
[65] Id.
[66] Id. at 552.
[67] Patrick Clark, The World’s Top Economists Want to Work for Amazon and Facebook, Bloomberg (June 13, 2016, 10:47 AM), (“If you want to be aware of what interesting questions are out there, you almost have to go and work for one of these companies”).
[68] Bill Franks, Taming the Big Data Tidal Wave 20 (2012) (discussing that the biggest challenge in big data may not be developing tools for data analysis, but rather the processes involved with preparing the data for the analysis).
[69] See Borgman, supra note 60, at 1070 (“Indeed, the greatest advantages of data sharing may be in the combination of data from multiple sources, compared or “mashed up’ in innovative ways.” (citing Declan Butler, Mashups Mix Data Into Global Service, 439 Nature 6 (2006))).
[70] Jack Clark, Apple’s Deep Learning Curve, Bʟᴏᴏᴍʙᴇʀɢ Bᴜsɪɴᴇssᴡᴇᴇᴋ, (Oct 29, 2015)
[71] Kewanee Oil v. Bicron Corp., 416 U.S. 470, 490 (1974).
[72] See infra Section III-B.
[73] Brenda Simon & Ted Sichelman, Data-Generating Patents, 111 Nw. U.L. Rev. 377 (2017).
[74] Id. at 379.
[75] Id. at 414.
[76] Id. at 415 (“[B]roader rights have substantial downsides, including hindering potential downstream invention and consumer deadweight losses . . .”).
[77] Id. at 417.
[78] Rampell & Pande, supra note 49.
[79] Lina Kahn, Amazon’s Antitrust Paradox, 126 Yale L.J. 710, 785 (2017) (“Amazon’s user reviews, for example, serve as a form of network effect: the more users that have purchased and reviewed items on the platform, the more useful information other users can glean from the site”).
[80] Simon & Sichelman, supra note 73, at 410.
[81] See infra Section III-A.
[82] Id.
[83] See infra Section III-B.
[84] Jack Clark, Apple’s Deep Learning Curve, Bʟᴏᴏᴍʙᴇʀɢ Bᴜsɪɴᴇssᴡᴇᴇᴋ (Oct 29, 2015),
[85] Peer, supra note 17 (“The idea that the data will be used by unspecified people, in unspecified ways, at unspecified times . . . is thought to have broad benefits”).
[86] See Harshavardhan Achrekar et al., Predicting Flu Trends using Twitter data, IEEE Conference on Comput. Commc’ns. Workshops 713 (2011),
[87] Jordan Crook, Uber Taps Foursquare’s Places Data So You Never Have to Type an Address Again, TechCrunch, (May 25, 2016)
[88] See Rampell & Pande, supra note 49.
[89] See Mattioli, supra note 61, at 554 (“A final limitation on patentability possibly relevant to big data is patent law’s requirement of definiteness”).
[90] See Nautilus, Inc. v. Biosig Instruments, Inc., 134 S. Ct. 2120 (2014).
[91] See Alice Corp. Pty. Ltd. v. CLS Bank Int’l, 134 S. Ct. 2347 (2014).
[92] Robert R. Sachs, Two Years After Alice: A Survey of the Impact of a “Minor Case” (Part 1), Bilski Blog (June 16, 2016),
[93] Id.
[94] Stephanie E. Toyos, Alice in Wonderland: Are Patent Trolls Mortally Wounded by Section 101 Uncertainty, 17 Loy. J. Pub. Int. L. 97,100 (2015).
[95] Id.
[96] Alice Corp. Pty. Ltd. v. CLS Bank Int’l, 134 S. Ct. 2347, 2349 (2014).
[97] Id.
[98] Id.
[99] Id. at 2350 (emphasis added) (citation omitted).
[100] Id. at 2357 (emphasis added) (alteration in original) (citation omitted).
[101] Id. at 2350.
[102] Id. at 2351.
[103] Enfish, LLC v. Microsoft Corp., 822 F.3d 1327, 1330 (Fed. Cir. 2016).
[104] Bascom Glob. Internet Servs. v. AT&T Mobility LLC, 827 F.3d 1341, 1349 (Fed. Cir. 2016).
[105] McRO, Inc. v. Bandai Namco Games Am. Inc., 837 F.3d 1299, 1308 (Fed. Cir. 2016).
[106] Id.
[107] Bascom, 827 F.3d at 1349.
[108] Enfish, 822 F.3d at 1330.
[109] McRO, Inc., 837 F.3d at 1314.
[110] Id.
[111] Bascom, 827 F.3d at 1349.
[112] Id.
[113] Id.
[114] See Toyos, supra note 94, at 121; DDR Holdings, LLC v., 773 F.3d 1245, 1257 (Fed. Cir. 2014).
[115] See supra Section II-A.

The ‘Enhanced No Economic Sense’ Test: Experimenting with Predatory Innovation

The ‘Enhanced No Economic Sense’ Test: Experimenting with Predatory Innovation
By Dr. Thibault Schrepel*

Download a PDF version of this article here



“To evaluate is to create.

Hear this, you creators!”

– Friedrich Nietzsche

This paper originates from a long-standing anachronism of antitrust law with regard to high-tech markets. Conventional wisdom assumes that antitrust law mechanisms are well suited to the study of practices in technology markets and that only adjustments should be made to these mechanisms, and sparingly at that.[2] This is untrue. Several practices fall outside the scope of antitrust law because the mechanisms for assessing the legality of practices are not adequate. In fact, no one can accurately identify a typical legal approach for non-price strategies, a truth which gives way for a chaotic jurisprudence to emerge from this lack of universal understanding, which we will show.[3]

This article further aims to contribute to the literature by advancing a new test, the “enhanced no economic sense” (“ENES”) test, to be applied to non-price strategies.[4] We show why applying it with consistency will help to simplify the law while avoiding legal errors – two goals that all of the tests aiming to assess the legality of practices under antitrust law should reach.[5] Some of these tests, which are too permissive, generate many type-II errors but are easily understandable, and thus, increase legal certainty. Others, which are better suited and, in theory, allow avoidance of legal errors, are too complex to be applied by the courts and, above all, to be understood by laymen.[6] But one must not give up. Antitrust law is not condemned to remain blind to certain technical problems or, on the contrary, to be incomprehensible to the ordinary man. The “ENES” test brings a solution of reason to this long-standing issue.

We should not adopt a new test without first ensuring that it would allow courts and antitrust authorities to take a position in each individual case and that rulings would benefit consumers as a result.[7] Here again, the “enhanced no economic sense” test meets this double objective. It also helps to understand why several decisions made in the past are, we will argue, legal errors. The Microsoft case[8] is one of them.

In turn, this paper makes a proposal to rethink the way most of the new practices implemented in technology markets are actually evaluated. This study is particularly timely because the development of the issues related to these markets, and the growing interest shown by competition authorities[9] calls for an identified position; one which will not hinder their extraordinary growth.

This paper proceeds in three parts. The first part presents the “enhanced no economic sense” test, ranging from its foundations up to detail of its implementation. The second part probes the test’s empirical efficiency, exploring the most important predatory innovation cases on non-price strategy reassessing them through the prism of the “enhanced no economic sense” test, which helps to establish the test’s effectiveness. The last part expands these empirical findings and presents our conclusions.


The “no economic sense” test is best suited to assess non-price strategies—in other words, all those which are unrelated to pricing changes. It complies with most of the characteristics that a test should have—its mechanism is easily understood and most of its criticisms fall short. And yet, the test may be improved to create less type-II errors while retaining its best features. This section details, accordingly, how to design a new version of it.

A. How to Determine Which Test to Apply

Determining which test is best-suited to assess non-price strategies implies considering which goals are to be assigned to antitrust law, and, accordingly, which characteristics the ideal test should have.

1. Regarding the Goals of Antitrust Law

The choice of which test is the most suitable for analyzing non-price strategies involves considering several elements. The first element is related to the goals that must be assigned to antitrust law.[10] These objectives may be grouped into three theories:

1. The “efficiency theory”:[11] according to this theory, antitrust law’s primary goal is to increase economic efficiency. Type-I errors[12] are seen as the greatest evil because they deter investments. Under this theory, there is no presumption of anti-competitive practices simply because a company holds a monopoly power.

