De Minimis Copying: An Empirical Study

Jessica Silbey* & Samantha Zyontz**

Download a PDF version of this article here.

In the internet age, the copyright de minimis defense has increased in relevance as copyright lawsuits (and IP generally) are more mainstream and infringement liability more widespread. This Article is the first empirical analysis of copyright de minimis defense cases, collecting and analyzing all such decisions since the mid-19^th century. It traces the doctrine’s development over the past century and its evolution in the digital era, when copying has become even more ubiquitous but its triviality remains widely disputed. The Article’s aim is not only to map the de minimis defense to learn more about it doctrinally—asking when is copying “too little” and how is that evaluated—but also asks the deeper, unresolved question about the harm copyright law aims to prevent and the benefits towards which it aims. In doing so, it proposes new rules for the de minimis defense designed to revive its appropriate function and filter out unmeritorious and inefficient copyright cases.

The Article begins with the hypothesis that, in the digital era, the copyright de minimis defense should be more relevant and more successful: as much as there is more copying on the internet, most of it is trivial. The data shows the first to be true (rise in relevance) but not necessarily the second (litigation success rates). Indeed, invocation of the de minimis defense has grown, but not its success in litigation, suggesting judicial skepticism toward the idea of trivial copying. That skepticism bodes poorly for the information age in which copying small bits of expression (or whole works momentarily) is essential for communication. The Article’s deep dive into the de minimis defense further explains how its evolution affects other aspects of copyright law, opening opportunities and exposing pitfalls in the context of strategic litigation. In the end, freedom of expression and the progress of science are at stake, and revitalization of the de minimis doctrine is a key to preserving these fundamental tenets of U.S. copyright law.

Introduction

De minimis non curat lex. The law does not concern itself with trifles.¹It is a centuries’ old legal defense across many areas of law and essential to efficient dispute resolution.² Despite its ubiquity and importance, little scholarship or theoretical analysis exists elucidating the nature of a “trifle” or the circumstances when the rule is best applied in specific legal fields.³ The challenge of studying the de minimis doctrine may arise because of the nature of the defense: in the face of an insubstantial injury, courts dismiss claims with little analysis or reasoning.⁴ What is there to learn from these terse decisions except that judges appear to know a trifle when they see one?⁵

In the age of artificial intelligence scooping up massive amounts of information, the meme-ification of digital communication on social media, and the ubiquity of copy/paste technology embedded in all of our electronic devices, the de minimis doctrine is (or should be) having a heyday, especially for copyright law. What counts as “too little” or “insignificant” copying to stall a frivolous lawsuit? The answer may be dramatically different from what counted as insignificant over a century ago when the de minimis doctrine entered copyright law. The de minimis doctrine is the difference between a lawsuit quickly dismissed (or never instigated) and one that costs millions of dollars to litigate. The line should be clearer, but it’s not. This Article investigates the changes in that line over a century, grounding the doctrine to its ancient roots in the common law.

De minimis injuries range from quantitatively small harms to those that are qualitatively insignificant.⁶ Intriguingly, some kinds of cases appear to be less susceptible to a de minimis defense, such as those concerning trespass and real property, and fiduciary law.⁷ Other kinds of claims, such as defamation and nuisance, more frequently elicit a de minimis defense.⁸ What makes one type of complaint de minimis and another worthy of adjudication within a specific field of law?

Within copyright law, plaintiffs launch time-consuming and costly legal claims presumably because they believe their copyright injuries deserve redress.⁹ And yet the de minimis defense succeeds because a court determines the injury is not cognizable. Frustrated copyright plaintiffs are told that despite defendant’s technical legal violations—an unauthorized copy was made—the court “does not concern itself with trifles.”¹⁰ At the heart of many cases resolved on de minimis grounds is a fundamental disagreement about the seriousness of the plaintiff’s harm and the importance of judicial efficiency to justice writ large.

When a court balances the call to adjudicate a cognizable copyright injury with preserving necessary judicial resources, the question presented is the nature and extent of the copying, not whether copying has occurred. Copying is not infringement. It’s the nature of the copying that matters.¹¹ The copyright de minimis analysis thus becomes a question of the kind of harm copyright law aims to prevent, because copying per se is not unlawful.¹² This is a harder question than might appear.¹³ In copyright law, what matters is the use to which the copying is put, not the fact of copying.¹⁴ A study of de minimis copyright may thus be helpful to clarify the nature and scope of cognizable copyright harms, which have been enduring and evolving questions in the field.¹⁵

The copyright de minimis doctrine has a long history that has shapeshifted over time.¹⁶ In the internet age, the copyright de minimis defense has increased in relevance as copyright lawsuits (and IP generally) are more mainstream and infringement liability more widespread.¹⁷ The digital age has made the copyright de minimis defense more relevant than ever. As artificial intelligence trained on copyrighted works has begun to dominate all areas of technology and society, whether those AI systems are making de minimis copies or causing substantial copyright harms is a question that could make or break the technology.¹⁸ In everyday life, we copy whole photographs and articles to send to friends and colleagues via email, text, and social media as essential forms of communication and discussion. When engaging in research, making our own art, and writing about the world around us, we inevitably copy bits of other copyrighted work; sometimes we copy the whole work and put it to productive and non-harmful uses. All these ubiquitous and quotidian activities implicate the copyright de minimis defense. This Article examines the patterns in assertions and analyses of the copyright de minimis defense over the past century and specifically in our digital era when copying is omnipresent but its triviality widely disputed. The Article’s aim is three-fold: (1) to map the de minimis defense as a way to learn more about it doctrinally—when is copying too much?; (2) to clarify the standard for the de minimis defense to better achieve its goal of judicial economy; and in doing so (3) to address the deeper questions about the harm copyright law aims to prevent in order to facilitate (and not chill) more communication and authorship in the digital age.¹⁹

This is the first empirical study of copyright lawsuits in which courts engage in some analysis of the de minimis defense.²⁰ The study begins with the hypothesis that in the digital era the copyright de minimis defense should be more relevant and more successful because, as much as there is more copying on the internet, most of it is trivial.²¹ The data shows the first to be true (more relevant) but not the second (litigation success rate). There is indeed a rise in the use of the de minimis defense in litigation, but not of its success in court.²² Under what circumstances does the defense succeed? This Article answers this question within the range of contexts reflected in the diversity of cases.

In doing so, related copyright doctrines also come under scrutiny, such as: the “substantial similarity test” for reproduction infringement; the fair use defense, which unlike the de minimis defense, arises after a finding of infringement; and the anti-aesthetic discrimination principle, which counsels courts to avoid aesthetic judgment when evaluating copyright infringement claims.²³ Each of these copyright rules relate to the evolution of the de minimis doctrine in some way. And, thus, mapping the de minimis doctrine over the past century adds to our understanding of these core and more common copyright doctrines.

As copyright adapts to technology that enables promiscuous copying and prolific collage, the vitality of the de minimis defense needs closer inspection. Doing so reveals opportunities to clarify the de minimis doctrine for future application, reestablishing its independence from infringement and fair use analyses and making de minimis a more useful defense. Because copying per se is not necessarily copyright infringement, there must be some copying less than “substantial” that is beyond copyright law’s reach. As this Article demonstrates, courts have long been engaging in some form of quantitative or qualitative analysis of copying when applying the de minimis defense. Analyzing the features and trends of those assessments may help preserve the tort-like nature of copyright, whereby threshold liability does not depend on proving the act occurred (copying) but instead on balancing harms from copying with benefits of dissemination and use.²⁴ Doing so also sharpens our focus on other critical aspects of copyright law, such as a bloated substantial similarity test or a stingy fair use analysis, which may both be causing trouble for copyright’s constitutional prerogative of promoting the progress of science.²⁵

There are only four scholarly analyses of the copyright de minimis doctrine, but none are empirical.²⁶ A few court cases from the past two decades discuss the copyright de minimis defense in detail and have become doctrinal benchmarks for its initial assessment.²⁷ The facts of those cases range widely, from a fleeting glimpse of a whole copyrighted pictorial work in the background of a film to the use of a musical fragment from one musical work in another.²⁸ But whether they reflect long-standing past practice or a change is unclear without the bigger picture. This Article will for the first time combine doctrinal, historical, and empirical perspectives on the copyright de minimis defense to put these more recent cases in perspective.

In the end, the Article concludes with a prescriptive proposal for the application of the de minimis defense that tracks the dominant trends in the cases and conforms with the doctrine’s 19^th century origins. The de minimis defense should be green-lighted at the motion to dismiss stage when copying is quantitatively or qualitatively insignificant as compared to the benefits normally derived from the exploitation of the copyrighted work. As described more fully in the Article, with some outlier exceptions, this is how the de minimis defense has been applied, but not always at the motion to dismiss stage and too often (most recently) with additional and unnecessary complexity. This Article cuts through the complexity, resets the de minimis defense for effective use by courts and litigants, and leaves the thornier questions of copyright infringement and fair use defenses for more involved copyright disputes.

The Article proceeds as follows. Part I discusses the early history of the de minimis doctrine and its 19^th century origins in equity, situating its essential role outside the statutory framework to maximize judicial resources and shortcut wasteful lawsuits. Part II puts this history in context of three eras of de minimis cases, “the Early Era,” the “Transition Era,” and the “Contemporary Era.” It explains the various characteristics of the cases from each of these eras and describes several exemplary cases to illuminate each time period. In doing so it highlights patterns that eventually resolve into a de minimis defense that is clearer and more efficacious. Part III describes the de minimis data set as a whole, the method for collecting and analyzing the cases, and notable trends among the three eras, further explaining the patterns in Part II. The analysis in Part III specifically connects the themes of the “Eras Tour” in Part II with a quantitative analysis of the data, including focus on the characteristics of the cases.²⁹ Surprising features of more recent de minimis cases emerge from this analysis, some are detrimental to judicial efficiency and others support it. Detrimental features include a wider variety of de minimis standards that complicate the terrain. Also, some standards improperly confound de minimis with substantial similarity, thwarting early dismissal and rendering the defense ineffective. Beneficial features include reviving the “technical use” standard for the de minimis defense (a non-actionable copying of a whole work), which is a historical anchor of the defense. Another beneficial trend is courts’ frequent countenance of defendant’s copying whole pictorial works in the internet age when images are more essential than ever to communication.

Part IV combines these trends and their features to propose a clarified de minimis standard to strengthen the efficacy of the defense. It explains that the narrowing and complicating of the de minimis defense is a recent trend only in the “contemporary era,” when copying has become ubiquitous and harm from de minimis copying is debatable. But laudable patterns preserving the historic defense are worth highlighting and strengthening for a revamped and clarified standard to serve judicial economy. The de facto narrowing of the de minimis defense through more complicated standards and broader liability for even insignificant copying needs to stop. Instead, a standard available on motion to dismiss, either sua sponte or by defendant’s motion, would combine quantitatively small amounts of copying with whole copies that are qualitatively insignificant as within the scope of the de minimis defense. This embraces and facilitates the digital era’s new modes of communication that rely on whole visual works such as photographs, images, and graphic art (think emojis, memes, and background visuals) which are and should continue to elude copyright liability when used referentially, informationally, in transitory or minimized fashion, or are otherwise substantively insignificant to the Defendant’s work. This comports with copyright policy and, we argue, can be solidified with modest practical changes that are anchored in the already established doctrinal framework for the defense. Doing so will revitalize the de minimis copyright defense for the digital age so that it assures judicial efficiency and aligns with purposes of copyright in the 21^st century.

In a case decided in the first year of the 21^st century, the copyright de minimis defense did not prevail but the Court of Appeals for the Second Circuit nonetheless explained its importance.

The de minimis doctrine is rarely discussed in copyright opinions because suits are rarely brought over trivial instances of copying. Nonetheless, it is an important aspect of the law of copyright. Trivial copying is a significant part of modern life. Most honest citizens in the modern world frequently engage, without hesitation, in trivial copying that, but for the de minimis doctrine, would technically constitute a violation of law. We do not hesitate to make a photocopy of a letter from a friend to show to another friend, or of a favorite cartoon to post on the refrigerator….Waiters at a restaurant sing “Happy Birthday” at a patron’s table. When we do such things, it is not that we are breaking the law but unlikely to be sued given the high cost of litigation. Because of the de minimis doctrine, in trivial instances of copying, we are in fact not breaking the law. If a copyright owner were to sue the makers of trivial copies, judgment would be for the defendants. The case would be dismissed because trivial copying is not an infringement.³⁰

This Article explores three eras of de minimis cases in light of the above statement of doctrine. It improves our understanding of copyright’s de minimis defense, which enhances judicial efficiency by cabining actionable claims to non-trivial copyright infractions. The Article thereby elucidates the contemporary harm copyright law aims to prevent and the benefits that non-actionable copying promotes. Doing so advantageously adjusts copyright’s contours for its optimal application in the digital age.

I. Historical Origins

Webb v. Powers (1847) may be the earliest incarnation of the de minimis doctrine in U.S. copyright law.³¹ It is the oldest case we identified that expressly uses the term “de minimis” in a context separate from fair use. Although fair use arose several years earlier in 1841, and was present in fair abridgement cases even earlier, Webb marks a distinct development.³² Notably, both Folsom v. Marsh, which is credited with creating the modern fair use doctrine,³³ and Webb v. Powers, the first U.S. copyright case we identified as relying on the de minimis defense, were decided by the same court in the same decade.

Webb quotes Folsom only once, yet the court’s analysis proceeds like a modified fair use analysis, focusing exclusively on the quantity and significance of the material copied.³⁴ From this decision, we see the origins of the copyright de minimis doctrine as an independent defense to infringement. At this early stage, the de minimis inquiry centered on at the “nature and value of the parts copied” and whether they are “so few and unimportant in number and value compared to the whole work.”³⁵ That formulation remained the central inquiry well into the late twentieth century.³⁶

In Webb, publisher of the book Flora’s Interpreter alleged that the defendant’s The Flower Vase infringed his copyright. Both books were botanical references containing categorizations and descriptions of flowers. The plaintiff alleged, in particular, that the defendant copied 20 of 148 definitions, about 156 words in total, from Flora’s Interpreter.³⁷ The defendant admitted she relied on and “meant to use the plaintiff’s book.”³⁸ The court nonetheless found no infringement, in part because the defendant’s book was “suited for a different class of readers.”³⁹ The court further explained that plaintiff’s own work clearly derived from other botany books, and many of the flowers “could not be described in any other way, if described naturally and truly.”⁴⁰ Beyond what may be considered imperceptible market harm and copying of uncopyrightable subject matter, the court confirmed that defendant did copy some “small part of the novelty in the arrangement” and “though small in value, such an imitation or appropriation… may be actionable…and the subject of an injunction, perhaps if easily separated from the rest of the book.”⁴¹ The court nonetheless determined that no separation was possible (on account that the copying of original material was based on organizational features), and that not only was the Master correct to refuse an injunction, but no damages would issue either.