2. The “consumer protection theory”:[13] this theory is based on the idea that the ultimate objective of antitrust law is to benefit consumers, not necessarily to increase economic efficiency. It contemplates criteria other than pure economic ones, such as preventing big mergers or overprotecting small businesses.[14]

3. The “growth-based theories”:[15] these related theories aim to enhance economic efficiency and prevent unwarranted transfers of consumers’ surplus. As a result, innovation is at the center of the debate, unlike theories which are only centered on economic efficiency and which do not necessarily involve the protection of innovation in and of itself.[16]

Most non-price strategies assume there are technological aspects directly related to innovative fields.[17] Accordingly, while evaluating practices’ efficiencies, courts must take great care not to impair innovation. For that reason, choosing a test included in the growth-based theories is ideal. The “no economic sense” test evaluates the reason a company has implemented practices and also makes it possible to take innovation and technological breakthroughs into account.

2. In Terms of its Efficiency

The second key component to be studied in order to identify the most appropriate test to detect anti-competitive practices is efficiency.[18] It cautions against multiplying the situations in which the courts are unable to judge whether some practices must be condemned.[19] It also means avoiding type-I and II errors.

In its 2005 study entitled “Competition on the Merits,”[20] the Organization for Economic Co-operation and Development (“OECD”) identified six criteria for evaluating various tests’ efficiencies.[21] The first relates to their accuracy: the tests must be based on accepted economic theories and, meanwhile, must avoid type-I and II errors. The second is linked to their administrability, understood as ease of applicability.[22] The third refers to their applicability: the tests should cover all of the issues raised by one type of anti-competitive practice. The fourth relates to their consistency, which should lead to homogeneous solutions.[23] As underlined by one author, the tests applied to predatory innovation are plural.[24] This should be changed. The fifth emphasizes their objectivity. The sixth and final criterion is related to their transparency. The tests must aim at defending the goals that have been identified as being the most crucial to antitrust law, i.e. the growth-based theories. The ENES test will meet these criteria, as explained below.

B. On Why the “No Economic Sense” Test Is Suitable

1. Its Main Characteristics

The “no economic sense” test is based on the simple idea that a practice should be regarded as anti-competitive if it makes sense from an economic point of view only because of its tendency to eliminate or to restrict competition.[25]

While this test is often presented as being close to the “profit sacrifice” test,[26] the fact of the matter is that it differs greatly from it. For instance, when applying the “no economic sense” test, a practice may be sanctioned if it makes no sense—besides creating anti-competitive effects—even though it did not involve any losses for the company.[27] At the same time, a practice that involves losses may still be seen as being pro-competitive if it is justified by potential gains in economic efficiency.[28] To the contrary, the profit sacrifice test condemns all practices that involve significant short-term sacrifices.[29] In short, the “no economic sense” test raises the question of why the defendant agreed to bear losses, which the profit-sacrifice test does not.

Gregory J. Werden, one of the “no economic sense” test’s greatest defenders, also underlined that, according to this test, practices are seen as being anti-competitive when, in addition of having no objectives other than eliminating or restricting competition, they also have the potential effect of restricting competition.[30] Establishing whether the practices have the potential to eliminate the competition is then a prerequisite,[31] and the burden of proof lies on the complainant.[32]

Also, the “no economic sense” test does not imply an ex-post evaluation of a practice’s effects. Rather the court’s duty is to evaluate the practice by taking into account all of the elements available to the dominant firm at the time of its implementation.[33] A practice may have been extremely profitable for reasons that were not anticipated by the company and this should not lead to a finding that the practice is pro-competitive. Also, as Werden notes,[34] a practice may have anti-competitive effects that were unpredictable at the time of its implementation and it should not be used as a means of late condemnation.

In fact, not evaluating the ex-post effects produced by a practice and not focusing on its costs are the reasons why the “no economic sense” test—even though it may be used to address non-price and price strategies[35]—is even more efficient for practices involving low costs, which is an important difference from the profit sacrifice test. It is then particularly suitable for evaluating predatory innovation, which is why it has led many jurisdictions to apply it in related cases.[36]

Another test called the “sham test” could also be suitable because its grounds are similar to the “no economic sense” test.[37] If we consider that innovation is an economic justification in itself, then, applying the “no economic sense” test asks whether the “innovation” is genuine or a “sham.”[38] An “innovation” will be considered a “sham” if it exists only for its negative effects on competition.[39] In other words, the definition of a “sham innovation” is any product modification that does not improve the consumer’s well-being in the short or the long term. The similarity of the two tests is particularly enlightening on the definition to be given for predatory innovation. In assessing whether innovation is real,[40] these two tests stand as opposing the vision of pioneer scholars Janusz Ordover & Robert Willig: that even genuine innovations can be anti-competitive.[41]

In short, unlike several other tests, which lead to a finding of anti-competitiveness in genuine innovations that improve the consumer welfare whenever anti-competitive effects are significant,[42] the “sham” and the “no economic sense” tests do not. The reason why the “no economic sense” test is preferable to the sham test is because it also encompasses non-price strategies that are not directly linked to products’ modifications, like those related to “low-cost exclusion” as a “refusal to deal”.[43] Further, its terminology is self-explanatory to the extent that the terms indicate its analysis mechanism, which can only increase legal certainty.

All of these reasons led to the application of the “no economic sense” test in the US courts[44] in Aspen Skiing,[45] Matsushita Industrial Co., Ltd. v. Zenith Radio Corp. and Brooke Group,[46] and US v. AMR Corp.[47] The Department of Justice and the Federal Trade Commission also argued for applying it in Trinko.[48] Instead, the Court applied the profit sacrifice test, without naming it, by deciding that the dominant company aimed to sacrifice short-term profits in order to compensate for long term ones.[49] Accordingly, criticisms of that decision are irrelevant to the “no economic sense” test.

European courts have also applied the test on several occasions, although they rarely use the exact terms to describe it. For instance, the Court of Justice implemented it in British Airways[50] and the General Court of the European Union did the same in UFEX[51] and Telefónica, referring to practices as “irrational from an economic point of view.”[52] The test was also applied by national competition authorities in France in British Airways v. Eurostar.[53]

2. Inoperative Criticisms

As we have shown, when dealing with predatory innovation, the “no economic sense” test is the best alternative. For that reason, Herbert Hovenkamp proposes applying it to all cases of this kind.[54] The vast majority of criticisms raised against it fall short.

As for consumer protection. Some have denounced all tests based on analyzing predatory innovation’s effects on the company that implemented it in any given instance.[55] They argue that antitrust law is about addressing the direct effect of practices on consumers[56] and that the “no economic sense” test does not protect consumer welfare.[57] But this criticism is based on false premises. Assessing whether a practice makes economic sense for reasons other than its anti-competitive aspects is the same as protecting consumer welfare because factors favoring consumer welfare align with economically sensible reasons for modifying an existing product. In fact, only two situations exist under the “no economic sense” test:

1. If the practice does not pass the test and thus is deemed to be unlawful, the damage to the consumer is certain since the anti-competitive effects of the practice have driven its implementation;

2. If the practice passes the test and is thus considered to be lawful, it may be:

a. entirely pro-competitive, a situation in which the consumer welfare is necessarily increased;

b. pro and anti-competitive at the same time (the practice is “hybrid”), but the practice is deemed legal as a whole in order not to discourage investments that ultimately benefit the consumer. There is a real danger in micro-analyzing innovation, practice by practice,[58] and applying this test avoids this pitfall to the extent that when practices are considered together a practice may be justified by the presence of another practice that is linked to it.

As for type-II errors.[59] One of the strongest criticisms[60] made to the “no economic sense” test is the courts’ supposed inability to evaluate hybrid practices that produce both positive and negative effects on competition.[61] Situations where a practice is economically justified but also involve anti-competitive features are indeed problematic. Predatory innovation is a good example of this; it would be easy for a company to modify one of its products in a small, pro-competitive way, while also implementing a very effective anti-competitive strategy. In fact, creating an analytical framework for this type of situation is the raison d’être of the “no economic sense” test.[62] This test seeks to avoid balancing the positive and negative effects to the extent that such a process is expensive and uncertain. The ENES test allows for some type-II errors rather than engendering type-I errors that discourage innovation.[63] In short, the “no economic sense” test prefers a result with type-II errors[64] to the possibility of implementing a more intricate test without consistency and accuracy.[65]

As for its simplistic features. Several authors have stressed[66] that some practices could be anti-competitive while being justified from an economic standpoint. This may be the case, for instance, when a company decides not to reveal the existence of several of its patents during a standardization process.[67] In such a case, the firm is implementing an anti-competitive practice although it is economically justified by the fact that it won’t necessitate providing extensive information about its patent—which may be a long and expensive process. In essence, the idea is that the “no economic sense” test is too Manichean,[68] and that a dichotomy between practices that are economically justified and practices that are not is far too removed from economic reality to hold true.[69]

Once again, this criticism seems to disregard the very foundations of the “no economic sense” test. The latter does not advocate for penalizing all practices that do not make economic sense, but only does for those that do not make economic sense without anti-competitive reasons behind them.[70] The example in which a company refuses to provide information to standard organizations is then covered.[71] In fact, let us ask the following question: why condemn the implementation of a practice that a company can justify? Antitrust law is not about imposing on all companies to have an overview of the markets on which they operate. Whenever they can justify a practice, it should be authorized, irrespective of its effects on competition.

As for its manageability. Some have pointed out that the “no economic sense” test was inapplicable when predatory practices involved low costs.[72] This statement is incorrect, however, to the extent that this test condemns practices that exclude competitors, regardless of the costs incurred.[73] Similarly, the criticism related to the inability to implement the test in situations where a company would have mixed intentions must be rejected since this test is indifferent to the subjective intention of the company.[74] One may think, finally, that the test is unsuitable for disruptive innovations in which development makes “no economic sense” because they are, for example, driven by an extravagant philanthropist’s irrational desire. But the “no economic sense” test does not lead to condemning such innovations either if they do not have solely anti-competitive consequences.[75]

As for giving leeway to the judge. The “no economic sense” test is said to create “safe harbors”[76] which would deprive judges of their utility. In this way, the application of the “no economic sense” test could be contravening the spirit of the rule of reason because it would create per se legalities. In reality, applying the “no economic sense” test does not remove the judge from the decision-making process. The judge remains in charge of deciding, according to the law, what constitutes a valid economic justification.[77] Such a role is particularly important as it gives legitimacy to the test as a whole.

As for subjective intent. A distinction should be made between objective and subjective intent. The first—objective intent—is the result of hard evidence, for instance, emails in which the company’s CEO has confirmed his intention to modify a product for the sole purpose of reducing competition.[78] The second one—subjective intent—implies inferring executives’ intentions by analyzing the facts. Some European authors argue for taking subjective intention into account[79] and the Chicago School’s scholars have also done so by stressing that the intention is important[80] when evidence of anti-competitive harm cannot be provided by other means.[81] Courts could adopt three different approaches:[82]

1. not taking subjective intent into account;

2. taking subjective intent into account only when it is proved to be useful and without asking for the anti-competitive intent to be shown in order to condemn a practice;

3. taking subjective intent into account so to establish a violation of antitrust law.

In fact, taking subjective intent into account is championed by Phillip Areeda & Herbert Hovenkamp.[83] Courts have taken subjective intent into account in the C.R. Bard v. M3 Systems[84] case. It was also the position adopted in the Microsoft case.[85] Other cases have followed since then.[86] In fact, several of the tests assessing the legality of unilateral practices take subjective intent into account. The “no economic sense” test does not, which is fortunate for our purposes.[87] In our review, only hard facts and empirical data should be used to establish practices’ illegality. In fact, taking a company’s subjective intent into account is far too unpredictable because it requires frameless control by the judge. Antitrust law is about pursuing economic efficiency. It should lead to imposing penalties only when the evidence of damage is indisputable.[88] Also, because innovation is inherently predatory,[89] taking subjective intention into account makes little sense.

Accordingly, some of the European and North American doctrine on this topic opposes the inclusion of subjective intent.[90] Frank H. Easterbrook, for instance, particularly highlighted the fact that analyzing intention did not allow one to distinguish between true monopolization and attempts to monopolize.[91] He stressed that evaluating a company’s subjective intention was expensive and had the effect of reducing legal certainty.[92] Further, distinguishing anti-competitive intent from pro-competitive intent may be more difficult than it seems. The language used between business executives may wrongly imply an anti-competitive intention. Considering subjective intention could be particularly harmful for small businesses where the language used by the staff may be more explicit than it is in larger corporations where executives are more sensitized to antitrust rules.

As for behavioral economics. Some authors[93] advocate for incorporating “behavioral economics according to which agents sometimes show a bounded rationality.[94] But the “no economic sense” test wisely rejects any consideration of behavioral economics.[95] No one would argue that agents are rational in every situation, and further, behavioral economics lacks empirical evidence to be integrated into the decision-making process.[96] Studies of behavioral economics fail, to the best of our knowledge, to reveal consistent trends that might be integrated into the rule of law.[97] It must therefore be set aside—as least for ex ante purposes—until it becomes more sophisticated.

As for its long-run effects. Professor Salop highlighted that the “no economic sense” test could lead to legalizing practices that provide an immediate benefit to consumers, such as improving a product, but eliminate competition over the long term, for instance, by removing product compatibility.[98] This would have the effect of increasing prices and thus harming consumers.[99] Such an example, however, assumes that the dominant firm is willing to reduce the usefulness, and therefore the value, of its product by removing its compatibility with other products. This hypothesis also presumes that the company enjoys an absolute monopoly power, because, otherwise, it would be safe to say that competition would actually push it towards compatibility with the aim of creating a network effect.[100] Plus, in the absence of a monopoly power, eliminating compatibility could lead consumers to adopt competing products.[101]

This hypothesis assumes, furthermore, that the dominant firm is present on both the upstream market, concerned by the changes, and a downstream market, where compatible products are. If this is not the case, removing compatibility would be illogical.[102] The example presumes, also, that the market conditions will stay unchanged. It also excludes dynamic efficiency considerations. Indeed, it is required for the dominant firm to know that its market shares will remain at a constant level, otherwise, it may not recoup its losses.[103] With high-tech markets, where market shares are evolving very quickly, it seems unlikely that companies would take such a risk. Disruptive technologies appear abruptly and create new markets, which very often eliminate all possibilities for “locking” the consumer into a market that no longer exists.[104] This example presumes, lastly, that it is better for consumers to have compatible products of lower quality than incompatible products with a higher quality. These different assumptions put together tend to prove the ineffectiveness of this criticism.

As for empirical evidence. Part of the scholarship on this topic notes that the “no economic sense” test excludes empirical evidence on the effects of practices.[105] It seems, conversely, that it allows a fair balance between legal certainty, on the one hand, and the need to consider empirical evidence in order to improve the analysis on the other.[106] If new empirical evidence shows that a practice, previously legal, is in fact having anti-competitive effects, the “no economic sense” test will lead to condemn it. The contrary is true as well. A practice seen as illegal for years may become permissible if new economic evidence is adduced by the company.

C. How to Improve the “No Economic Sense” Test

We have shown that the “no economic sense” test is particularly efficient for analyzing non-price strategies such as predatory innovation. And yet, as explained, applying the test as it was originally designed implies a trade-off: creating some type-II errors in exchange for legal certainty. Although this could be defended, this article intends to demonstrate how this trade-off may be avoided by applying an improved version of the “no economic sense” test, which would maintain legal certainty while avoiding any type of legal errors.

The “no economic sense” test might be improved, but it is important not to create a new version that would have high implementation costs. While keeping this objective in mind, we propose an “enhanced version of the ‘no economic sense’ test,” which answers criticisms related to the underinclusivity of the test and potential difficulties with applying the test when pro and anti-competitive modifications coexist.[107] Two situations may thus be distinguished, one in which product changes are separable from one another and one in which the changes cannot be isolated.