[T]he part of the arrangement claimed to be original by the plaintiffs…was hardly sufficient to justify an injunction. A novelty in arrangement, especially so trifling as this, without any new material connected with it, seemed…to be of questionable sufficiency to be protected by a copyright. The Master seemed to be of the same opinion, on the grounds of `de minimis non curat lex.’ ~ldots~It would generally be equitable and just to let the party, under such circumstances, seek redress for it by damages in a suit at law. It is not a suitable case for the exercise of a peremptory injunction, which is chiefly to aid legal rights, and in cases of copyrights runs against specific parts copied. It is usually done to work substantial justice between the parties, rather than destroy the whole book of the defendants for the small infringement in the arrangement, if otherwise it was novel, and unexceptionable, and useful to the community. This conclusion renders it unnecessary to decide on the point how far the conduct of the plaintiffs…should bar this application, though redress might perhaps be still had at law.⁴²

In this passage, we see combined four related observations: (1) minimal copyrightable subject matter, (2) no substitutional market harm, (3) quantitatively small and qualitatively insignificant amounts of copying, and (4) the problem of enjoining the distribution of a whole work based on “unexceptional” and “useful” similarities. The decision refers throughout (and cites in full) to the Master’s report, the crucial part of which says “whether I regard quantity or value, I am compelled to conclude that the part of the materials copied by the defendants, is too insignificant in character for the law to notice. I therefore find that there is in no respect an infringement, by the defendants, of the copyright of the plaintiffs.”⁴³ It is a rich case to inaugurate the de minimis defense in copyright, tied as it is to the many other reasons a copyright infringement claim could fail. Notably, it lacks an analysis of substantial similarity because, of course, the plaintiff’s and defendant’s books did not resemble each other at all. The case concerned copying only very small parts, not a whole work, and thus it challenged existing conceptions of how copyright law protects authors.

Note how in Webb, the court accepts the Master’s determination in large part because injunctive relief was sought to bar the publication and distribution of Defendant’s book for the cause of reproducing only 156 words. The court also accepts that there might have been an action in law for nominal damages – which means there is perhaps a technical violation of copyright – but that the infringement was “too insignificant in character for the law to notice.”⁴⁴ Legal damages were not usually available in a court of equity, and the case was dismissed. That the court seemed limited by the injunctive remedy, which was the means of copyright enforcement at the time, reinforces the conclusion that defendant committed no cognizable copyright harm.⁴⁵ Scant remedies at law – nominal monetary damages – may still have been feasible in this case “only to the extent or value of the encroachment”⁴⁶ by the defendant although both the Master and the court seemed doubtful given the very small amount of copying.

In the 1840s and earlier, copyright largely extended only to books as published; abridging or quoting from books was not infringement.⁴⁷ Webb v. Powers is therefore an odd infringement case, except that it comes on the heels of Folsom v. Marsh, in which Justice Story held that an abridgement was an infringement.⁴⁸ That case concerned defendant’s 866 page abridgment, which copied 353 pages from a 6,763 page biography of George Washington.⁴⁹ L. Ray Patterson writes convincingly that the legacy of Folsom v. Marsh is to expand copyright’s monopoly by redefining infringement to include some abridgments, despite the court admitting that “a fair and bona fide abridgment of an original work is not a piracy of the copyright of the author.”⁵⁰

In Folsom, Justice Story explained that defendant’s use is not “fair” because “so much is taken, that the value of the original is sensibly diminished, or the labors of the original author are substantially to an injurious extent appropriated by another.”⁵¹ Justice Story’s explanation is based on “a theory of unjust enrichment”⁵² and focuses on quantity taken and market harm, but is further qualified by other considerations, such as “the nature and objects of the selections made, the quantity and value of the materials used, and the degree in which the use may prejudice the sale, or diminish the profits, or supersede the objects, of the original work.”⁵³ Justice Story explains that “none are entitled to save themselves trouble and expense, by availing themselves, for their own profit, of another man’s works, still entitled to the protection of copyright.”⁵⁴ The Supreme Court would eventually overrule this latter statement about “sweat of the brow,” but not until 1991 and after a century of disputes about whether in the absence of originality the author’s labor alone justifies copyright protection.⁵⁵ In the meantime, the de minimis doctrine and copyright fair use take root only six years apart as two related but distinct considerations in an equitable assessment of infringement defenses.

Six years later, Webb did not engage in a fair use analysis presumably because 156 words were just too few for a market substitutional effect. The 1831 Copyright Act gave the copyright holder “the sole right and liberty of printing, reprinting, and vending” the copyrighted work.⁵⁶ There was no statutory basis for a finding of infringement “that includes the copying of a portion (even a small portion) of a copyrighted work either in form or substance.”⁵⁷ But since Folsom opened the door to infringement determinations based on the quantity and quality of the copied material “and the degree in which the use may…supersede the objects[] of the original work,”⁵⁸ courts could now consider harm to plaintiffs from partial copying as weighed against the benefits of such minimal copying to readers and other writers.

Webb was not about piracy; it was a case about whether copying helpful material from one book to include in another without laboring to produce those portions oneself is the kind of copying copyright law should care about. Webb adapted Folsom v. Marsh for those situations in which no copies were made, but some small amount of copying occurred. When the court in Webb decided the case as a matter of equity in defendant’s favor, it correctly dismissed the case, characterizing it about “trifles with which the law should not concern itself.”⁵⁹ Webb sprouted a new copyright doctrine from Folsom‘s fair use language and held that some small amount of copying was “hardly sufficient to justify an injunction” and was defensible not because it was a fair use but because it was trivial and too “insignificant in character for the law to notice.”⁶⁰ Webb v. Powers is cited well into the mid-20^th century for this holding that defendant’s copying of 156 words from Plaintiff’s botany textbook was an “insubstantial invasion” of Plaintiff’s copyright for the court to take notice.⁶¹

The de minimis defense originates with the Court of Chancery as a “maxim of equity” and thus was correctly applied in Webb.⁶² With roots centuries old, the function of general common law doctrine is “to place outside the scope of legal relief the sorts of intangible injuries, normally small and invariably difficult to measure, that must be accepted as the price of living in society.”⁶³ As Minnesota’s highest court explained in 1908, the maxim addresses “mere trifles and technicalities” that “must yield to practical common sense and substantial justice.”⁶⁴ Over centuries, the de minimis doctrine was applied across many legal areas in which it declared diverse notions of “value” trivial (including “absolute values and relative values, monetary values and human values, public values and private values”).⁶⁵ Beginning with Webb v. Powers, copyright law was part of this history.

Between 1847 and 1909 (the enactment date of a new Copyright Act, which introduced statutory damages⁶⁶), there is one reported case that arguably considers the de minimis defense to defeat an infringement claim.⁶⁷ The one exception is List Publishing Co. v. Keller (1887). It concerns rival “social” directories, and the court determined that defendant infringed plaintiff’s directory based on the copying of 39 fake entries.⁶⁸ Its statement that “the compiler of a general directory is not at liberty to copy any part, however small, of a previous directory, to save himself the trouble of collecting the materials from original sources,” expresses the contestable notion that there is no such thing as de minimis copying.⁶⁹ It also paraphrases the “sweat of the brow” doctrine from Folsom quoted above, which the Supreme Court eventually overruled.⁷⁰ List Publishing is plausibly about how much copying is too much (just 39 fake entries) but also about how to prove unauthorized copying of a factual work (by seeding fake entries into a factual database). List Publishing is cited in only two mid-century cases discussing the de minimis defense.⁷¹ And its statement that a subsequent author cannot save themselves time by copying from another is no longer good law.⁷²

These are the early origins of copyright’s de minimis defense. The nineteenth century de minimis doctrine evaluated the nature and value of the parts copied compared to the whole work in the context of copyright law’s goals: promoting the progress of science through incentivizing the production of authored works by preventing market substitution.⁷³ From here, the de minimis defense blossomed into an independent defense for most of the twentieth century.⁷⁴ More recently, however, and since the beginning of the twenty-first century, our data shows that the significance of the de minimis defense has been dwarfed by fair use and the de minimis analysis partially colonized by the substantial similarity test.⁷⁵ To be sure, the de minimis defense began as an element of fair use, sharing with it the effect of exempting from infringement copying that was not a market substitute.⁷⁶ But fair use today does not only concern works that merely contain small parts of another copyrighted work. It covers much more, as the statutory preamble explains.⁷⁷ The de minimis defense became useful and relevant, as Webb v. Powers demonstrates, in its distinction from fair use and misappropriation (one standard for which became “substantial similarity”).⁷⁸ Yet, as both fair use and misappropriation standards evolve to become more complex and fact-driven, so does the de minimis defense when it was supposed to be a quick end to a frivolous lawsuit. We track that evolution below.

II. The Eras Tour (With Apologies to Taylor Swift)

Our comprehensive dataset of de minimis cases between 1847 and 2022 includes only 178 court opinions. (We discuss the collection and coding of these cases in Part III.) At first, we were surprised by this small number. But in the context of assessing the contested evaluation of a “trifle,” finding 178 examples is significant.

As already mentioned, the earliest de minimis case is Webb v. Powers from the District of Massachusetts in 1847.⁷⁹ After that, the de minimis defense does not occur with any regularity until the new century.⁸⁰ In 1903, copyright’s originality doctrine expands to assure that more works are covered by copyright protection.⁸¹ And then the new 1909 Copyright Act further broadens copyrightable subject matter, scope, and remedies by adding a new statutory damage regime.⁸² Both events make exceptions and limitations to infringement even more important, which we believe roused the de minimis defense from its mid-19^th century roots. As we describe below, this first Era contained twenty-one de minimis cases.

After the 1976 Act, defendants raise the de minimis defense with more frequency. Notably, fair use was codified in the 1976 Act, but the de minimis defense was not. Nonetheless, the variety and quantity of cases alleging de minimis copying as a defense to infringement rose substantially after 1978 to 157 cases, affirming its common law roots and its significance to court cases.⁸³ Several features of a changing society fuel this upward trend of de minimis cases. First, unauthorized copying becomes ubiquitous with copying technology commercialized and accessible (think photocopy machines, home recording technology, and the rise of personal computers).⁸⁴ Second, the entertainment industry’s licensing and distribution systems become sophisticated and complex, and freelance and independent authors are asserting copyright claims.⁸⁵ Third, mid-20^th century infringement standards evolve, including the doctrines of fragmented literal similarity and substantial similarity, allowing for reproduction infringement based on much smaller parts of the whole work.⁸⁶ These technological and industry changes, combined with infringement doctrines evolving to become notoriously complex and quixotic,⁸⁷ create the need for a more robust de minimis defense, which is asserted with more frequency in the Third Era.

In this part, we divide the cases into Eras related to these trends and legal milestones. Within each Era, we discuss patterns in the parties’ identities, works, and the legal standards. A noticeable trend is that in the Early Era cases are between competitors and the works at issue arguably compete in the same market (e.g., book publisher suing another book publisher). By Era Three, however, the disputes are less often between parties offering arguable market substitutes, and instead they are between putative licensees and licensors (e.g., a visual artist suing a television company for hanging art in the background of a television show set). This later breed of lawsuit includes plaintiff complaining of defendant’s failure to license plaintiff’s work (or a part of it) as an input into defendant’s work, even when the requirement to pay remains debatable under copyright law. This new claim for copyright harm–not clearly market substitution, but instead the rise of a kind of in-licensing or rights/clearance practice–puts pressure on the de minimis defense, which immunizes fragmentary, partial, or fleeting copying of authored works. As we explain infra, the rise of this new kind of in-licensing claim harms the de minimis defense and too easily shifts the court’s attention to the more fact-intensive and costly fair use analysis. We urge a correction of this trend at the end of the Article with clearer de minimis guidelines and recommendations to litigants and courts.

A. The Early Era (until 1977)

The Early Era, covering the 19^th century and the first half of the 20^th century, contains only 21 cases. Of those, only 5 ended in a successful de minimis defense (including Webb v. Powers).⁸⁸ The cases in the Early Era are largely between competitors in the same or closely related industries. They mostly concern directories, advertising catalogues, manuals and textbooks, high-information literary works of the technical or non-fiction kind. In all but four of the cases, copying is fragmentary or partial, and the court’s de minimis analysis while very brief features quantitative considerations.⁸⁹ It makes sense that few de minimis cases arose in the first half of the 20^th century, because copyright law still largely was thought to prohibit only copying of the whole work akin to market substitution.

When addressing defendant’s de minimis defense, the Early Era court often invokes a cursory standard with a single metric (usually quantitative) or no standard at all, and it simply says the use is or is not de minimis. Because most of these cases are fragmentary or partial copying cases, when the court finds infringement there is usually a determination that the copying has replaced the value of the original work in the marketplace. The substantial similarity standard for infringement was still developing into the one we recognize today, and it is rare in the Early Era for courts to use the phrase. Instead, courts say, for example, that “it is necessary that a substantial part of the copyrighted work be taken”⁹⁰ or the copying was “not substantial” or “important.”⁹¹

One such case, Mathews Conveyor Co. v. Palmer-Bee Co. (1943), is a dispute between competing manufacturing companies in which Plaintiff alleged a range of contract and unfair competition claims.⁹² The copyright claim concerns their respective catalogues and occupies only a small part of a much larger opinion devoted to discussing the agency and contract relationship between the parties.⁹³ Defendant allegedly reproduced as sketches several photographs from Plaintiff’s catalogue.⁹⁴ The court described the defendant’s catalogue sketches of machine parts and materials as not being exact copies and being from different angles than the plaintiff’s catalogue photographs.⁹⁵ Also, the status of plaintiff’s copyright was disputed because the catalogue appeared to be distributed without notice of copyright.⁹⁶

Notwithstanding these weaknesses in plaintiff’s case, the Court of Appeals went on to decide

“the issue on this phase of the case…from a consideration whether there has been such a substantial infringement of plaintiff’s right…[and] that in order to sustain an action for infringement of copyright, a substantial copy of the whole, or a material part, must be reproduced.”⁹⁷

As to the de minimis defense, the court said “on the principle of de minimis non curat lex, it is necessary that a substantial part of the copyrighted work be taken.”⁹⁸ It cites Folsom v. Marsh for the proposition that “whether so much has been taken as would sensibly diminish the value of the original” is also a factor.⁹⁹ The court then affirms the denial of Plaintiff’s copyright infringement claim, citing several reasons, including “that any injury to the plaintiff from defendant’s action in regard to these two sketches…is obviously infinitesimal; and that the alleged appropriation of the idea or form or perspective of plaintiff’s two cuts from among the hundreds in its catalogue, is insubstantial as an infringement of plaintiff’s copyrighted book.”¹⁰⁰

We see in Mathews the characteristic feature of competitors fighting over market share and harnessing dubious copyright claims to do so. In this case, the court conceives of copyright infringement as a cause of action that prevents injury to the value of the work as a whole and does not consider the individual parts of the work (here photographs made into sketches) as separate from the catalogue.¹⁰¹ It further assesses the reproduction of two of the photographs into sketches as “insubstantial” and any injury suffered as “infinitesimal.”¹⁰² The de minimis analysis, infringement assessment, and fair use question reinforce each other, and also reinforce the conclusion of no copyright liability, but they are nonetheless described as distinct.

Another case from this era, Advertisers Exchange v. Hinckley (1951), involved a contract dispute between industry players in advertising.¹⁰³ Plaintiff provided copyrighted advertisements to defendant-advertiser for redistribution in local papers in a select geographic area for a fixed annual fee.¹⁰⁴ Defendant purportedly cancelled the contract with Plaintiff but continued to use the copyrighted materials (twenty-nine advertisements) for more than eighteen months thereafter in a local newspaper with a circulation of 3,261 copies.¹⁰⁵ The court spends almost no part of the opinion deciding the issue of de minimis raised by the defendant. The extent of its ruling on that matter is this: “Compared to the amount which had been sent to the defendant from which he might choose such as he desired to use, it is true the amount was small, but it was important, and I think a substantial part of the material was used by defendant over this period of approximately 20 months.”¹⁰⁶

Most of the opinion discusses whether the contract was in fact terminated and if so, what the damages should be. As to the latter issue, the court questions whether copyright or contract damages makes sense. A calculation of copyright damages “runs into a ridiculous amount,” the court says, and “it never was intended that the statute should apply in a case of this kind.”¹⁰⁷ Because the contract was for services—the provision of advertising copy for a fixed annual sum no matter how much advertising copy the defendant used—the court rejected the idea of per-copy damages.¹⁰⁸ “The only actual damages the plaintiff could possibly have suffered would be the loss of the sale of service to some other merchant in the community, which amounted to $156 a year, under the terms of the contract.”¹⁰⁹

We see in Hinckley, as in Mathews, industry siblings with arm’s-length agreements in place. When the relationship breaks down, the plaintiff tries to use copyright law (which is arguably a marginal aspect of the business relationship) to recover damages from what is an obvious breach by the defendant of contractual terms. As in Mathews, the Hinckley court focuses on quantitative metrics to assess whether the copying is de minimis because, while each of the advertisements were copied as a whole, the amount of time the advertisements circulated was considered central to Plaintiff’s injury. The works are copyrightable literary works with pictorial components, and yet the court assesses the harm from copying in terms of lost contract revenue, substantially mitigating the damages under copyright law that would have been substantially higher given the statutory damages regime. This mitigation perhaps reflects the “small” number of unauthorized copies relative to the amount to which Defendant had access under the contract, counterbalanced by the “importance” of the unauthorized copies relative to the parties’ relationship and the duration Defendant continued to use the advertisements after terminating the contract. Unlike Mathews, Hinckley contains discussion of neither the infringement standard nor fair use. When the Defendant raises the de minimis defense in response to the Plaintiff’s copyright claim, the court simply says that the Defendant’s actions are not de minimis.¹¹⁰ Instead of assessing copyright damages, the court rewards the plaintiff under the contract, indicating sub silentio that the plaintiff’s injury only tangentially relates to unauthorized copying.¹¹¹

In these cases, courts accept the fact of unauthorized copying of copyrightable material and assess whether that copying crossed a threshold of harm that copyright law aims to prevent. That harm appears to be the expected investment in the material and/or the value of the material copied measured by the context of the overall business transaction. The pithy de minimis discussion in light of the lengthier descriptions of the business transactions reflects the courts’ appreciation of the relevance of the parties’ business relationship to assessing harm from copying of the business materials. We think it fair to characterize these as competition cases in which de minimis relates to the extent of unfairness of the copying (but not the copying itself) and whether the copying interfered with the plaintiff’s business expectations. Whether we call these examples of copying “technical violations”—copying without cognizable harms—or copying that does not meet the quantitative threshold for infringement is reasonably debatable. But neither of these cases (and most cases in this era) do not evaluate infringement in a manner that resembles the substantial similarity analysis we have come to know.¹¹² Nonetheless, the cases from this Early Era anchor the common law de minimis defense in copying that is unrelated to, or has so insubstantial an effect on, the plaintiff’s business interests.

B. The Transition Era (1978-1997)

The second era covers the first two decades of the 1976 Act. It contains thirty cases, of which eleven ended in successful de minimis defenses (only three of which also involved fair use, a feature of this Era we discuss infra in Part III). All but one of those prevailing defendant cases concerned copying of literary works.¹¹³ The number of de minimis cases in this era rises slightly as a percentage of all copyright case decisions (from 2% to 4%), but defendant win rates remains relatively the same.¹¹⁴ Distinct from the Early Era, the Transition Era included literary works encompassing both fictional and non-fictional works, as well as software cases for the first time. Like the Early Era, most litigation pairs in the Transition Era are parties in the same or similar fields, and cases brought to prevent industry competition.¹¹⁵

A difference in the Transition Era and the Early Era concerns the nature of the industries and the amount of the copying. As to the industries, in this second era, we see cases disputing the copying of software.¹¹⁶ And there are more entertainment companies, including television, music, and film as both plaintiff and defendant.¹¹⁷ As to the amount of copying, this era introduces defendants copying plaintiffs’ works in whole for inclusion in different works, not just parts of plaintiff’s works, such as a plaintiff’s sculpture in a film¹¹⁸ and plaintiffs’ photographs as part of an internet distribution service.¹¹⁹ Thus, the Transition Era appears to inaugurate the potential for in-licensing whole works as parts of other works.¹²⁰ Newer copyright-rich businesses are more prevalent in this era.¹²¹

This Transition Era also inaugurates more case variation as between plaintiffs and defendants, size of the copying, and the legal standards courts employ to evaluate the de minimis defense. Quantitative metrics still predominate as they did in the Early Era, but in this second era courts more frequently employ a standard that approximates fragmented literal similarity to assess whether the defendant’s copying of the plaintiff’s work is either de minimis or constitutes a cognizable copyright harm.¹²² As in the first era, there are many cases in which the standard is conclusory, and the court simply accepts or rejects a finding of “de minimis non curat lex.” But in contrast to the first era, this second era contains cases in which courts evoke “recognizability” and “substantial similarity” as relevant to the de minimis assessment. These more fact-intensive and plaintiff-friendly analyses become common touchpoints for the infringement standard in the third (and contemporary) era and, we argue, unnecessarily complicate the de minimis defense, rendering it less able to achieve its purpose of judicial economy.

This is a Transition Era because we surmise that when media companies diversify, contributors to those media companies vie for market share in the value of the distributed content, and newer media companies and networks compete with older and more established ones.¹²³ As copyright industries diversify, and more copyright authors claim a share of the revenue, courts struggle to determine when the defendant’s copying is trivial in relation to the parties’ business as a going concern and when, in contrast, copyright law should protect the plaintiff’s investment in their work by forcing defendant to pay or refrain from its use. Two illustrative cases follow in which the court correctly held the defendant’s use was defensible but not necessarily because it was de minimis. Both cases contain seeds of the doctrine’s unfortunate complexification, which sprout in the Third Era.

Computer Associates International v. Altai Inc. is well-known for its implementation of the “abstract-filtration-comparison” test for copyright infringement of computer software.¹²⁴ The case “deals with the challenging question of whether and to what extent the `non-literal’ aspects of a computer program…are protected by copyright.”¹²⁵ The court mentions the de minimis defense exactly once: in the context of approving the district court’s infringement analysis (using a fragmented literal similarity test) when comparing the competing programs’ parameter lists and macros as part of a fragmented literal similarity test. The court writes:

[F]unctional elements and elements taken from the public domain do not qualify for copyright protection. With respect to the few remaining parameter lists and macros, the district court could reasonably conclude that they did not warrant a finding of infringement given their relative contribution to the overall program. See Warner Bros., Inc. v. American Broadcasting Cos., Inc.,720 F. 2d. 231, 242 (2d Cir. 1983) (discussing de minimis exception which allows for literal copying of a small and usually insignificant portion of the plaintiff’s work); 3 Nimmer § 13.03[F][5], at 13–74.¹²⁶

In support of the non-infringement determination, the court cites a case and a treatise, both confusingly relating the de minimis defense with the infringement standard.¹²⁷ The court relies on Nimmer for his discussion of the fragmented literal similarity test, occurring when copying is exact but not comprehensive.¹²⁸ Exact copying of small portions of a copyrighted work should be de minimis, but not when the copying is extensive. Nimmer’s discussion of fragmented literal similarity, and the cases adopting it, is a major step toward what Oren Bracha has called the doctrinal “slippery slope,” whereby de minimis (as measured in small bits) becomes part of the substantial similarity test for infringement (as measured both quantitatively and qualitatively).¹²⁹ The court also cites Warner Brothers v. ABC, a case from this same time period when plaintiff film company unsuccessfully alleged infringement by defendant television company of Superman (the character) in the new television show “The Greatest American Hero.”¹³⁰ The defendant evoked the de minimis defense as part of a fragmented literal similarity test to support its argument that only a small part of a large work was copied (a character) and arguably only an idea (of a superhero).¹³¹ The appellate court in Computer Associates, applying this precedent, approved the district court’s finding of no infringement, saying “the district court reasonably found that [plaintiff] failed to meet its burden of proof on whether the macros and parameter lists at issue were substantially similar.”¹³² Here, we see the use of fragmented literal similarity test evolving to bridge the de minimis defense with the infringement standard, melding quantitative and qualitative evaluations into a broader substantial similarity analysis.¹³³ This move arguably guides courts to decide de minimis cases when quantitative metrics are objectively discernible and the use of the small bits are insubstantial parts of the plaintiff’s work. This is a laudable and clear baseline standard for a de minimis defense on a motion to dismiss, and one we take up at the end of this Article. It nevertheless gets marginalized as the de minimis doctrine evolves to appear entangled with more complicated infringement standards (as it does in this case) and more frequent fair use defenses.

Computer Associates retains some features of the Early Era. It maintains a quantitative focus on the nature of the copying and describes a competitive situation between parties whose literary works – computer software – are thinly protected by copyright but are nonetheless valuable products for sale. The de minimis analysis in Computer Associates bridges the earlier cases, in which discrete amounts of copying were measurable as against copying of the whole work, with the newer cases in which valuable but small parts of the work are being copied (literary characters or software macros). Computer Associates mentions the de minimis defense, fragmented literal similarity analysis, and the substantial similarity test all in one paragraph,¹³⁴ evidencing the court’s struggle to determine when fractional but material copying is cognizable as an infringement or when it is otherwise defensible. The defendant-friendly holding makes good sense to us, and it is followed in future decades as significant precedent.¹³⁵ But the de minimis aspect of the case is underappreciated and, we think, played an important role in the defendant’s success.

Carola Amsinck v. Columbia Pictures Industries Inc. presents another kind of transition case.¹³⁶ It is not a case between competitors, but one in which a self-employed graphic artist, Carola Amsinck, sues a film company for its use of her whole copyright work (pastel-colored teddy bears) in several movie scenes.¹³⁷ This case is thus a precursor to the claims in which parties dispute whether distinct copyrighted works (e.g., graphic art) incorporated as an input into larger and more complex copyrighted works (e.g., a film) is copyright infringement.¹³⁸ Amsinck’s teddy bears were made, with her permission, into a baby mobile by a toy company.¹³⁹ The company making the mobile credited Amsinck with the original design and paid her.¹⁴⁰ But Defendant Columbia Pictures neither gave Amsinck credit nor licensed the work (as a toy or graphic art) for use in the film.¹⁴¹ The mobile appears in several scenes in the film for between two to twenty-one seconds, “with a total exposure of approximately one minute and thirty-six seconds.”¹⁴² The court describes the mobile as at times “barely visible” and at other times “in a close-up shot.”¹⁴³

Defendant Columbia Pictures prevails in a combination ruling that the film was not a “copy” of the mobile in the way copyright cares about, and, in the alternative, because the film’s use of the copy was fair.¹⁴⁴ As to the first point, the court concludes that Columbia Pictures did not commit substitutional copying, which is the kind copyright law aims to prevent.¹⁴⁵ The court states that

[T]o establish `copying,’ the plaintiff must prove that the defendant `mechanically copied the plaintiff’s work’; there must be some degree of permanence or the maxim `de minimis‘ applies, requiring a finding of no liability. In determining whether a use constitutes a copy, the courts look to a functional test to see whether the use has `the intent or the effect of fulfilling the demand for the original.¹⁴⁶

This invocation of de minimis resembles the cases from the Early Era in which courts found the defendant’s copying of parts of the plaintiff’s work to be trivial in the context of market substitution. The difference here is that the defendant is copying the whole of the plaintiff’s work and the court’s assessment of triviality relates to the minimal time the work appeared in the film and the lack of market for its filmed appearance. The court concludes that

[T]he defendants’ display of the Mobile bearing Amsinck’s work is different in nature from her copyrighted design. In this matter, the defendants’ use was not meant to supplant demand for Amsinck’s work; nor does the film have the effect of diminishing interest in Amsinck’s work. Defendant’s use was not a mechanical copy. Defendants’ use, which appears for only seconds at a time and can be seen only by viewing a film, is fleeting and impermanent. This Court therefore concludes that the defendants’ use is not a copy for the purposes of a copyright infringement action.¹⁴⁷

This resembles the technical use defense that Judge Leval evokes in Davis v. Gap, Inc. when he describes the singing of Happy Birthday by restaurant patrons, quoted earlier.¹⁴⁸ Whereas defendant (Columbia Pictures) used and enjoyed plaintiff’s work, defendant’s use and enjoyment are insignificant in light of the defendant’s work as a whole and as compared to how plaintiff (Amsinck) ordinarily earns a living from her copyrighted work (e.g., selling copies of the teddy bears or licensing copies of them to make mobiles). This court determines triviality, or the existence of a “trifle,” in terms of the non-existence or miniscule market harm to the plaintiff. It does not consider the use of the plaintiff’s derivative work in the film as a market on which the plaintiff reasonably relies, and even if it existed, the use was so fleeting or trivial to be inconsequential.