When the changes may be separated from each other. Unlike innovations in the traditional sectors of the economy, innovations developed in high-tech markets often have the advantage of being readable[108] to the extent that it is possible to analyze each line of the source code.[109] Whether innovations introduced in high-tech sectors intend to create a new software structure, to add new features to a product, to facilitate product use, or to increase its security, most of these objectives are achieved by the means of one or more lines of code.[110] The purpose of each of these lines is clearly identifiable.[111]

The software (or product) as a whole is thus distinguished from each of the features that can be adjusted individually.[112] It may also be distinguishable from its updates if the introduction of new software always benefits to the consumer,[113] some of its updates may play against its interest. Accordingly, some authors have raised[114] the point that predatory innovation is more easily identifiable on high-tech markets than others.[115]

Consequently, the “no economic sense” test may be improved by identifying the purpose of every update of a product, and more specifically, each element of the update. Litigious situations where pro and anti-competitive effects are recorded simultaneously tend to disappear to the extent that each of these effects may be separated from the others. As a result, applying the “no economic sense” test is even more relevant. It is not for the judge to interfere with the company’s management and/or express disapproval with the strategic choices so to punish companies for not having implemented “less anti-competitive[116] practices. Rather, the judge is tasked with punishing actors that implement practices that could have been implemented without harming the consumers’ well-being.

Applying the enhanced version of the “no economic sense” test will have the following structure: in the first instance, the complainant will have to establish the injury he suffered, and in response, the defendant will try to justify each of the changes made to the product. The judge will guarantee the proper conduct of this analysis and will identify all product modifications that have a solely anti-competitive effect. This role is crucial. Consider the situation where a company decides to eliminate its products’ compatibility with those of its competitors. Imagine that the dominant firm is justifying this change by showing that the new version of its product allows geo-location. In this example, the company has an economic justification—adding a new function to its product—but because it is unrelated to the removal of compatibility, doing so is anti-competitive and should be condemned. The judge then has the responsibility to reject all economic justifications made in bad faith. When doing so, he ensures that companies cannot escape liability, as they might under the traditional “no economic sense” test, by presenting “ghost” justifications[117] unrelated to their practices.

Rules for eligibility of evidence in these inquiries can be inspired from the Daubert criteria,[118] which fit in line with the work of Karl Popper.[119] In its famous ruling, the Supreme Court held that pursuant to Rule 702 of the Federal Rules of Evidence governing expert testimony, only those using recognized “scientific method[s]” are to be taken into account. The court admitted the need to “filter” the scientific evidence produced in court. A similar filter should be implemented when applying the “no economic sense” test to invalid economic justifications, evidence or, expert reports that do not have a scientific value but have the sole purpose of obscuring the procedure.

When the changes are indivisible. Although the changes made to a technological product are generally separable from each other, that is not always the case. Let us imagine a situation in which a dominant firm decides to remove the compatibility of its products with the specific aim of increasing their safety. It would be possible to track what changes in the coding led to elimination of product compatibility. However, if the economic justification provided by the company—namely increasing products’ safety—is directly linked to the compatibility removal, the practice will be considered pro-competitive.[120] In this situation, a single line of code is added (or removed) to delete product compatibility, but it produces two consequences that cannot be separated. Imagine an alternative situation in which a company decides to increase the execution speed of its software by using Wi-Fi rather than Bluetooth. In addition, suppose that, for security purposes, all compatible devices use Bluetooth rather than Wi-Fi. Once again, the judge could easily track which lines of code have enabled the addition of new functionality, on the one hand, and the removal of product compatibility with Bluetooth on the other. Yet, the practice must be deemed pro-competitive because increasing the speed of the product is a valid economic justification caused by replacing the Bluetooth functionality. In this case, even though several lines of code are modified, they cannot be analyzed separately.

Accordingly, the judge must (1) ensure that only valid economic justifications are brought by the parties and (2) determine which modifications may be separated from each other.[121] These two steps are the keystones of a decision process free of judicial errors. This reasoning seeks to encourage investments in the short term and allow continued sophistication of antitrust law by better matching business justifications. In order words, it is about creating the smallest safe harbor possible—when practices cannot be separated—so that antitrust law is effective. It allows, at the same time, a drastic increase in the level of legal certainty by providing an understandable legal framework.

D. Modeling of the Proposed Test

The enhanced version of the “no economic sense” test is comprised of four steps. Together, the steps ensure a legal analysis that detects predatory practices as precisely as possible.

Step 1: Does the practice implemented by the dominant company tend to reduce or eliminate competition? If the answer is negative, the practice is deemed to be legal. If the answer is positive, the analysis moves on the second step.

Step 2: Does the practice provide a benefit to the dominant firm solely because of its tendency to reduce or eliminate competition?[122] If the answer is positive, the practice must be condemned. If the answer is negative, the analysis moves on the third step.

Step 3: If the judge suspects that some of the effects created by the practice are pro and anti-competitive, he must determine whether it is possible to distinguish between the modifications made to the product and their economic justifications. If the answer is negative, the practice is deemed to be legal as a whole. If the answer is positive, the analysis moves on the last step.

Step 4: Modifications that make economic sense for reasons that are not anti-competitive must be allowed, while modifications that only tend to reduce or eliminate competition must be condemned. The penalties should be proportional to the intensity/number of anti-competitive practices.

This test, by avoiding judicial errors, ensures short-term efficiency by raising the level of legal certainty for companies while eliminating practices that, without any doubt, are predatory. Moreover, allowing courts to analyze practices related to product modifications will have long-term effect of improving their expertise. Finally, it must be noted that some practices deemed to be legal will be proved to be anti-competitive in a near future. The opposite is true as well.

This graphical representation summarizes the 4 steps detailed beforehand.[123]



Now that the enhanced version of the “no economic sense” test has been explained, it will be applied to the major cases dealing with predatory innovation so as to demonstrate its efficiency. The same methodology is used for each case. We begin with a brief summary of the facts, then we explain the test applied by the court and the court’s decision.[124] We then apply the new test to the facts and discuss alternatives to the courts’ conclusions.[125]

A. The Berkey Photo v. Eastman Kodak case (1979)[126]

Facts. In the late 1960s, Kodak dominated the cameras and compatible products markets. In 1972, the company decided to introduce a new system, the “110 Instamatic,” as well as a new device, the “Kodacolor II film.” The new device was smaller and simpler to use than the previous ones. It was also incompatible with the products of one of its competitors in an ancillary market. Berkey brought suit against Kodak for having illegally removed the compatibility of its products.

The test applied by the court. The Second Circuit applied a “reasonableness” test[127] and centered its analysis on the fact that a single improvement could justify all modifications made to the product.

The solution. First, the Second Circuit held that an “innovation” may be prohibited by antitrust law if it is proved to be anti-competitive.[128] The court then noted that the new camera introduced by Kodak, the Kodacolor II film, had several lower quality features than its former model. In particular, the Court found that its autonomy was shorter[129] and that it generated more “red eye” in photos. The new device, however, had a better grain[130] and was smaller. The judges then held that comparing the features of the two devices was not a probative element given the different characteristics of the two, which were performing better in different areas.[131] The Court held that Kodak was not liable.[132]

Application of the enhanced version of “no economic sense” test. The Court’s holding justified a set of practices on the ground that one of them was pro-competitive.[133] But, as we have discussed, whenever it is possible to distinguish between the modifications made to a product and their justifications, the judge must condemn the anti-competitive ones. In the present case, Kodak—which initially asked for a per se legality to be applied—argued that the few improvements made to its product justified all the modifications, including those that may have anti-competitive aspects.[134] Berkey argued, on the contrary, that Kodak had introduced a less efficient camera for the sole reason that it would no longer be compatible with Berkey products, which justified holding them liable.[135] The Court found that even though some of the new features of the Kodacolor II were inferior to those of Kodacolor X, they did not translate into anti-competitive strategies in themselves because they could have reduced the attractiveness of the new camera to consumers without directly affecting competition. Berkey’s argument based on the characteristics of the Kodacolor II film was accordingly deficient.[136] In addition, Berkey failed to prove that Kodak could have marketed a smaller camera while improving all characteristics of the earlier version. Nevertheless, one must ask whether removing the new camera compatibility with Berkey’s products was a necessity in order to achieve its goal. In the present case, because Kodak did not demonstrate the causal link between the new design of its device and the need to remove it, it arguably could have been held liable.