The court’s determination that the defendant did not engage in copyright infringement could have ended the case. But the court nonetheless also conducted a fair use analysis, possibly, to insulate its first ruling with a second in case of appeal.¹⁴⁹ The fair use analysis starts with factor four and, essentially, repeats the conclusion from the first part of the opinion: “In situations where the copyright owner suffers no demonstrable harm from the use of the work, fair use overlaps with the legal doctrine of de minimis, requiring a finding of no liability for infringement.”¹⁵⁰ The court explains that

The film does not pose a threat to the market for Amsinck’s work in either licensing artwork or in future sales to motion pictures. Indeed, the Court believes that, notwithstanding the lack of identification of the Mobile, its use in the film might actually increase the demand for mobiles in general, thereby benefitting plaintiff indirectly.¹⁵¹

From this statement, we learn the court does not consider the parties to be competitors in related markets; it considers the defendant’s use of the plaintiff’s work as beyond the reasonable or customary use that would require payment.¹⁵² The court’s analysis of the third fair use factor (amount and substantiality) similarly repeats its earlier de minimis analysis.¹⁵³ While conceding that the defendant displays the design “in its entirety” and “used every element of the work,” the court explains, “the Mobile is seen for only a few seconds at a time, and the artwork itself is visible for less than 96 seconds.”¹⁵⁴ The court concludes that “[a]lthough generally a use does not constitute fair use if it reproduces the entire work, the Court declines to find that the defendants’ short-term display of the Mobile in a film precludes a finding of fair use.”¹⁵⁵

The court’s quantitative analysis (the amount of time the works was visible in the film) combined with an assessment of market harm as a measure of whether the use was de minimis resembles earlier cases about partial copying and market substitution. It also resembles later cases discussed infra in the Contemporary Era insofar as: the copying is whole (however brief); the parties are not competitors; and the plaintiff’s work is an input to the defendant’s work, albeit not in a central way (e.g., not in the manner of sequels or translations).¹⁵⁶ Notably, however, the Amsinck court does not ask whether the defendant’s work is “substantially similar” to the plaintiff’s. Doing so would not make sense in this context. But the court does consider how visible the plaintiff’s work is to viewers and how much the defendant’s work focuses on the plaintiff’s work, resembling a “recognizability” evaluation that eventually (and unfortunately) becomes an alternative for “substantial similarity” in similar cases in the Contemporary Era.¹⁵⁷ This case thus bridges the earlier era with the subsequent one, combining a quantitative and qualitative assessment, perhaps because of the rising frequency of whole copying.¹⁵⁸ Fair use is also more helpful now that it has been expressly codified in the 1976 statute requiring analysis of market harm, purpose and character of the use, amount copied, and nature of the work.¹⁵⁹

Despite the prolonged fair use analysis in the Amsinck case, it is a laudable decision that – although not decided on a motion to dismiss – reaches the right result for the right reasons. Quantitatively miniscule and qualitatively insignificant, the use of plaintiff’s derivative work in the background of a film is not copyright infringement. Notable is that this case is about the whole of a plaintiff’s work, not parts of it, and still the court determines that the copying is trivial. To be sure, judicial and lawyering resources were wasted deciding a case about such a triviality. Hopefully, the patterns and standards described in this Article can streamline cases like Amsinck and guide courts and defendants in the future.

As we describe more fully in Part III, after the 1976 Act and in the Transition and Contemporary Eras, de minimis cases include many more defendants arguing that whole copies are nonetheless de minimis.¹⁶⁰ This may seem counterintuitive, but it’s an important evolution of the case law that should save more judicial and lawyering resources if followed. Only four such cases existed between 1960-1977, comprising 19% of the cases before 1978. Starting in 1978, and in this Transition Era, cases alleging whole copying in which the defendant raises a de minimis defense compromises 30% of the cases. This increase might be a result of earlier courts treating the “work” as a more fixed and less flexible category.¹⁶¹ As explained supra, in the first half of the 20^th century and in the Early Era, parts of works copied and disputed as de minimis may have concerned photographs or sketches in catalogues or manuals, where the work was the catalogue and the part was the photograph or sketch contained within it.¹⁶² But in the second and third eras, a plaintiff might bring such a case for unauthorized copying of a photograph or a work of visual art (as in Amsinck) because that work is separately authored and owned.

The growing number of “whole” work de minimis cases after 1978 might also be a result of increased access to copying technology, the growing acceptance of copying whole works for personal use, ephemeral use, or as raw material for the rapidly evolving entertainment and high-technology industries.¹⁶³ Whatever the reason for the rise of whole copying cases, they put pressure on the de minimis standard, which was primarily (although not always) concerned with small parts of works, not the inconsequential use of whole works. This pressure should be alleviated in light of the revealed patterns in these cases adjudicating whole copying as de minimis when uses are quantitatively small and qualitatively insignificant.

We consider both Computer Associates and Amsinck examples of early versions of what emerges in the Contemporary Era as a fragile basis for the de minimis defense, delineated by milestone case Ringgold v. Black Entertainment Television.¹⁶⁴ As the next section explains, Ringgold describes and affirms both a “technical violation” (a whole copy without consequence) and “quantitative threshold” standard for the de minimis defense. The first is “a technical violation of a right so trivial that the law will not impose legal consequences”;¹⁶⁵ and the second is copying that “has occurred to such a trivial extent as to fall below the quantitative threshold of substantial similarity.”¹⁶⁶ Ringgold also includes a qualitative analysis of the use to which the copy is put as part of its description of the de minimis defense, which closely resembles a substantial similarity analysis.¹⁶⁷ Before and during the Transition Era, the substantial similarity test was evolving and expanding. We believe this shift was partly because of the changing nature of the claims being brought by parties who demand licenses for a broad range of uses of their copyrighted works (e.g., Carole Amsinck). When plaintiffs bring infringement claims with more frequency of whole copying as background input into complex audiovisual works, the quantitative metric of the de minimis standard adjusts and the qualitative metric rises in significance. We discuss this trend, combining quantitative and qualitative metrics complexifying the de minimis defense in Parts II.C and III below.

C. The Contemporary Era (1998-2022)

The Contemporary Era differs in significant ways from both the Early and the Transition Eras. As Part III describes in more detail, there are many more cases in the Contemporary Era. Also, the rate of defendant successes declines (e.g., the de minimis defense is less successful overall). The third era is also characterized by more diversity: the identities of the litigants are more varied, plaintiff-authors appear more frequently to file opportunistic lawsuits for licensing revenue with questionable claims of market harm, and courts’ legal evaluations of de minimis are more varied and complex.

There are 127 cases in this era, which as a percent of all copyright decisions is more than double the previous eras.¹⁶⁸ It might make sense, then, that legal standards would be more diverse as courts struggle to assess de minimis defenses across a wide range of claims and industries.¹⁶⁹ Although the de minimis defense is more common in the Contemporary Era, it prevails less frequently as compared to the Transition Era.¹⁷⁰

As in previous eras, this one includes competitors in industries not reasonably relying on copyright, who nonetheless harness copyright to contest certain business practices. These include financial services and manufacturing companies competing on sales of goods and services but who sue for, among other things, copying parts of catalogues or business materials.¹⁷¹ Copyright is largely a side-show in these cases in which trade secret misappropriation, tortious interference with contractual relations, and breach of fiduciary duty play more central roles.

Different from earlier eras, however, in the Contemporary Era, there are more music cases and many more software cases.¹⁷² This tracks the growth of both industries in the internet age and to us is unsurprising, especially when copying small parts of either music or software is easier with digital technology and bits of music or software are reasonably considered lawful input into new music or software. These cases become notorious and contested disputes, as the “value” of the bit copied (e.g., a musical riff or software code module) arguably has independent market significance. These features confound the de minimis doctrine, as we see in the discussion of VMG Salsoul v. Ciccone below.

The bulk of the cases in this era involve entertainment or media companies and distributors, and they are between individuals and companies or individuals and other individuals (e.g., not between competitors).¹⁷³ The kind of copying now is fairly split between whole and partial (or fragmentary), with whole copying outpacing both in the latter part of this era between 2008-2022.¹⁷⁴ Plaintiffs’ claims of whole copying of their pictorial/image-based works predominate in the Contemporary Era, and the landmark cases discussed below illustrate this trend. This is when the de minimis doctrine should be most useful, in the vein of Amsinck and Computer Associates, and yet its contours blur in tandem with the more complex infringement standards and fair use assessments that are also evolving.

Over half of the courts in cases of this third era employ either no evaluative standard or only a quantitative standard to determine whether the copying was de minimis.¹⁷⁵ This is not surprising, because that is how de minimis started and it reflects the earlier eras. However, in the Contemporary Era, qualitative evaluations of de minimis become more common, with as many as 42 of the cases (a third) referring to “substantial similarity.”¹⁷⁶ “Substantial similarity” often includes “recognizability” as part of the standard, which we think makes Plaintiff’s case easier to prove and contributes to the unfortunate weakening of the de minimis defense.

The Contemporary Era cases span a range of approaches reflecting this era’s diversity. Cases such as Ringgold v. Black Entertainment Television (1997),¹⁷⁷ in which a visual artist plaintiff wins against a television company, and Gottlieb v. Paramount Pictures Corporation,¹⁷⁸ in which a visual artist plaintiff loses against a film company, are different sides of the same coin. We discuss them below.¹⁷⁹ These have come to be known as landmark cases in for the doctrine.¹⁸⁰ But whatever their milestone status, their doctrinal approach to de minimis, including qualitative evaluations that resemble aesthetic assessments, remains a minority approach. A contrasting case, the music sampling dispute VMG Salsoul v. Ciccone, reflects approaches from earlier eras assessing fragmentary copying in quantitative terms and elevates the de minimis standard in a way that would effectuate judicial economy and copyright policy if decided on a motion to dismiss.¹⁸¹ As discussed below, the defendant’s win in VMG aligns with the trends in the earlier eras and continues in this contemporary era to maintain the vitality of the de minimis defense.

Finally, the case of Bell v. Wilmott Storage Services (2021), rejecting a de minimis defense for whole copying of a photograph on an obscure website, is an outlier and, we argue, incorrect. But it is nonetheless significant.¹⁸² Plaintiff prevails in Wilmott for what in earlier eras would be a technical violation of copyright (a whole copy without harmful effect). Defendant’s display of plaintiff’s photo on a hard-to-access website inflicted no market harm.¹⁸³ Courts in earlier eras would likely have dismissed as de minimis the single use of a photograph in an undistributed (or not widely distributed) publication to avoid wasting judicial resources.¹⁸⁴ That the plaintiff-photographer prevails in Bell v. Wilmott is unfortunate and a mistake. The court’s reasoning in Wilmott relies in part on the absence of de minimis in the statute’s text, and thus is a head-in-the-sand formalistic reading of copyright’s scope defying the century’s long history of the common law de minimis defense.¹⁸⁵ Wilmot is also inconsistent with many of the cases in this era of whole copying which result in a relatively high number of defense wins. As we explain more fully in Parts III and IV, this contemporary era’s increasing tolerance for whole copying of pictorial works, contra Wilmot, inaugurates an updated and laudable version of the quantitatively minimal copying standard from prior eras. In this Contemporary Era, the quantitative analysis applies consistently to whole works when used for a short time and with insufficient visibility.¹⁸⁶ If courts in the future fail to follow this trend and instead rely on Bell v. Wilmot, something we do not advocate, the de minimis doctrine is likely to become less helpful to defendants and courts, frustrating the principle of judicial economy that it serves.

1. Ringgold and the Second Circuit Cases

Ringgold is a dispute about the unauthorized display of a “story quilt” (in poster form) as background set decoration for a television show.¹⁸⁷ (See Image 1.) The poster of the quilt was displayed intermittently during the half-hour sitcom for a total of 27 seconds.¹⁸⁸ The court nonetheless held that that use of the poster was not de minimis and violated Faith Ringgold’s copyright in the quilt. The court also rejected fair use as a defense.¹⁸⁹

The Ringgold court recognized that a “copyrighted work might be copied as a factual matter, yet a serious dispute might remain as to whether the copying that occurred was actionable.”¹⁹⁰ Notably, Ringgold described three possible invocations of de minimis as a defense to copyright infringement: (1) “a technical violation of a right so trivial that the law will not impose legal consequences”; (2) copying that “has occurred to such a trivial extent as to fall below the quantitative threshold of substantial similarity, which is always a required element of actionable copying”; and (3) as part of the fair use analysis to assess its third factor, “the amount and substantiality of the portion used in relation to the copyrighted work as a whole.”¹⁹¹

This tripartite division of the copyright de minimis defense is a fair reading of its history across the eras. The Ringgold court explains that the first category of “technical violations” is “rarely litigated” because of their very nature.¹⁹² In support, it references Judge Pierre Leval’s 1997 Nimmer Lecture (reproduced in his opinion in Davis v. Gap, Inc. and quoted supra) in which he explains that singing “Happy Birthday” in a restaurant or making a photocopy of a cartoon to hang on the refrigerator are examples of “de minimis non curat lex.”¹⁹³ The Ringgold court also cites its Knickerbocker Toy Co. v. Azrak-Hamway Int’l Inc. decision.¹⁹⁴ In that case, the court dismissed the plaintiff’s copyright infringement claim on the basis of a technical but trivial violation in which an office copy of the plaintiff’s work was made but never widely distributed, concluding the harm from the unauthorized copying was so small that legal enforcement was not justified.¹⁹⁵

Ringgold‘s main contribution to copyright law is its application of the second category of the copyright de minimis defense, which it uses to analyze the sitcom’s use of the story quilt. In considering whether the second kind of de minimis use prevails as a defense, Ringgold applies a standard that copying of protected material must “fall below the quantitative threshold of substantial similarity.”¹⁹⁶ The court says this necessitates looking “to the amount of the copyrighted work that was copied, as well as (in cases involving visual works), the observability of the copyrighted works in the allegedly infringing work.”¹⁹⁷ Observability demands consideration of factors such as “length of time the copied work is observable…and such factors as focus, lighting, camera angles, and prominence.”¹⁹⁸

Applying this standard, the court engages in an aesthetic interpretation of the sitcom to reject a de minimis defense. The court assesses the television show’s use of the story quilt – how the show displayed the story quilt (prominently and in focus) and the role the quilt played in the sitcom’s story (affirming themes and deepening character delineation). It describes a 4-5 second segment in which its “own inspection of a tape of the program reveals that some aspects of observability are not fairly in dispute.”¹⁹⁹ In this description, the court discusses camera angle, position of actors in relation to the quilt, scene framing, and camera focus. It concludes, saying

An observer can see that what is hung is some form of artwork, depicting a group of African-American adults and children with a pond in the background. The brevity of the segment and the lack of perfect focus preclude identification of the details of the work, but the two-dimensional aspect of the figures and the bold colors are seen in sufficient clarity to suggest a work somewhat in the style of Grandma Moses….All the other segments are of lesser duration and/or contain smaller and less distinct portions of the poster. However, their repetitive effect somewhat reenforces the visual effect of the observable four-to-five second segment just described.²⁰⁰

Based on this quantitative and qualitative analysis of the sitcom’s use of the quilt, the court determines that the quilt is “recognizable,… and with sufficient observable detail for the `average lay observer’ to discern African-Americans in Ringgold’s colorful, virtually two-dimensional style. The de minimis threshold for actionable copying of protected expression has been crossed.”²⁰¹ Finding no de minimis use and substantial similarity, the court then also rejects the defense of fair use, concluding the balance of the factors favors the plaintiff despite the defendant prevailing on factor three.²⁰² Ringgold‘s effect is that later cases also conduct this layered aesthetic analysis by folding the de minimis defense into the infringement assessment as a component of the substantial similarity test, grounding copyright misappropriation in recognizability by the “average lay observer.”²⁰³

The Ringgold test, while plaintiff-friendly, does not inevitably result in a defense loss. A year after Ringgold, the Court of Appeals for the Second Circuit decides Sandoval v. New Line Cinema Corp, a case raising a similar circumstance as Ringgold.²⁰⁴ In Sandoval, plaintiff photographer Jorge Antonio Sandoval disputes the use of ten of his photographs by New Line Cinema in the background of the film “Seven.”²⁰⁵ The photographs were used in the full-length film and “visible, in whole or in part, for a total of approximately 35.6 seconds.”²⁰⁶ The court cites Ringgold for the applicable de minimis standard: whether “the copying of the protected material is so trivial as to fall below the quantitative threshold of substantial similarity, which is always a required element of actionable copying.”²⁰⁷ As in Ringgold, the court closely analyzes the use of the photographs, describing their observability and contextualization in the film. In the court’s description,

[T]he photographs never appear in focus, and except for two of the shots, are seen in the distant background, often obstructed from view by one of the actors,…figures in the photographs are barely discernable, with one shot lasting for four seconds and the other for two seconds. Moreover, in one of the shots, after one and a half seconds, the photograph is completely obstructed by a prop in the scene.²⁰⁸

As in Ringgold, the Sandoval court reviews video evidence, becoming a kind of film critic.²⁰⁹ But unlike in Ringgold where plaintiff prevails, in Sandoval the plaintiff loses. The Court of Appeals affirms the district court’s grant of summary judgment for the defendant, finding “that the defendant’s copying of Sandoval’s photographs falls below the quantitative threshold of substantial similarity….[b]ecause Sandoval’s photographs appear fleetingly and are obscured, severely out of focus, and virtually unidentifiable, we find the use of those photographs to be de minimis.”²¹⁰

Notice how in both Ringgold and Sandoval, defendants copied all of the plaintiff’s work – the whole quilt and each photograph. Defendants’ copying was neither fragmentary nor partial. These cases are thus unlike those from previous eras that contest the harmful uses of fragments or parts (sentences or sketches) that compromise pieces of a larger whole. Nonetheless, these courts evaluate the whole copying of pictorial works under a de minimis standard as either “a technical violation of a right so trivial that the law will not impose legal consequences” or as quantitatively trivial and qualitatively insignificant.²¹¹ In both Ringgold and Sandoval – as in Carola Amsinck – the defendant’s use of the plaintiff’s work is arguably not within an existing or foreseeable reproduction or licensing market. Plaintiffs and defendants are in distinct industries with separate markets for their authored works (fabric arts and television (Ringgold); still photography and Hollywood (Sandoval)). And the harm to plaintiffs in both cases is not a competitive substitutionary harm; defendant’s use is not a market replacement for plaintiff’s work or a labor-saving device.

A decade later, the Ringgold de minimis analysis is so well established that it even enables dismissal for defendants at an early stage. Gottlieb Development v. Paramount Pictures Corp. presents a factual and legal analysis similar to Ringgold and Sandoval, except the court decided the de minimis defense on a motion to dismiss.²¹² In Gottlieb, artwork on the side of a pinball machine was visible in the background of a movie scene.²¹³ (See Image 2.) As with Ringgold and Sandoval, the court engages in an aesthetic assessment of the recognizability and context of the defendant’s use of the plaintiff’s art in the film. The court says plaintiff’s work was used “sporadically…always in the background…play[ed] no role in the plot,…[was] almost always partially obscured, and [was] fully visible for only a few seconds during the entire scene.”²¹⁴ The court concludes that the “`average lay observer’ would not be able to discern any distinctive elements of Gottlieb’s designs” and thus any copying fell “below the quantitative threshold of substantial similarity.”²¹⁵ Gottlieb cites both Sandoval and Ringgold for authority to dismiss the plaintiff’s copyright claim as “de minimis and therefore not actionable”²¹⁶ on the basis of the assessment of observability “factors such as focus, lighting, camera angles, and prominence.”²¹⁷ Gottlieb is a culmination of de minimis doctrine in the Contemporary Era achieving its goal of judicial efficiency.

These cases explain that whole copying can be de minimis. Additionally, however, they engage in more complex and nuanced aesthetic analyses to assess whether copying is both quantitatively and qualitatively below the infringement threshold, sometimes undermining a swift dismissal. These detailed aesthetic analyses featured in the Contemporary Era highlight the inevitable and oppugnant presence of aesthetic judgment in copyright cases, even in cases alleging only a very small or insignificant amount of copying.²¹⁸

The case of Davis v. Gap, Inc. further develops the third era’s doctrinal trend.²¹⁹ In Davis, the court rejects a de minimis defense in the context of an advertisement that features models donning the plaintiff’s “sculptured metallic ornamental wearable art” (ornamental eyeglasses) without his permission.²²⁰ (Image 3.) Judge Leval, writing for the court, determines that the “de minimis doctrine is not applicable” because “the infringing item is highly noticeable…in part because [they] are strikingly bizarre.”²²¹ It may be “a trivial matter for persons to be shown wearing their eyeglasses or wristwatches” in advertising, but not in the circumstances under review.²²² This is a remarkable statement in the context of the court’s other examples of de minimis copying (singing “Happy Birthday” or copying a cartoon to hang on a refrigerator).²²³ The defendant’s for-profit status cannot be the distinguishing factor, as defendants in Sandoval and Gottlieb were also for-profit (although they were feature films and not print advertisements for a clothing company). Perhaps the fact that the defendant’s print advertisement is static (not fleeting) and “widely displayed throughout the United States” makes a difference?²²⁴ It is otherwise hard to distinguish the situation in Davis from those in Sandoval and Gottlieb, except to take the court at its word and accept its aesthetic evaluation: the eyeglasses are distinctive and noticeable.

Each of these cases demonstrates how the de minimis standard evolves to include qualitative factors and begins to more closely resemble the substantial similarity test by including “recognizability” and “observability” as factors.²²⁵ Although courts still focus on the quantitative element (how much and for how long), the courts in the Contemporary Era also delve into a qualitative analysis with an aesthetic interpretation of the defendant’s use and its reasonable perception in the context of the whole work. The mixing of de minimis defense factors with the infringement standard is not an inevitable state or evolution of either doctrine, as earlier eras demonstrate. Commentators criticize this doctrinal shift as broadening the misappropriation standard and affecting the scope of substantive law.²²⁶ We agree. But as this Eras Tour shows, the de minimis defense was not always and is not consistently part of the “substantial similarity” inquiry. As Part III shows in more detail, it is often independent from the exclusive rights of reproduction and display, for example, and it is also frequently independent from the fair use defense.²²⁷ To align with its purpose and copyright policy, we urge the reclaiming of the de minimis doctrine’s independence with a focus on quantitative triviality (even in the case of whole copying) and qualitative insignificance.

2. VMG Salsoul and the Ninth Circuit

The Ninth Circuit’s 2016 decision VMG Salsoul v. Ciccone²²⁸ resembles the cases discussed above, except for two important features. First, it is a case between two music authors, rather than between parties in different copyright industries. And it is a de minimis case about fragmentary copying, not whole copying. Despite resembling cases from previous eras, it is an important case in this third era because its analysis of the de minimis defense includes both quantitative and qualitative factors, and it expands the application of the de minimis defense to sound recordings.

VMG Salsoul v. Ciccone involves Madonna Ciccone’s use in her song “Vogue” of a one-second sample from VMG’s sound recording. The sample was a single horn hit (a note or a chord) used four times.²²⁹ VMG Salsoul decision was a milestone de minimis case because it was the first court to split with the 2005 case of Bridgeport v. Dimension Films, which held sound recording infringement was immune to the de minimis defense.²³⁰ For a decade after Bridgeport, debates raged over whether Section 114 of the Copyright Act, directed to sound recordings specifically, precluded a de minimis defense.²³¹ The Ninth Circuit concluded that it did not.

Other than Bridgeport,…we are aware of no case that has held that the de minimis doctrine does not apply in a copyright infringement case. Instead, courts consistently have applied the rule in all cases alleging copyright infringement. Indeed, we stated in dictum…“that the rule applies throughout the law of copyright, including cases of music sampling.”…[N]othing in the neutrally worded statutory definition of “sound recordings” suggests that Congress intended to eliminate the de minimis exception.²³²

The court explains that Section 114 articulates the principle that “infringement takes place whenever all or any substantial portion of the actual sounds…are reproduced.”²³³ Therefore, copying that is less than “substantial” is not infringement.

The court held that the de minimis defense prevails because a single horn hit, repeated at most four times in a song, was not “substantial.” The court reaches this determination in several ways. First, it explains that the horn hit was unrecognizable to an average audience and thus was not actionable misappropriation.²³⁴ Second, the court cites its own de minimis precedent, Fisher v. Dees,²³⁵ and an early substantial similarity case from 1926, Dymow v. Bolton, to argue that de minimis copying combines quantitative triviality with unrecognizability.²³⁶ “As a rule, a taking is considered de minimis only if it is so meager and fragmentary that the average audience would not recognize the appropriation.”²³⁷ Cases about fragmentary and partial copying continue to cite the “substantive” standard as elaborated in VMG, but whether “recognizable” means anything more than “too small to notice” remains unclear. Sandoval and Gottlieb, discussed above, demonstrate that there are cases in which copying a whole work may not be recognizable to ordinary observers; however, such copying would fail to meet the “meager and fragmentary” requirement embedded in the VMG de minimis standard.

VMG evaluates de minimis largely as a quantitative test, while also attaching the more modern “recognizability” factor to measure substitutional value.²³⁸ The court’s quantitative emphasis reappears when it describes the copyright misappropriation doctrine as the right to prevent a “substantial copy of the whole or of a material part.”²³⁹ Audience observability or recognizability is relevant, the court explains, because without it “the copier has not benefitted from the original artist’s expressive content. Accordingly, there is no infringement.”²⁴⁰ The court thus implies that misappropriation requires copying that is “substantial,” and observability or recognizability is evidence in support of meeting that standard. But it is not true that recognizability alone is sufficient to find infringement, given the many de minimis cases in which copies are made and identified but are still deemed trivial.²⁴¹ In other words, not all recognizable or observable copying is misappropriation, even under VMG.

Rather than using the whole of plaintiff’s work in an insubstantial way,²⁴² the defendant in VMG used an insubstantial portion of the work – a fragment – in her song in a manner it could not be detected by an ordinary listener.²⁴³ The horn hit was a fragment of the plaintiff’s work, and that fragment as used in defendant’s sound recording was both unrecognizable and de minimis. The Ninth Circuit’s aesthetic analysis in VMG nonetheless followed that of the Second Circuit cases by asking how an average audience would perceive both the quantitative and qualitative aspects of the copying:

After listening to the audio recordings,…we conclude that a reasonable juror could not conclude that an average audience would recognize the appropriation of the horn hit….The horn hit is very short – less than a second,…occurs only a few times in Vogue….[and is] easy to miss. Moreover, the horn hits in Vogue do not sound identical to the horn hits in Love Break….[Furthermore,] a highly qualified and trained musician listened to the recordings with the express aim of discerning which parts of the song had been copied, and he could not do so accurately. An average audience would not do a better job.²⁴⁴

This assessment resembles a substantial similarity analysis but is distinct due to its sole consideration of the use of a fragment of a sound recording and the fundamental dissimilarity of the songs. “Vogue” is not a copy of ”Love Break.” VMG Salsoul thus resembles cases in the Early Era in which the defendant copied fragments of the plaintiff’s work (a photograph in a manual or words from a book) but in which the defendant’s and plaintiff’s respective works were not copies of each other and thus not market substitutes.²⁴⁵ Unlike the Early Era, the infringement standard now condemns partial (even fragmentary) copying as actionable infringement when the copy is “recognizable” as a measure of input-value—perhaps especially in musical work and sound recording cases. In VMG Salsoul, the fragment was undetectable and that was the end of the case.

A more recent Ninth Circuit decision, Bell v. Wilmott Storage Services, develops VMG‘s analysis in a way that conflicts with the Second Circuit cases discussed above.²⁴⁶ Specifically, it rejects the possibility of a “technical use” of another’s work as de minimis – a copy without harmful consequence. (This was the first iteration of the de minimis defense from Ringgold.²⁴⁷) In so doing, Bell leaves only fragmentary copying as plausibly insubstantial and thus non-infringing. This holding conflicts with case trends across the eras and ignores the long history of the de minimis doctrine.