Applying the enhanced version of the “no economic sense” test would have resulted in a different outcome from the one found by the Second Circuit. Kodak could have been held liable for having removed the compatibility of its new camera—if it was an anti-competitive strategy, which the Court did not address. Legal certainty would have been enhanced by a court decision that provides clarity and predictability to all companies on the market, which would, in turn, minimize impeding innovation by a type-II error. Not to mention, it would have also benefited consumers.

B. The North American Microsoft case (2001)[137]

The approach adopted by European and North American courts regarding predatory innovation practices differ on several points. Notably, European judges extend the essential facilities doctrine to high-tech markets while North American courts refuse to do.[138] The Microsoft case illustrates this fundamental distinction. This case also remains one of the pillars of predatory innovation doctrine. It is, in fact, the first case in which a court studies software encoding in such detail.[139]

Facts. The Microsoft case was the subject of a long series of jurisprudence, which ended in June 2001 (with “Microsoft III”) with a decision of the United States Court of Appeals for the District of Columbia.[140] One of the practices considered[141] was the way Microsoft integrated Internet Explorer into its operating system. In doing so, the company:

• deleted the function allowing one to remove the software from the operating system (practice n1);

• designed the operating system in order to override the user’s choice to use a different browser (practice n2);

• designed the operating system so that when certain files related to Internet Explorer were removed, bugs appeared (practice n3).

The plaintiffs stressed that Microsoft had set up several anti-competitive practices to eliminate Netscape, which was a browser that provided some basic functions similar to that of an operating system.[142] Moreover, the complainants denounced the fact that Microsoft had developed a Java script incompatible with Sun Microsystem’s products.[143]

The test applied by the court. This decision is one of the first related to high-tech markets to have considered the balancing test,[144] even though the court did not apply it. The burden of proof was initially on the plaintiff to demonstrate the anti-competitive effects of the practices.[145] The defendant then had to demonstrate that the pro-competitive effects prevail over anti-competitive ones.[146] In theory, if the defendant did so, the plaintiff then had to prove that the defendant was incorrect. This balancing theoretically ended once a party showed conclusively that the dominant effects were pro or anti-competitive. In the present case, even though the court was “very skeptical about claims that competition has been harmed by a dominant firm’s product design changes,”[147] it opined that applying a per se legality to such practices was inappropriate because they may have various effects on competition.[148]

The solution regarding Internet Explorer.[149] The Court held that practices n1 and n3 were illegal.[150] On the other hand, it found that the practice n2 was pro-competitive. The court ruled that the “technical reasons” provided by Microsoft were exculpatory considering that the plaintiffs failed to demonstrate a greater anti-competitive effect.[151] The Court also ruled that a company’s technical justifications may not be sufficient to establish the legality of a practice. This implies that a genuine technical innovation may be considered anti-competitive if it causes great damage to the competitive process. This reasoning was not applied in this case because the plaintiffs failed to demonstrate any anti-competitive effect,[152] but it did challenge the claim that antitrust law protects innovation.[153] The consequences in terms of disruptive innovation are particularly dangerous because they create a destructive effect on existing markets. In short, a sole pro-competitive effect was proven in practice n2, but only anti-competitive effects were argued for practices n1 and n3. As a consequence, the judges did not conduct a balancing of any of the effects[154] and the decision does not illustrate the practical application of the balancing test.[155]

The solution regarding Java. The plaintiffs also challenged the practices implemented by Microsoft in the development of Java for Windows. They complained, in particular, that Windows Java was incompatible with Sun Microsystems’ products.[156] Windows, in opposition, argued that its own Java should be allowed because of its superiority to Sun’s Java.[157] The Court found that a company’s dominant position does not prohibit it from developing products that are incompatible with those of its competitors,[158] and thus, the court justified Microsoft Java’s incompatibility with Sun because of its greater power and speed.[159]

Application of the enhanced version of “no economic sense” test. The Microsoft decision suffers from numerous flaws that would have been eliminated by applying the enhanced version of the “no economic sense” test.[160] According to the information available in the decision, Microsoft gave no pro-competitive justification for practices n1 and n3.[161] It is unclear whether Microsoft gave justifications that were rejected earlier in the procedure, but in the absence of any, these practices made sense solely because of their tendency to reduce or eliminate competition. A fine should have then been imposed.

Practice n2 was justified by Microsoft on the grounds that using Netscape’s browser prevented the use of the “ActiveX” which allowed the proper functioning of “Windows 98 Help” and “Windows Update.”[162] Microsoft also justified the forced use of Internet Explorer by explaining that when the user started Internet Explorer from “My Computer” or “Windows Explorer,” a different browser would not have enabled them to keep the same window.[163] Surprisingly, the court did not analyze the way Windows operated in more detail. Indeed, if it had been shown that the use of ActiveX was prevented when using another browser because of Microsoft’s anti-competitive desire, or, in other words, that Microsoft could have allowed Active X even with another browser, the company’s technical justification would have been nullified. Similarly, if it had been shown that another browser would have created the same ease of use with “My Computer” and “Windows Explorer,” the holding that only Internet Explorer could do it would have been refuted. In the absence of additional information, it is impossible to say whether the practice implemented by Microsoft was pro or anti-competitive. The lack of conviction in this case tends to minimize the poor reasoning led by judges, but a more detailed decision would have increased legal certainty, and ultimately would have promoted innovation.

Regarding the practices related to Java, the Court found that because Microsoft’s Java was more powerful than Sun’s, its design was de facto justified.[164] The dominance Microsoft enjoyed at the time did not create any duty to design compatible products with its competitors. It should be emphasized, however, that a different outcome would probably have been reached in the European system because of the principle of “special responsibility” for dominant firms. In any event, the application of the enhanced version of the “no economic sense” test would have resulted in an acquittal on behalf of practice n2. The application of that test would have also increased legal certainty by providing businesses a more comprehensive grid of analysis than the enigmatic one given by the Court.

C. The European Microsoft Case (2004)[165]

The Microsoft case, along with the Google case,[166] remains to this day the most iconic case in terms of abuse of dominant position. In addition to its importance regarding penalties, it is the first decision in which the Commission found that network effects could be used to strengthen the barriers to entry in high-tech markets.[167] Unlike the North American decision, the European Commission did not analyze the way[168] Microsoft integrated its software (here a media player) into its operating system. The North American judges condemned the company because the integration of Internet Explorer came along with micro-practices possessing anti-competitive effects that were distinguishable from the integration. The European decision, in contrast, disputed the integration of Windows Media Player in itself.[169]

Facts. On December 10, 1998, Sun filed a complaint against Microsoft on the grounds that Microsoft had refused to provide information allowing for the interoperability of Sun’s products with Microsoft’s PC.[170] In February 2000, the European Commission launched an investigation against Microsoft.[171] A second statement of objections against Microsoft involved interoperability and the integration of Windows Media Player within its operating system.[172] And yet another statement of objections was sent to Microsoft in 2003 following a market survey.[173]

The test applied by the court. Above all else, it should be underscored that the European Commission rejected Microsoft’s argument that antitrust law could not be applied to the New Economy.[174] The Commission pointed out that “the specific characteristics of the market in question (for example, network effects and the applications barrier to entry) would rather suggest that there is an increased likelihood of positions of entrenched market power, compared to certain ‘traditional industries’,” and required a strict application of antitrust law.[175]

In fact, the European Commission seemed to apply a similar test to the one used by the North American Federal Court, but it is difficult to say if the Commission applied a “balancing” or a “disproportionality” test[176] given that the term “proportionality” does not appear in the decision.[177] But the similarities stop here. The North American and European procedures did not use the same semantic field[178] and practically showed a will to reach different outcomes. The European Commission multiplied the references to “the prejudice of consumers,”[179] “network effects,”[180] and “interoperability,”[181] while the Department of Justice reported the “predatory”[182] nature of the practices and the need to protect “innovation.”[183] The European Commission indirectly intended to ensure consumer protection by enabling products interoperability while the Department of Justice developed a broader view through a defense of innovation that aimed not to create type-I errors.[184] Most of the European Commission’s decision is devoted to the rules of “tying,”[185] which highlights the amalgam between this notion and the one of predatory innovation.[186] It also shows why “tying” does not cover all of the issues related to predatory innovation.[187] The Commission found that Microsoft had not put forward any efficiency gain that would justify the integration of its Windows Media Player into the operating system.[188] It held, however, that anti-competitive effects may have outweighed efficiency, at least in theory.[189]

The solution. The European Commission sanctioned Microsoft for numerous antitrust violations, which were confirmed by the General Court.[190] In analyzing Microsoft’s integration of the Windows Media Player (“WMP”) into its operating system, the Commission began by observing that there were no technical means to uninstall the player.[191] This observation led the Commission to analyze whether the interdependence of Windows Media Player with the operating system was necessary, thereby avoiding one of the main pitfalls of the American decision.[192] Microsoft replied, “if WMP were removed, other parts of the operating system and third party products that rely on WMP would not function properly, or at all.”[193] Here, Microsoft intended to prove the absence of predatory innovation, which deserved to be discussed. Unfortunately, the European Commission did not answer the argument.