Bell v. Wilmott Storage Services concerns the reproduction of a photograph of the Indianapolis skyline on a website. The defendant was unaware of the photograph, since it had recently acquired the website (along with thousands of others) in a database purchase; moreover, the site was inaccessible to most internet users.²⁴⁸ The site mostly consisted of text and had few photos; it comprised a part of a server database associated with a website called VisitUSA.com.²⁴⁹ According to the court, the photograph “was only accessible to those users who conducted a reverse image search—as Bell had—or those who knew the precise address of the image database archiving the photograph.”²⁵⁰

Given this context, there was little that a qualitative or aesthetic analysis of the kind in Ringgold and VMG could accomplish. Instead, the court relies on the fact that “the `degree of copying’ was total—the infringing work was an identical copy of the copyrighted Indianapolis photo. There is thus no place for an inquiry as to whether there was de minimis copying.”²⁵¹ The court distinguishes earlier cases, including Fisher v. Dees and VMG Salsoul as partial-copying cases.²⁵² Both are inapposite, the court explains, because in Bell the whole photograph was copied and was fully and clearly displayed on the website, however inaccessible it was to most ordinary internet users.²⁵³

The defendant contends that, although the whole photograph was copied and displayed, no cognizable harm occurred and litigation was therefore unwarranted. ²⁵⁴ This de minimis defense relies on several facts. First, the website was inaccessible and the photograph hard to find.²⁵⁵ Second, the copying was unintentional and without value to the defendant.²⁵⁶ The court rejected the relevance of these facts, stating that the de minimis defense does not relate to insignificant or minimal use (as in only one time on an obscure server database).²⁵⁷ It instead relates to the “quality =and quantity of the protected work that was `used’ by the defendant to make the allegedly infringing copy.”²⁵⁸ This conclusion is incorrect, ignores relevant context, and is inconsistent with most other de minimis cases, including those in the Contemporary Era.

Wilmott‘s restatement of the de minimis doctrine purposefully and directly disregards earlier cases concerning “technical use” as de minimis (cases that Ringgold said comprised category (1)).²⁵⁹ These past cases expressly considered whether the extent of the use is “trifling” in terms of the market harm or substitutionary effect and not worth the costs of litigation.²⁶⁰ Why did the Wilmott court cast aside this early history of the de minimis defense? Perhaps because more recent cases, including VMG, appear to anchor infringement and the “substantiality” determination in terms of “recognizability.” As the Wilmott court says “we have consistently applied the de minimis principle to determine whether a work is infringing by analyzing the quantity and quality of the copying to determine if the allegedly infringing work is a recognizable copy of the original work, in other words, whether the works are substantially similar.”²⁶¹

This is an inaccurate statement of the de minimis doctrine and is a contestable statement of the substantial similarity test. ²⁶² One problem with this formulation is that it prioritizes de minimis copying that is partial, meager or fragmentary over whole copying that occurs only once, infrequently, or is otherwise immaterial. As the Eras Tour shows and Part III further elaborates, the de minimis defense applies to both whole and partial copying. Indeed, the de minimis defense arises frequently when the plaintiff claims unauthorized copying of their whole work – in as much as 40% of all the cases and with more frequency since the late 1990s. There is no doctrinal reason why whole copying should be any less defensible than partial copying. It is the nature of the harm that matters, not whether copying occurred. ²⁶³ Another problem is that Bell v. Wilmott fails to account for harm to the Plaintiff in the specific context of Defendant’s use. It is therefore out of line with the overall trend in de minimis cases, which consider substitutional harm and materiality of copying to measure whether the plaintiff’s case is worth the court’s time in the first instance. The Wilmott court’s rigid reading of the Copyright Act as precluding “technical copying” is wrong as a matter of both doctrine and fact. Indeed, as we explain infra, defendants prevail at approximately the same rates when raising a de minimis defense to the unauthorized copy of a whole work as with a fragment of plaintiff’s work.²⁶⁴ Nonetheless, the Wilmott court rejected both Ringgold and Davis that embraced “technical violations” as de minimis.²⁶⁵

We reject [defendant’s] `technical violation’ theory of a de minimis defense…Copyright is a creature of statute, and the only rights that exist under copyright law are those granted by statute. Where an unauthorized material use of the copyrighted work does fall within one of those exclusive rights, infringement occurs, unless the use is excused by one of the privileges, exemptions, or compulsory licenses found in sections 107 through 122….A `technical violation’ defense…is plainly in tension with this right because it suggests that making a single copy is somehow not enough to show a violation absent some further material use or action. But crucially, the Act is agnostic as to the use of the copy once it is made; the unlicensed copying itself is the violation.²⁶⁶

Bell v. Wilmott recharacterizes whole copying as never de minimis, despite prior cases that say the opposite.²⁶⁷ It attempts to narrows the doctrine to situations involving only fragments of a copyrighted work and rejects the common-law understanding of de minimis as a trifling use unworthy of judicial attention, solely because the Copyright Act contains no explicit defense. ^{268 269} Bell marks a doctrinal break from copyright history and policy, disregarding the long tradition of excusing even complete copies when the resulting harm was trivial. This history is especially relevant in the internet age when whole but fleeting copies are made with ubiquity as part of digital communications.

Part III analyzes the dataset in detail to show that while the use of the de minimis defense has grown, its success rate has not. And yet the idea of trivial copying is more vital than ever and remains a total defense to copyright infringement. Plaintiff’s contemporary complaints involving (what in earlier eras might have been) trivial copying and dismissed early in the case, now occurs simultaneous with court’s more complicated analyses of the de minimis defense, infringement, and occasionally fair use for drawn-out litigation. The result is that what qualifies as trivial copying appears to be neither judiciously resolved nor as straightforward in the Contemporary Era.²⁷⁰ But as we also show in Part III, Wilmott remains an outlier. It is true that the de minimis doctrine has come to resemble the infringement analysis with more frequency, with smaller parts of works more relevant to misappropriation claims and whole copies sometimes more easily deemed infringing. But the de minimis defense is not defunct. To the contrary, it is even more relevant for defendants in a certain category of cases – whole copying of pictorial works, especially non-photos – that resemble the technical violations Wilmott rejects. It is to these features of the data set we now turn.

III. The Data

This Part describes the de minimis case data set, the manner of its collection, and its coding. It then disaggregates the different kinds of de minimis copying across the three eras and relates them to evolving legal standards, party identities, nature of the works, and fair use claims. Finally, this Part identifies the circumstances under which a de minimis defense is more likely to prevail in the Contemporary Era. Drawing on this trend, it then proposes a clarified de minimis standard to revitalize its judicial economy purpose and conform with copyright policy.

A. Collecting Cases [Figures 1-4]

Litigated cases described here are a very small slice of litigated copyright disputes and transactions in which the de minimis defense is relevant. The nature of the defense is such that most de minimis uses are not ever litigated. As such, we do not capture in our dataset the many circumstances in which cases are settled before or shortly after a case is filed. We assume, given the nature of the de minimis doctrine, those circumstances far outnumber the cases in our dataset.²⁷¹ This study does not cover all litigated cases in which the defense is raised, only those reported in Westlaw in which the defense was mentioned in a written court opinion. Furthermore, Westlaw is not comprehensive, especially in earlier decades.²⁷² As such, this Article’s data cannot be used to draw conclusions about cases where a de minimis challenge was raised but never discussed by a court. Nonetheless, litigated cases – and especially on-line available court opinions – influence legal doctrine’s application and development.²⁷³ Lawyers rely on litigated cases, such as those described supra in the Contemporary Era to advise clients. A significant aspect of legal practice concerns understanding what those cases say and how what they say has changed over time.

The data set includes 178 court opinions.²⁷⁴ Our initial search resulted in 710 cases in which the terms “de minimis” and “copyright” co-occurred in one paragraph in the ALLCASES database. (We used other search terms for “de minimis,” such as “deminimis,” “de minimus,” and “deminimus” due to spelling variations.) We extracted from this large set cases in which de minimis applied to subject matter analysis (whether a de minimis amount of creativity existed in the purportedly authored work) and to damages analysis (after a plaintiff-friendly infringement analysis). We were only interested in copyright cases in which the de minimis defense arose to undermine an infringement claim in the context of the indisputable copying of a copyrighted work.

To check for completeness, we ran several other searches. We ran a search for “trifl! /p copyright! % “de minimis,”” which returned only 36 cases, of which one was relevant. We added that case to the database.²⁷⁵ We also ran a search for “insubstantial! /p copyright! % “de minimis”” that returned 346 cases. A randomized analysis of those cases returned no relevant cases.²⁷⁶ And we ran a search for “infring! /p substant! /p copy /p whole /p material! /p part,” derived from an early standard for de minimis, which returned 56 cases, none of which were relevant.²⁷⁷ We also found an additional three cases by following citations to West Publishing Co. v. Edward Thompson Co. (E.D.N.Y. 1909), which appears to inaugurate an early version of the de minimis defense without using that phrase.²⁷⁸ We added those three cases and West Publishing to our database as well.

For the purposes of this Article, we call this final set of 178 the “de minimis cases.”²⁷⁹ The cases range in time from 1847-2022.

Below are four graphic representations of the total number of de minimis cases in our data set. The take-away is that the de minimis defense to copyright infringement has been on the rise since the first decade of the twenty-first century. But that trend began with the 1976 Copyright Act and the Transition Era. Specifically, de minimis decisions hovered between 2-4% of all copyright decisions for 70 years and then doubled to 8% for the years 2011-2020. We cannot know the cause of this rise, but we hypothesize it is due to several factors making the de minimis defense all the more important, including: the broadening of the copyright cause of action as previously described, the availability of statutory damages making litigation more attractive even for weak claims to promote settlement, and the rise of the internet facilitating all sorts of copying both trivial and nontrivial.

Figure 1 is the number of de minimis case decisions over time. There were only 21 cases in the Early Era, which ends in 1978.²⁸⁰ There were 157 cases after that date. Due to the small number of observations, the data is bucketed in decades and later five years to highlight the growth over time. The buckets have a break between 1977-1978 to indicate the number of case decisions before and after the Copyright Act of 1976.

Figure 2 shows the same data as Figure 1 but by Era. The Contemporary Era dominates the de minimis dataset. This comports with our hypothesis that the internet age has driven the explosion of de minimis cases, highlighting the question whether prolific but trivial copying as part of our everyday communications and transactions requires adjustments to copyright law, including the scope of defenses.

We also measured the rate of de minimis decisions in relation to copyright decisions generally. Figure 3 shows the raw number of de minimis decisions in relation to copyright decisions from 1941 to 2020. Notably, in the first decade of the twenty-first century, de minimis decisions begin to increase as copyright decisions decrease.²⁸¹ The number of copyright (blue) and de minimis (red) decisions over time highlight and compare the trends over time. This confirms the rise in relevance of the de minimis defense to copyright defendants in the Contemporary Era. As unauthorized copying becomes an inevitable part of everyday life, fewer copyright lawsuits are filed but more of them contain a de minimis defense, justifying the copying as trivial.

We visualize this upward trend of de minimis decisions in Figure 4, which shows the de minimis case decisions as a percent of all copyright case decisions from 1941 to 2020. De minimis decisions hovered between 2-4% of all copyright decisions for 70 years and then doubled to 8% for 2011-2020. This shows that the de minimis case decisions made up a larger share of copyright decisions in court.²⁸² When combined with Figures 2 and 3, this data shows not only the rise of legal relevance of the de minimis defense to copyright defendants in the internet age, but also the relevance of the defense to courts deciding copyright cases. As judges write more opinions on de minimis challenges over time, varying the standards and their application in particular cases, there is need for clearer standards and doctrinal predictability to serve the doctrine’s purposes of judicial economy. Judicial economy is not the same over time; as cases become more complex with different kinds of parties asserting different kinds of works, as is true in the Contemporary Era, concerns of judicial economy should increase and de minimis claims become more salient.

B. Coding Cases

A team of four research assistants and one author (Silbey) coded the case facts and their doctrinal details. At first, we coded the same cases and met once a week during the coding to check intercoder reliability and discuss questions raised. After several weeks, cases were divided among four research assistants and each coded only their cases. One of us (Silbey) reviewed all coded cases independently. At the conclusion of the coding, each research assistant wrote a memo about their cases’ themes and outliers.

We coded for a variety of facts and doctrinal details.²⁸³ We coded for: the kind of work (both plaintiff’s and defendant’s);²⁸⁴ the stage of litigation at which de minimis was discussed; form of de minimis (e.g., fragments, special parts, or whole, and sub-categories therein);²⁸⁵ the nature of the court’s de minimis analysis (legal standards, quantitative or qualitative thresholds);²⁸⁶ the identity of the plaintiff and defendant;²⁸⁷ presence or interaction with fair use; and success or failure of the de minimis defense.²⁸⁸

1. Types of Works and Case Pairs [Fig. 5-6.2]

Many of these categories required careful delineation for coding purposes. For example, because we hypothesized that the kind of works at issue might affect the de minimis analysis, we expected to see patterns arise concerning works that are more easily fragmented (and thus de minimis more available as a defense). Since copyright law does not define a “work,” and some regularly appearing works are also left undefined in the Copyright Act, we had decisions to make about genres, forms, and boundaries.²⁸⁹

Parties frequently debate the boundaries of the work, often to their strategic advantage, especially when the work contains parts that could be separately copyrightable (such as screen shots, characters, or chapters).²⁹⁰ We sought to identify those strategies when they occurred, and so divided works along these lines even if not part of the Copyright Act (e.g., “article” separate from “book” whereas the Copyright Act speaks only of “literary work”). We categorized each work in terms of its most reasonably appreciated whole and in terms of the work that the author originally conceived, even if the part that was copied could be independently copyrightable.²⁹¹ For example, when defendant copied the pictorial or graphic works within plaintiff’s manufacturing manual, we coded that as a manual (for plaintiff); if defendant was using the images in its manual, that was coded as a manual as well. This accurately captures the kind of lawsuit it is (between similar companies preparing similar products). In later eras, and especially in the Contemporary Era, the same dispute is more likely to be characterized as between a Plaintiff’s pictorial work (if Plaintiff was the individual author and owner of the image) and a Defendant’s manual.

The evolution in the Contemporary Era of disputes between putative licensees and licensors, as opposed to between industry competitors, is captured by this shift in the nature of works the parties claim. In the Contemporary Era, it is more common for photographers or visual artists to sue as independent authors for use of their works as input into books or audiovisual works. These cases were coded in relation to the authorial claim of each (e.g., visual artist versus movie producer), and we understand these cases to represent a different kind of lawsuit than in prior eras – not between competitors but between potential licensees. After the Ringgold decision in 1998, we see a rise in these different lawsuits in which the works are disaggregated, especially with regard to pictorial works, whole works (not fragments), and when the lawsuit is between potential licensees as opposed to industry competitors. These kinds of lawsuits may lend themselves more often to de minimis defenses as defendants contend that small-value inputs into a larger work (e.g., a background poster for a television sitcom set) is not copyright infringement, because that was true in earlier eras. Of course, whether it is a de minimis copy or copyright infringement is precisely the question these new lawsuits in a new era raise. And the varying approaches in the Contemporary Era fail to clarify the question.

Figure 5 describes litigation pairs by era showing this pattern of rising lawsuits between putative licensees-licensors. If a plaintiff claims a defendant’s book included a copy of an excerpt from plaintiff’s book, that was coded as Book_Book. The vast number of cases still involve plaintiff and defendant authors of the same kind of work. We believe this approximates direct competition in the marketplace, with de minimis claims focusing on whether substitutional harms exist, a bedrock copyright complaint. The post-Ringgold period (and the Contemporary Era) includes many more cases involving plaintiffs and defendants authoring different kinds of work. We believe these cases are less obviously about direct competition and instead reflect a new plaintiff strategy to seek licensing revenue for uses that were either previously considered de minimis, fair use, or not substitutional copies at all (e.g., not substantially similar copies). The vast majority of these new cases are brought by individual plaintiff authors of pictorial, graphic, and sculptural works (“PGS” works²⁹²) against defendant authors of movies, television shows, and videos. These kinds of cases grew from nothing in the Early Era to 11% of the cases in the Contemporary Era (the middle bar in Figure 5).

In general, we found that the distribution of the kinds of works over time was related to the prevalence and growth of copyright industries at the beginning of the internet age. For example, we see more music cases after 1993 (although they existed in earlier eras). And computer code cases begin in the late 1980s and rise in the mid-2000s, when personal computers became available and the industry rapidly grew.²⁹³ Claims by pictorial work authors begin to rise in the mid-1990s and spike in the mid-2000s, with photography cases in particular growing in number. Not only are pictorial works easier to copy and distribute in the internet age, but the image industries (photography and visual arts) are booming currently and are populated by independent and freelance authors who claim copyright in their work and who license them to media outlets for publication.²⁹⁴ As Figure 5a shows, the percent of cases involving similar parties have decreased over the Eras with less than half the cases (45%) in the Contemporary Era involving similar parties. These cases are getting replaced more often by those brought by individual plaintiffs. Further, individuals suing other individuals are 21 of those “Similar Parties,” confirming our hypothesis that independent and individual authorship of different kinds of works, which become parts of and input into other works, may fuel these disputes in which de minimis defense is salient.

Figures 6.1 and 6.2 shows the percent of plaintiff and defendant works over the total number of cases over time, bucketed by eras.²⁹⁵ The main observation with these two Figures demonstrating the growth of new industries is that the types of works involved in these cases become more varied over time for both plaintiffs and defendants. The Early Era is dominated by books, lists, articles, and music. The Transition Era sees a rise of computer code, PGS works, and television. The Contemporary Era includes the whole range of works, but most prevalent are computer code, video, and PGS works. Over time, articles, books, and lists appear less often in de minimis cases. We interpret this trend as confirming the utility of the de minimis defense to efficiently resolve disputes between legacy copyright industries common in the Early Era. The prevalence of newer industries in the later eras in which de minimis is more frequently debated suggests that norms of permitted copying and required licensing in new industries may still be developing and judges struggle to arrive at a clear standard, increasing opportunism among litigants. Increased variety and nascent, ambiguous copying/licensing norms negatively implicates the de minimis doctrine’s judicial economy rational. As the next section demonstrates, norms from the Early Era about whole copying in a quantitatively trivial manner begin to take root in the Contemporary Era especially with regard to PGS works, which is a laudable development. But the doctrine’s growing complexity still challenges its underlying rationale and requires further clarification.

2. Forms of De Minimis [Fig. 7.1, 7.2, 8]

Because we hypothesized that the form of de minimis might matter to the court’s analysis, we coded for three forms measured quantitatively: “fragment” “some”, and “whole” (described further below and in Appendix Table 2). And then we further coded for specific forms identified qualitatively, such as photographs, videos, textual sentences, etc. (also described further below and in Appendix Table 3).

As explained, whole copying of a work dramatically increased over time, first appearing with any regularity in the Transition Era and outpacing other kinds of partial copying (fragmented or some) in the Contemporary Era. Figure 7.1 shows this distribution bucketed by eras. It shows how the categories of “some” and “whole” completely change relative prevalence between the Early Era and the Transition Era. In the Early Era, “some” copying was 57% of all cases and “whole” was 19%. By the Contemporary Era, “some” was only 28% and “whole” was 46%. There is reversion back to the mean in the Contemporary Era for “fragments” at 26% of all cases in the eras. The two take-aways from this distribution are (1) copying the whole work and asserting a de minimis defense has become much more common over time; and (2) when combined with defendant win rates, fragments and whole copying are defendant-friendly factors in an infringement suit when alleging de minimis defense.²⁹⁶ That whole copying can be (and frequently is) adjudicated to be de minimis is an important commendable, and surprising feature of copyright law in the internet age. Courts and litigants should take note.

Figure 7.2 shows this same distribution of de minimis forms roughly by decades. As with Figure 7.1, in Figure 7.2 the percent of de minimis case decisions controls for the growing trend of de minimis claims over time. Whole works did not appear until the 1960s, presumably because by then copying technology was popularized and accessible. Whole works have remained a consistent and growing percentage of de minimis cases. As we explain infra regarding prevailing defendants, although fragmentary copying remains the predominant way for a defendant to win when asserting a de minimis defense, courts also appear solicitous of defendants who make whole copies of pictorial works in certain contexts, separate and apart from a fair use defense.²⁹⁷ Notably, the rise of whole copying cases in which de minimis is raised as a defense and succeeds undermines the account that Ringgold and the doctrinal developments in the Contemporary Era have made it harder for defendants to prevail. We consider this a significant corrective to the narrative that Ringgold’s complex and time-consuming approach governs de minimis analyses, opening up more opportunities to build upon a trend that is economical and comports with copyright law and policy.

We hypothesized that specific kinds of fragments, parts, or works, might make a difference in the prevalence and success of defendant’s de minimis defense. ²⁹⁸ As Figure 8 shows, sentences (including lede sentences and single pages) dominated the Early Era with 62% of the cases, but have become much less prevalent, reducing to 21% of cases in the Contemporary Era.²⁹⁹ These early cases were about titles, headings, lines of code, and short sentences (including definitions, such as in informational works).³⁰⁰ Taking the place of these kinds of cases in the Transition and Contemporary Eras are cases involving the whole copying of pictorial works and phrases (including musical phrases and visual phrases as ideas and stock images).³⁰¹ Cases concerning pictorial works went from 10% of cases to 34% over the eras. And cases concerning phrases comprised 5% of the Early Era cases as compared to 22% of the Contemporary Era cases. These are cases such as Ringgold and Gottlieb (pictorial works) and the many musical copyright cases (musical phrases), such as VMG Salsoul. The works subject to “entire” copying in this Figure 8 does not include two-dimensional visual art (such as images) but does include sculpture, audiovisual works (television and film), and literary works such as computer programs and manuals.³⁰² The majority of whole and entire work cases in the Early Era occur in the last decade, as Figure 7.2 indicates.

To us, Figure 8 confirms several themes arising in the dataset: (1) the growing diversity of plaintiff works involved in de minimis cases in the Contemporary Era; (2) the rise of de minimis cases concerning pictorial works in the internet age and Contemporary Era (usually asserted against defendants who are putative licensees, not industry competitors); and (3) the shrinking of de minimis cases in which defendants are accused of copying textual fragments of the plaintiff’s work, which is how the de minimis doctrine began. One way to think of this evolution is that a new kind of de minimis case is arising in the Contemporary Era: whole copies of image-based works used in audiovisual or literary works. And these new cases are a version of the old de minimis case in which textual quotation or referral is common and acceptable. This is a critical new feature of the de minimis doctrine because today we communicate as much by images as by words.³⁰³ Allowing speakers to copy and paste images in order to communicate, without being subject to copyright infringement, is speech-enhancing and effectively harmonizes copyright law and the First Amendment. We understand this trend to be an admirable adaptation of Early Era de minimis decisions that is consistent with copyright law and policy: immunizing from frivolous lawsuits defendant’s quantitatively trivial and qualitatively insignificant copying while also promoting knowledge through distribution of expressive works.

3. Legal Standards & Evaluative Factors [Fig. 9-11]

How do courts assess de minimis and what factors do they consider in a de minimis determination? Because so many cases contain only a conclusory assessment of the de minimis defense, there was often nothing to code. We nonetheless sought to understand how many of the cases in the data were perfunctory judicial hand-waves (as either de minimis or not). And we wanted to understand the details of courts’ de minimis analyses when they were more complex. So, when courts engaged in more specific analysis, even if only a sentence or small paragraph, we looked closely at the factual and legal considerations.

We identified six general approaches to the de minimis defense and tracked them in terms of their substance and predominance. They included: (1) no standard at all; (2) a simple statement of “de minimis non curat lex”; (3) a quantitative assessment; (4) an assessment of aesthetic significance, which we considered a qualitative measure; (5) consideration of financial investment or harm; and (6) analysis resembling “substantial similarity,” which often included “recognizability” or “observability” as a key factor.³⁰⁴

Figure 9 visualizes the range and distribution of standards over the eras.³⁰⁵ This data analysis confirms three themes from the previous discussion. (1) Courts in the Contemporary Era engage more diverse legal standards per case than in the past, which is why the Contemporary Era’s bar goes past 120%, confirming that the de minimis analysis has become more complex over time. (2) The largest change to the de minimis analysis is the incorporation of a standard that resembles a “substantial similarity” infringement analysis, growing from 5% in the Early Era to 46% in the Contemporary Era (“Recognizability/SS”). And, (3) the perfunctory quantitative origins of de minimis (“Amount/FLS”) remain constant over the three eras and continue to account for a majority of the cases. Additionally, financial investment rationales for assessing de minimis decrease from 24% to 6% between the Early and Contemporary Era (this is the “Labor Value/Accessibility” factor). We attribute this in part to the milestone case of Feist Publication Inc. v. Rural Telephone Services, 499 U.S. 340 (1991) – itself a de minimis case – which ended copyright’s sweat-of-the-brow doctrine and explained that the “primary objective of copyright is not to reward the labor of authors, but to promote the progress of science and the useful arts.”³⁰⁶ The upshot is that de minimis is becoming more a complicated doctrine (and thus less efficient), undermining a key purpose of the defense. But also, importantly, some categories of cases are ripe for quick de minimis dismissals and identifying them is key to maintaining the defense’s vitality in the digital age.

Before evaluating the trends in prevailing de minimis defense cases – i.e., which of these standards appears more frequently when defendant wins – we dig further into the quantitative and qualitative metrics that courts consider as part of a de minimis defense. The discussions in these cases resemble aesthetic analyses of defendant’s use of plaintiff’s works. Despite courts’ admonitions that judges should avoid art criticism,³⁰⁷ copyright law is shot through with aesthetic questions.³⁰⁸ The de minimis decisions evaluating quantitative and qualitative metrics are aesthetic analyses, and they highlight how much aesthetic considerations may matter even in cases alleging only trivial copying.

We were not surprised by the rise of more complex de minimis analyses, given that we hypothesized that Ringgold and Gottlieb were landmark cases for the evolving doctrine. However, we were surprised by the distribution of quantitative metrics over time.³⁰⁹ As Figure 10 shows, the fraction of cases considering whole copying is similar in the Early and Contemporary Eras at around 20% (and is less in the Transition Era at 13%). The number of cases considering partial copying is also similar between the Early and Contemporary Eras at around 15%, and higher in the Transition Era (at 27%). The second era therefore stands out as a transition era for these quantitative metrics, and so appropriately named. Factors that decline over time include repetitious use of small amounts (“Repetition”), dwindling from 14% in the Early Era to 3% in the Contemporary Era. Courts appear to discuss the de minimis defense less frequently when defendants make small but repetitious use of copyrighted material. And courts are less likely to consider specific fractional amounts of copying (“Amount”) as relevant to the de minimis determination over time, declining from 43% in the Early Era to 33% in the Contemporary Era. Fractional amounts nonetheless remain a significant factor in the cases (about a third of de minimis analysis contain them).³¹⁰ The element that increases over time, is the absence of a quantitative metric (“No Metrics”), which may be one way to explain the relative waning success of the de minimis defense in the Contemporary Era.³¹¹ A third of cases are more likely to have no quantitative metric at all in the Contemporary Era at 34% as compared to 24% in the Early Era, and this bodes badly for defendants. As we explain infra, the presence of a quantitative metric helps defendants in borderline cases, especially in the case of copying a PGS work.³¹² Given that the de minimis doctrine began with quantitative assessments of triviality and its goal is to be an efficient equitable defense, it makes sense for both courts and litigants to anchor the analysis in quantitative metrics.

Qualitative metrics follow a similar pattern, with some exceptions.³¹³ As Figure 11 shows, 37% of cases in the Contemporary Era use no metric at all to evaluate the de minimis defense (“No Metrics”). This is a decrease from the Early Era in which more than half (52%) of cases contained no qualitative metric. This is not surprising given the de minimis defense began as an assessment of quantitative triviality in terms of fragments and small parts of larger works; courts infrequently considered relevant the aesthetic qualities of works. When courts do consider meaning or message of the work copied, the next common factor is “relevance” – when courts consider the nature of the work copied salient to the meaning or value of the defendant’s work. The “relevance” factor ranges from 23% to 37% over the eras and approximates courts’ substantive assessment of “triviality.” We believe this reflects court’s concern with market substitution for Plaintiffs expressive form. Qualitative assessments concerning alterations to the plaintiff’s or defendant’s work only appear in the Contemporary Era (Alterations). This makes sense given how much harder it would have been in other eras to copy and alter another author’s work and insert it into one’s own. Perhaps more surprising – but consistent with the doctrinal evolution described supra in Part II – is the rise in qualitative metrics of “recognizability” and “observability” from 10% in the Early Era to 36% in the Contemporary Era. This follows the trend of co-mingling the de minimis with the substantial similarity analysis. And it raises the important question whether in some large number of “trivial” copying cases, the de minimis standard and the substantial similarity standard are insufficiently doctrinally distinct. Their co-mingling is an obstacle to achieving the de minimis doctrine’s goal of judicial efficiency.