In fact, the European Commission did not seem to address this issue because it was described under the broader label of tying. Instead of addressing the issue, the judges were almost exclusively concerned by the fact that Microsoft did not market a version of Windows without Windows Media Player.[194] But the practice in question concerned the functionality of the operating system. And the notion of tying was unfit to be applied because there was a technical reason to combine the products into one.[195]

Here lies one major criticism[196] one can make regarding European Commission decision: by refusing to analyze the issue of predatory innovation, the Commission deprived its decision of a legal basis for examining all of the anti-competitive effects identified by the complainant. The European Commission also noted that “it can be left open whether it would have been possible to follow Microsoft’s above line of argumentation had Microsoft demonstrated that tying of WMP was an indispensable condition for simplifying the work of applications developers.”[197] It was added that “Microsoft has failed to supply evidence that tying of WMP is indispensable for the alleged pro-competitive effects to come into effect,”[198] thus concluding that it was “technically possible for Microsoft to have Windows handle the absence of multimedia capabilities caused by code removal (and the resulting effect on any interdependencies) in a way that does not lead to the breakdown of operating system functionality.”[199] In this respect, the Commission held that Microsoft had not provided tangible evidence to support their argument that the integration of Windows Media Player simplified the work of application designers.[200] Conversely, the Commission did not underline a discussion regarding whether the integration of Windows Media Player to the operating system had allowed a more complete experience of Windows. Nevertheless, it concluded “it is appropriate to differentiate between technical dependencies which would by definition lead to the non-functioning of the operating system and functional dependencies which can be dealt with ‘gracefully.’”[201] In other words, Microsoft could have kept the other functions of the operating system but all functions directly or indirectly related to the use of such a player would have been inoperable without it or another player.[202] The distinction between technical and functional interoperability held by the European Commission is, in fact, essential. It indicates a difference between the interoperability necessary to the functioning of a product on the one hand and allowing a new feature to work properly on the other.

The Commission imposed a €497 million fine on Microsoft for its practices.[203] In a press release related to the United States’ investigation of the same matter, the Department of Justice called the sizable fine regrettable, especially because unilateral competitive conduct is “the most ambiguous and controversial area of antitrust” law.[204] The Commission also required Microsoft to sell a version of its operating system free of its Windows Media Player.[205] Lastly, the Commission ordered the company to disclose information that it had refused to previously provide for the development of products compatible with its own.[206]

Application of the enhanced version of “no economic sense” test. To the extent that the integration of Windows Media Player had strong pro-competitive effects—such as providing cost and time savings to consumers—applying the proposed test would have necessarily led to not condemning Microsoft. The Microsoft decision is therefore a type-I error. It would have been helpful for the Commission to analyze the anti-competitive aspects separate from pro-competitive aspects, but the Commission did not, demonstrating how European courts and authorities deal improperly with predatory innovation.[207] Analyzed correctly, the enhanced version of the “no economic sense” test shows that it was not possible to dissociate the integration of Windows Media Player from enhanced consumer well-being. Forcing the consumer to download another media player carried a direct injury to the consumer in itself. Further, Windows Media Player was not exclusive. In other words, the integration of WMP made economic sense to Microsoft not only because of the potential anti-competitive effects, but also because it improved the user experience. Accordingly, the European Commission made a critical error by holding a real innovation anti-competitive. The fact is that it is not for judges to order functions’ removal if they benefit consumers. In such a world, the door is open for competition authorities to prevent all innovations producing a low anti-competitive effect despite having great pro-competitive effects.

D. The iPod iTunes Litigation

The iPod iTunes Litigation is the most recent case to have drawn a great deal of attention to a practice of predatory innovation. It is particularly important because it took place several years after the Microsoft and Intel cases, both of which had been influential for analyzing practices in the high-tech sector.[208] Yet, to the best of our knowledge, the iPod iTunes Litigation has never been examined closely.

Facts. On November 10, 2001, Apple introduced its portable music player, called the iPod.[209] One of its competitors, RealNetworks, analyzed the iPod’s software and managed to extract the code created by Apple for listening to downloaded files on the iPod. RealNetworks inserted this code in all MP3s sold in the RealNetworks Music Store. In 2001, Apple introduced iTunes, free software allowing users to manage audio files on the iPod. On April 28, 2003, Apple also introduced the iTunes Music Store (“iTMS”), an online store enabling direct music purchase.[210] Apple managed to modify the format of the regular audio files purchased on the iTMS by introducing a digital rights management (“DRM”) to restrict the use of regular AAC (“Advanced Audio Coding”) files.[211] This format was referred as “AAC Protected” and iTunes used a feature called FairPlay, which allowed Apple to manage this DRM. Apple also changed the iPod internal software so as to allow the proper reading of these “AAC Protected” files.[212] In July 2004, RealNetworks introduced the new version of its RealPlayer. It included a feature called Harmony that sought to imitate FairPlay compatibility and enable audio files purchased from the RealNetworks online store to be played on the iPod. In October 2014, Apple decided to release the version 4.7 of iTunes.[213] This version changed the FairPlay encryption method, removing any compatibility between Harmony and iTunes.[214] RealNetworks then restored this compatibility.[215]

A first complaint was introduced on January 3, 2005, denouncing the anti-competitive strategy implemented by Apple in order to eliminate competition on the online market for selling music.[216] On February 28, 2005, the plaintiffs also denounced the modification made by Apple of a free audio format, the AAC, into a protected audio format, the “AAC Protected.” They underlined Apple’s intention to exclude competitors from the market through an anti-competitive strategy[217]—predatory innovation. The applicants pointed out that the audio files purchased from the iTMS became incompatible with other audio players. Also, any files purchased from a store other than iTMS became incompatible with the iPod. Lastly, the complainants alleged that Apple was unlawfully tying by requiring the purchase of an iPod in order to listen to the files bought on the iTMS and by forcing users to purchase songs on the iTMS in order to be able to use the iPod.[218] On March 7, 2005, Apple responded to the complaint pointing out, in the company’s opinion, the inaccuracy of the complainants’ allegations.[219] Apple stressed that listening to the files bought on the iTMS using other music players remained possible by “burning” them into a CD-ROM and then by extracting the files from the CD on a computer to delete the protection.[220] As for tying, Apple underlined that a practice may not be condemned as such if, by buying the two products separately, it would be so expensive that no consumer would do so.[221] In September 2006, Apple released version 7.0 of iTunes, which, aside from introducing new features, once again deleted iTunes’ compatibility with Harmony.[222]

The test applied by the court regarding iTunes 4.7. Following a long process, the United States District Court for the Northern District of California was in charge of determining the legality of iTunes 4.7. Quoting Allied v. Tyco, the judges pointed out that the District Court has ruled that “there is not a per se rule barring Section 2 liability on patented product innovation.”[223] They also held that balancing the pro and anti-competitive effects of a new product was rejected in Allied.[224] The judges applied the same reasoning, holding that when a real improvement is shown, antitrust law should not condemn the modification.[225]

The solution found by the court regarding iTunes 4.7. Apple argued that the introduction of this new version of iTunes was motivated by the necessity to improve its security through the strengthening of anti-piracy protections.[226] Apple stressed, in particular, that (1) the earlier version of the software had previously been pirated, (2) that the proliferation of computer attacks that occurred at the beginning of 2004 led some music labels to require Apple to take corrective action, and, (3) that they changed the encryption method in accordance with the contractual provisions sent by music labels.[227] RealNetworks underlined that Apple’s true intention was to remove the compatibility with the audio files purchased from its online store.[228] RealNetworks also stressed that Apple began developing a new FairPlay a month after it refused to grant RealNetworks a license, proving Apple’s anti-competitive intention. RealNetworks said it had concluded numerous deals with music labels, which threatened Apple’s position on the market for selling online music.[229] The company also emphasized how it had increased its market shares after launching Harmony, whereas Apple’s market share had, for the first time, dropped below 70%.[230] RealNetworks lastly argued that Apple showed its anti-competitive intention by threatening to remove any compatibility with Harmony.[231]

The Court ultimately rejected RealNetworks’ arguments. The court ruled that the introduction of iTunes 4.7 was a real improvement that could not be condemned. In particular, they emphasized that the expert appointed by RealNetworks reported himself that the new version of FairPlay was indeed harder to hack, significantly increasing its safety.[232] The court also underlined that no other practices implemented by Apple could have been challenged under antitrust law.[233] This precision is particularly interesting because it seems to recognize the possibility of dissociating the various technical changes made by Apple, on one side, and the pro-competitive modification related to FairPlay security, on the other. But in fact, the Court merely analyzed whether Apple had breached antitrust law, without examining the technical aspects of the product in too much detail. This decision, thus, did not apply the enhanced version of the “no economic sense” test.

Application of the enhanced version of “no economic sense” test regarding iTunes 4.7. The court held that the practice implemented by Apple had the effect of reducing competition in the market for online music sales.[234] Yet, it is necessary to consider the possibility of distinguishing the improvement in terms of FairPlay’s safety from the eventual anti-competitive effects, namely, the removal of compatibility. Unfortunately, it is not possible to process this analysis as relevant expert testimonies are still covered by business confidentiality.[235] Accordingly, the merits of the decision adopted by the Northern District of California cannot be denied nor confirmed.

The litigation regarding iTunes 7.0[236] is of the same nature, though more complex. It occurred based on the fact that, in 2006, Apple introduced iTunes 7.0. Complainants highlighted similar problems to the 4.7 version.

The test applied by the court regarding iTunes 7.0. A jury was asked to evaluate the modifications made to iTunes 7.0.[237] The process that led to the question asked to the jury is to be analyzed carefully because its formulation significantly influenced the final outcome. Two points of disagreement arose between the parties. The first was related to the possibility of separating the practices from one another. The second concerned taking into account subjective intent, because even though the practices were seen as an indivisible whole, they could have been considered anti-competitive.

Regarding the modifications’ separability. In a November 18, 2014 document, which was jointly submitted by the two parties in order to define the legal issues, Apple’s counsel crystallized the case around the following questioning: was iTunes 7.0 a real improvement?[238] Apple’s counsel sought to apply the test set out in Allied Orthopedic, according to which if a product is improved, the modifications are deemed to be legal.[239] Accordingly, the new version of a product is either an improvement or a strategy seeking to eliminate competition.[240] The plaintiff noted that the way the question was formulated favored Apple.[241] They emphasized that evaluating whether the new version of iTunes included improvements was not sufficient.[242] They underlined, notably, that the changes made to the KeyBag Verification Code (“KVC”) and the Database Verification Code (“DVC”) should have been brought to the jury’s attention as being independent practices.[243] Conversely, Apple argued for these amendments to be considered as a whole along with the iTunes 7.0 improvements.[244] And indeed, Apple stressed that according to Allied v. Tyco, all the changes were to be considered as an indivisible whole,[245] and that anyway, the practices could not have been differentiated in practice. Apple also gave two technical justifications for its actions: an increase in security, as well as strengthening against product corruption.[246] Unsurprisingly, in the hearing held on December 1, 2014, plaintiff’s counsel argued that they needed to ask the jury separately about the different product modifications.[247] The Court decided mentioning coding issues to the jury would be confusing.[248] All product modifications were then presented as an indivisible whole.[249] In this regard, it should be stressed that avoiding legal errors cannot be done when the essence of the issues are not submitted to the discretion of the courts and juries. Moreover, even by assuming that mentioning coding would have confused the jury, nothing actually prevented the court from analyzing the coding beforehand and then submitting an adapted question to the jury.

Regarding subjective intent. The plaintiffs stressed that according to Allied Orthopedic, a company that has improved its product but also voluntarily created substantial anti-competitive effects should be condemned.[250] They alleged that the improvements made on iTunes had anti-competitive purposes[251] and asked for the improvements to be balanced with the anti-competitive intent. They suggested that the jury follow a two-stage approach, analyzing whether the improvements were real, and if so, whether Apple had been driven by anti-competitive intent.[252] Apple stressed that the intention did not matter considering the fact that competition by innovation is about harming competitors by introducing a new product.[253] Apple claimed that, as long as some improvements to the product were shown, it could not have been sanctioned regardless of other considerations.[254] The judges took Apple’s side, stressing that product improvements were not to be balanced with anti-competitive intent. We subscribe to this analysis.

The solution found by the court regarding iTunes 7.0. The jury was asked the following question: “Were the firmware and software updates in iTunes 7.0, which were contained in stipulated models of iPods, genuine product improvements?”[255] and answered in the positive, saying that the update of iTunes 7.0 was a real improvement, and, therefore, pro-competitive.[256] Because of the question’s wording, the modifications made to the DVC and KVC codes were not properly analyzed.

Application of the enhanced version of the “No Economic Sense” Test regarding iTunes 7.0. Both claims made by the parties are supported by precedents.[257] In fact, the diversity of the tests chosen by the various jurisdictions throughout the years and the lack of standardization of these tests have had the effect of creating a very unclear jurisprudence, resulting in hard-to-understand litigation. On the merits, Apple’s argument, according to which all modifications were indistinguishable, is not convincing. The link between allowing videos to be played on iPods and the need to enhance security is not particularly obvious, but judging whether other changes[258] did necessitate eliminating compatibility should have been performed.[259] Unfortunately, such an analysis may not be conducted in great detail as the expert testimonies are still covered by business confidentiality.[260] Let us simply underline that not analyzing each modification separately raises the possibility that a type-II error was actually pronounced. Unfortunately, the necessary information to conduct a deep analysis is not available.


This paper has examined the ENES test from a theoretical and empirical perspective. The main conclusion is that the test would improve the quality of the law for analyzing non-price strategies, which is greatly needed. More than a simple adjustment of the existing rules, it requires a new and standardized approach to these practices.

Conclusions drawn from the test cases. We have shown that the ENES test leans toward the conviction of practices that were deemed to be legal by the courts, like Berkey Photo v. Eastman Kodak. It also leads to exonerate Microsoft in the European case, contrary to what was done in the European Commission’s ruling. The same goes for CR Bard in its litigation. Lastly, applying this new test results in similar conclusions in four cases, the main difference being only that the ENES test would have greatly increased the level of legal certainty. A table in the appendix illustrates its effectiveness on a wider range of cases.

General contribution. We have demonstrated, in short, how the ENES test results in the creation of a uniform rule of law, which ultimately increases consumer welfare by encouraging companies to keep innovating. Consumer well-being is also improved by the elimination of anti-competitive strategies. As a matter of fact, the proposed test is easier to implement than most other tests, and yet, it limits legal errors more efficiently than others. Its quasi-mathematical aspect leads to a better understanding of the rule of law and it must, therefore, be implemented in all cases related to predatory innovation and other non-price strategies.

Appendix #1 – A reassessment of the major cases related to predatory innovation

A reassessment of the major cases related to predatory innovation

Name & date of the case

IBM (1979)

Berkey Photo v. Eastman Kodak (1979)

CR Bard v. M3 Systems (1998)

United States v. Microsoft Corp US (2001)

European Commission v. Microsoft Corp EU (2004)

Outcome found by the court

No conviction: The product modification is not “unreasonably anti-competitive

No conviction: Comparing the quality of two devices is not a conclusive evidence

Conviction: the new product is easier to use, but “the real reason” of the modification is anti-competitive

Conviction: Deleting the possibility to remove a software from the operating system (practice n1) and programing the operating system so to bug when certain files related to Internet Explorer are deleted (practice n3) has no pro-competitive justification

No conviction: Programing the operating system so to override the consumer choice to use another software than Internet Explorer (practice n2) is technically justified by Microsoft

No conviction: Windows’s Java is more efficient than the Sun’s Java

Conviction: The operating system may have properly functioned even in the absence of Windows Media Player

Application of the enhanced version of “no economic sense” test

No conviction: The improvements may not be separated from the anti-competitive effects

Conviction: Removing compatibility is unrelated from improving the camera

No conviction: Improving the needle system may not be done without removing compatibility

Conviction: Practices n1 and n3 made economic sense only because they produced an anti-competitive effect

Inability to judge: No information is available on the separability of the improvement with the compatibility removal

No conviction: Microsoft’s Java is better and Microsoft’s had no duty whatsoever to ensure the compatibility of new products with those of its competitors

No conviction: The integration of WMP to the operating system is not anti-competitive in itself

Name & date of the case

HDC Medical v. Minntech Corporation (2007)

Intel US (2010)

Allied c. Tyco (2010)

iPod iTunes litigation (2014)

Outcome found by the court

No conviction: Minntech provided an economic justification for its product modification that HDC could not prove to be false

Mutual agreement: Intel has agreed to amend its practices for the future

No conviction: The new design is an improvement establishing the superiority of the new product

No conviction: iTunes 4.7: this version of iTunes is a real innovation in that it increases the security of the software

No conviction: iTunes 7.0: This version of iTunes is also a real innovation in that it increases the security of the software

Application of the enhanced version of “no economic sense” test

No conviction: removing the product compatibility is inseparable from the improvement made to the product

Inability to judge: lack of information on the possibility to distinguish between the improvement and the deleting of compatibility

No conviction: Removing the product compatibility is the reason explaining the improvement

Inability to judge: The documents allowing to analyze whether changes made to iTunes 4.7 were justified for technical reasons are sealed

Inability to judge: The documents allowing to analyze whether changes made to iTunes 7.0 were justified for technical reasons are sealed

* Thibault Schrepel (Ph.D., LL.M.) is a research associate at the Sorbonne-Business & Finance Institute, University of Paris I Pantheon-Sorbonne. He leads the “innovation” team of Rules for Growth and is the Revue Concurrentialiste creator. His personal website is Your comments on this article are more than welcome ( I would like to thank the editors of the NYU Journal of Intellectual Property & Entertainment Law for their consstructive editing.

[1] Type-I errors, also called “false positives,” occur when a court—or a competition authority—wrongly condemns a company for having implemented practices which, in fact, are not anti-competitive. On the contrary, type II errors, also known as “false negatives,” occur in the absence of condemnation of a practice that is anti-competitive and should therefore have been condemned.

[2] For an overview, see Michael L. Katz & Howard A. Shelanski, “Schumpeterian” Competition and Antitrust Policy in High-Tech Markets, 14 Cᴏᴍᴘᴇᴛɪᴛɪᴏɴ 47 (2005).

[3] OECD Policy Roundtables, Predatory Foreclosure, DAF/COMP(2005)14, at 48-59.

[4] Predatory innovation, which we previously identified as being one of the key issues in terms of high-tech markets, illustrates our point. See Thibault Schrepel, Predatory Innovation: The Definite Need for Legal Recognition, SMU Sᴄɪ. & Tᴇᴄʜ. L. Rᴇᴠ. (forthcoming 2018); see also, Thibault Schrepel, Predatory innovation: A response to Suzanne Van Arsdale & Cody Venzke, Hᴀʀᴠ. J.L. & Tᴇᴄʜ. Dig. (2017),

[5] OECD Policy Roundtables, Competition on the Merits, DAF/COMP(2005)27, at 23.

[6] Id.

[7] Id.

[8] Case COMP/C-3/37.792—Sun Microsystems, Inc. v. Microsoft Corp., Comm’n Decision (Apr. 21, 2004), available at [hereinafter “Microsoft Decision”].

[9] The OECD has recently devoted several roundtables to the subject. See OECD Policy Roundtables, Algorithms and Collusion, DAF/COMP(2017)4; see also OECD Policy Roundtables, Big Data: Bringing Competition Policy to The Digital Era, DAF/COMP(2016)14. Most of the world-leading competition authorities have contributed to them too.

[10] See generally Nicholas S. Smith, Innovative Breakthrough or Monopoly Bullying? Determining Antitrust Liability of Dominant Firms in Exclusionary Product Redesign Cases, 84 Tᴇᴍᴘ. L. Rᴇᴠ. 995 (2012) (explaining antitrust law objectives).

[11] Id. at 1016.

[12] As a reminder, type-I errors, also called “false positives,” occur when a court—or a competition authority—wrongly condemns a company for having implemented practices which, in fact, are not anti-competitive.

[13] See Smith, supra note 10, at 1018.

[14] See John B. Kirkwood & Robert H. Lande, The Fundamental Goal of Antitrust: Protecting Consumers, Not Increasing Efficiency, 84 Nᴏᴛʀᴇ Dᴀᴍᴇ L. Rᴇᴠ. 191 (2008); see also Nicolas Petit, European Competition Law, 143 (Montchrestien, 2012) (text in French) (“the European competition law seems to have decided in favor of this theory.”).

[15] We argued that the Neo-Chicago School should seek that goal. See Thibault Schrepel, Applying the Neo-Chicago School’s Framework To High-Tech Markets, Rᴇᴠᴜᴇ Cᴏɴᴄᴜʀʀᴇɴᴛɪᴀʟɪsᴛᴇ (May 6, 2016), Two authors further developed the premises of that school of thought. See David S. Evans & A. Jorge Padilla, Designing Antitrust Rules for Assessing Unilateral Practices: A Neo-Chicago Approach, 72 U. Cʜɪ. L. Rᴇᴠ. 27, 33 (2005); see also Thomas A. Lambert & Alden F. Abbott, Recognizing the Limits of Antitrust, 11 J. Cᴏᴍᴘ. L. & Eᴄᴏɴ. 791, 793 (2015) (“Neo-Chicagoans reason that ‘market self-regulation is often superior to government regulation . . .’”).

[16] See Herbert Hovenkamp, ​​Antitrust and Innovation: Where We Are and Where We Should Be Going, 77 Aɴᴛɪᴛʀᴜsᴛ L.J. 749, 751 (2011) (“[T]he gains to be had from innovation are larger than the gains from simple production and trading under constant technology.”).

[17] OECD Policy Roundtables, Two-Sided Markets, DAF/COMP(2009)20, 14 (“Firms sometimes use non-price strategies, such as exclusive contracts and product tying, to limit competition or foreclose the market to rivals. These practices have been at the centre of several important competition cases involving two-sided markets.”).

[18] Mark S. Popofsky, Defining Exclusionary Conduct: Section 2, the Rule of Reason, and the Unifying Principle Underlying Antitrust Rules, 73 Aɴᴛɪᴛʀᴜsᴛ L.J. 435, 477 (2006); see Frank H. Easterbrook, Cyberspace and the Law of the Horse, 1996 U. Cʜɪ. Legal F. 207, 215 (1996) (explaining the necessity to create a straightforward legal standard).

[19] OECD Policy Roundtables, supra note 5 , at 23.

[20] Id. This study of several hundred pages is extremely rich and remains the reference on the law.

[21] Id. at 23.

[22] See John Temple Lang, Comparing Microsoft and Google: The Concept of Exclusionary Abuse, 39 Wᴏʀʟᴅ Cᴏᴍᴘᴇᴛɪᴛɪᴏɴ & Eᴄᴏɴ. Rev. 5, 5 (2016).

[23] In terms of predatory innovation, an author already underlined in 1988 that all decisions dealing with the subject had little coherence, and that remains unchanged to this day. See Ross D. Petty, Antitrust and Innovation: Are Product Modifications Ever Predatory, 22 Sᴜffᴏʟᴋ U. L. Rᴇᴠ. 997, 1028 (1988) (“The decisions to date offer little guidance on how to distinguish a predatory product innovation, if such exists, from a legitimate innovation.”).