Despite the trend of co-mingling de minimis with infringement analyses, as much as between a majority and a third of the cases do not discuss qualitative factors at all. Ringgold‘s in-depth analysis that resembles the substantial similarity test is thus still a doctrinal minority across all cases. And, overall, courts more frequently frame the de minimis analysis in terms of no metric and/or quantitative metrics.³¹⁴ A more rigorous qualitative analysis resembling the substantial similarity test only arises with relative frequency in the second half of the Transition Era. By the Contemporary Era, a qualitative evaluation is one of several analytic options. This shows that Ringgold and its progeny may have shifted the way de minimis defense operates, but only in a subset of cases. The more qualitative and aesthetically complex analysis is neither the default standard nor the only one. Furthermore, as we discuss below, the other de minimis standards are more defendant-friendly.³¹⁵ For the de minimis defense to remain independent and effective, therefore, courts and lawyers must appreciate the range of de minimis standards and the optimum contexts for their use.

4. Case Outcomes & Fair Use Interaction [Fig. 12-21]

A goal of this research was to learn under what circumstances the de minimis defense succeeds and if it succeeds at the earliest possible stages of litigation, as the defense optimally demands. To this end, we coded for the success or failure of the de minimis defense, its interrelation with the existence of a fair use defense, and the stage of litigation at which the de minimis analysis is dispositive.

As seen in Figure 12 below, only a portion of cases raising a de minimis defense also raise a fair use defense, and in the Contemporary Era the overlap in defenses is the smallest of all eras. This supports a determination of the independent relevance of the de minimis defense from fair use. Although the de minimis defense developed from fair use in the 1840s, as described in Part I supra, in a majority of cases since then it has been alleged independently.

This bodes well for the early intervention of de minimis in an appropriate case. The relative lack of overlap in the dataset between de minimis and fair use may also send the wrong signal that raising both defenses is inconsistent or risky for defendants. Given the de minimis data set clearly demonstrates the de minimis defense’s independent structure and role, lawyers should not avoid raising both defenses and courts should not default to the more complicated fair use analysis in the presence of a simultaneous de minimis defense. For de minimis to effectuate judicial economy, it must be available early in the litigation and, if inappropriate for dismissal, its early assertion should not preclude other defenses later in litigation.

In terms of win-rates,³¹⁶ as seen in Figure 13 below, de minimis claims prevailed at similar rates in the Early and Contemporary Eras (about 25%); and de minimis also prevailed over fair use raised in the same case in both eras (although in the Contemporary Era the rates of success are close).³¹⁷ The Transition Era follows a different pattern. Both de minimis and fair use (when claimed in a de minimis case) success rates are higher, 37% and 56% respectively. The rise in fair use successes over and above the de minimis defense in the Transition Era may be explained by fair use’s recent statutory codification in 1976. It is possible that courts were relying on new and express statutory fair use factors (which includes a factor for “amount and substantiality”) instead of an implied, albeit deeply rooted, common law de minimis defense to dispose of cases. Also notable in Figure 13 is that when (what eventually becomes) the Ringgold standard for evaluating the de minimis defense (SS Standard) rises in prevalence from the Early to the Contemporary Era, the rate of successful de minimis defenses reverts to its Early Era level. No matter, each are independently viable defenses to be raised optimally at different stages of litigation, which is an important observation given the de minimis defense appeared to come from fair use.

The de minimis defense is not often successful in court. In only 49 of the 178 cases did the de minimis defense prevail.³¹⁸ Figure 14 below compares the total number of de minimis cases with the fraction of those in which the defense prevailed. As shown in the figure, the rate of success is fairly constant over the century but then slows in the Contemporary Era around 2003 (when the two lines drastically diverge), which corresponds to the early days of the internet’s popularity.

We tried to understand this data regarding win/loss rates from different perspectives. We did so first with regard to the nature of the de minimis copying (whole, fragment, or some). We predicted that fragmentary or partial copying was more likely to be de minimis; and that whole copying (especially of pictorial works, such as photographs or other visual art), which was more common in the internet age, was less likely to be de minimis. Indeed, fragmentary and partial copying is more likely to be de minimis, but surprisingly whole copying is also determined to be de minimis with relative frequency, albeit under a different legal standard. We rely on this trend to structure our proposal in Part IV for a clarified de minimis defense.

In Figure 15 below, we see in the Early Era all the cases in which de minimis prevails concern fragmentary copying (and none are PGS works). In the Transition Era, prevailing de minimis cases are spread between fragments, partial, and whole copies, with fragments being the form that most frequently prevails. Notably, across eras all but one prevailing de minimis case concerning a whole work involved PGS works.³¹⁹ In the Contemporary Era, successful de minimis cases are also spread between all three forms, but fragmentary and whole copying rival each other for prevalence and “whole” copying prevails more often than in the past. Again, PGS works predominate the “whole” copying category. PGS arrive later in the de minimis data set and PGS plaintiffs lose at higher frequency.

What other factors correlate with a successful de minimis defense? In general, when courts assess a defendant’s copying in a quantitative manner (the “Few/Some” and “Amount” bars below in Figure 16), the defendant is more likely to win a de minimis defense.³²⁰ In particular, courts’ identification of specific amounts (fractions, percentages, or quantifiable parts) correlates with successful de minimis defenses. If there is no quantitative metric (“NS” below), the de minimis claim is more likely to fail than with any other metric. In other words, a de minimis defense has better odds of winning if there is an amount mentioned. From this data, we understand that quantifiability is crucial to a successful de minimis defense, which makes sense both historically and normatively. And we urge defendants and courts to draw on quantitative metrics to justify a dismissal on the basis of de minimis copying, which is consistent with case law and the roots of the common law doctrine.

Although de minimis claims lose more often than they prevail, we see this same pattern of preferences for quantitative metrics when we evaluate legal standards in general. As Figure 17 shows, the de minimis defense standard that has a better chance of success is quantitatively oriented (15% of all cases). Recognizability/Substantial Similarity (8%) and the general “de minimis non curat lex” (DM, 6%) are the next most successful. The de minimis standard of “Labor Value/Accessibility” also has a high chance of success relative to the cases in which it is raised. Three of these evaluative standards (but not “Recognizability”) originate in the Early Era and persist in the Contemporary Era. We think they reflect a historically accurate conception of de minimis as a judicially efficacious defense that ideally short-cuts Plaintiff’s claims early in litigation when defendant’s unauthorized copying is not a market substitute for Plaintiff’s work and thus causes no copyright harm.

The combination of the data from the above three Figures might seem to lead to the conclusion that whole copying is disfavored overall, even if more common in the Contemporary Era. But the picture is more complicated. Whole works show up in conjunction with all six possible standards, see Figure 18 below, but courts analyze whole copying under the “de minimis non curat lex” standard 2.5 times more often than they do for partial copying. (This general standard originates from the doctrine’s earliest days and is described in Ringgold as a “technical violation.”) Otherwise, the courts view the forms similarly across the other legal standards. As shown above in Figure 17, this general standard (DM) frequently drives successful de minimis defenses. We learn from this that certain forms of de minimis (whole copying) may be more likely to prevail under certain legal standards even without close or complex multi-factor analysis. This fact is worth emphasizing given the importance of early and efficient dismissals of non-meritorious copyright cases.

When we dug deeper into the cases concerning whole copying, we learned that successful de minimis cases involving whole PGS works most commonly involve a general standard (DM). This was surprising given the cases of Bell v. Wilmott, Davis v. Gap, Inc., and Ringgold, which feature whole copies of plaintiff’s PGS works in which the defendant lost. Despite these well-known cases, although the majority of prevailing defendants copy only fragments or parts of the plaintiff’s work, when the defendant copies a whole PGS work they are also more likely to succeed than in other situations. Litigants and courts should pay attention to this trend and avoid over-emphasizing these renowned cases. They may be notorious cases precisely because they are lengthy and complex analyses of a doctrine that is meant to be a quick off-ramp for defendants and thus the cases are anomalies, not the norm.

Below in Figure 19, we see the de minimis cases involving PGS works (56 in total), those that succeed, and under which standard. Although these PGS cases only prevail 25% of the time (14 of the 56 cases, 13 of which are “Whole” copies), the most common standards mentioned are quantitative, general, or “Labor Value/Access.” This suggests that courts are still assessing de minimis under a “technical violation” standard and are doing so more frequently in the context of PGS works. This reflects an importation of the origins of de minimis into the internet age in which PGS copying is more frequent but resembles the assessment of harmless but recognizable copying from the Early Era.

Finally, below are two logit regressions³²¹ that feature the factors associated with a prevailing de minimis claim. As Figure 20 demonstrates, one of the most important factors for a prevailing de minimis defense is whether the plaintiff’s work is a PGS work (but not a photograph), after controlling for other factors of the case. This is surprising because it contradicts the theory that cases like Ringgold determine de minimis case outcomes. It turns out that Ringgold is more of an outlier than prevailing wisdom presumes. Further, articulating a quantitative metric and invoking the legal standard of “Labor Value/Access” also may increase the odds of prevailing, all else being equal.

Unfortunately, raising a fair use claim alongside a de minimis defense lowers the chance of a de minimis defense success. This is also surprising. If there is a chance at a de minimis defense, one might think a fair use claim is also strong (given fair uses’ “amount and substantiality factor”). It is not, and we think this is a problem for copyright law and policy. One defense should not preclude the other. We hypothesize that when defendants raise a fair use defense together with a de minimis defense, because fair use is well-established and case law is plentiful (contra the de minimis doctrine), courts frequently default to fair use. This is unfortunate because fair use is more fact-intensive and often occurs at summary judgment, after time-consuming and expensive discovery, whereas de minimis defenses can and should be raised earlier in litigation ideally at the motion to dismiss stage to achieve the doctrine’s goal of judicial economy. We learn from this data that courts and litigants need to be better educated that these defenses can and should be raised in the same case albeit at different times in litigation. Also, as alternative defenses, the existence of fair use should not defeat serious evaluation of the de minimis defense.

In the context of PGS works, however, this logit regression together with Figure 21 below indicates that courts act as if whole copying is more defensible as de minimis, after controlling for other factors. This is also surprising, and a feature of the data set from which courts and litigants can also learn. It is relevant both substantively and procedurally. Substantively, courts appear appropriately solicitous of copying PGS works as forms of “quotation” or reference, which is how we communicate in the internet age. Like fragmentary copying of literary works is quintessentially de minimis, and has been since the origins of the doctrine, copying PGS works for a similar purpose or in a similar manner (as a fragment of a larger whole, or as informational or referential use) should also be de minimis.³²² Indeed, the logit regression in Figure 21 demonstrates that exactly.

The regression represented here is similar to Figure 20, with one major change. Here we look at the additional effect of being both a PGS work and a “whole” copy on the likelihood of having a prevailing de minimis defense (the line highlighted in yellow).³²³ When we look at the interaction of these factors, the majority of the effect (resulting in a successful defense) comes from that combination. These results suggest being both a PGS work and a whole copy is more likely to prevail than use of pictorial works in any other form.

These two Figures (20 and 21), when combined with Figures 15-19, demonstrate a surprising pattern in the dataset: although a de minimis defense rarely prevails, when it does prevail in the Contemporary Era and a whole PGS work is concerned, the original and still prevalent standard of “de minimis non curat lex” explains a defendant’s win, but not fair use which negatively correlates with a prevailing de minimis defense. Access/Labor Value is the standard with the largest influence. This pattern may reflect current communication trends and everyday on-line transactions dominated by visual images (e.g., emojis, memes, gifs) in which the pictorial images stand in for pithy messages and are more informational than expressive.³²⁴ The court’s acceptance of these trends suggests copyright law’s adaptability to new modes of communication, as is appropriate with a statute that was meant to evolve with technological innovation.³²⁵ The negative finding here is that the existence of a fair use defense appears to hurt the chance of a prevailing de minimis defense in the context of whole PGS works, and this puts defendants who copy such works in a difficult position, as they face choosing between an early dismissal on the basis of de minimis and potentially waiving (by failing to raise) the admittedly more fact-intensive but doctrinally more evolved fair use defense. We argue below in Part IV that courts and litigants should reverse this trend and accept alternative defensive pleading without negative consequences to defendant’s chances of success. Photographs alone are treated differently. Although they too are almost always whole works, defendants copying photographs have only won 4 times out of 26 cases (15%). All of these cases were in the Contemporary Era. Possibly courts are reluctant to treat unauthorized copying of photographs as de minimis, although the growing number of de minimis wins for whole PGS works in general suggests a doctrine in flux, and thus ripe for optimizing and clarification.

Finally, let us say something about when in litigation the de minimis defense prevails, which should be early to align with the doctrine’s purpose before significant time and resources are expended. When the de minimis defense prevails, 22% of the time (11/49 cases) it is at the motion to dismiss stage. Sixteen percent of the time (8/49 cases) it is at the preliminary injunction stage. That means that 39% of the prevailing de minimis defenses occur at what we consider the appropriate procedural stage of litigation, leaving 61% of the prevailing de minimis cases at the summary judgment or other later stage. This is not optimal and courts and litigants can do better. Ideally all de minimis wins should be at the motion to dismiss or preliminary injunction stages.

Of the 19 cases in which the de minimis defense prevailed early, 15 include a quantitative evaluative measure and 4 contain a conclusory statement of the rule (“de minimis non curat lex”) or no standard at all. They include a wide range of works (literary, PGS, and audiovisual) and no de minimis form predominates (fragment, some or whole). The amount of successful de minimis dismissals is relatively even between the Early Era and the Contemporary Era (20% and 27%, respectively). To us, this affirms the existence of a fertile ground on which to plant the seed for an updated, useful and generally applicable de minimis defense standard for the internet age. Such a standard will expand the number of cases properly dismissed at early stages of litigation, honor the common law evolution of the equitable doctrine and its purposes of judicial economy, and conform to copyright law and policy. We turn to that formulation in our Conclusion.

Conclusion: A Revised De Minimis Standard and Its Exemplary Application

We learn from this empirical evaluation of the de minimis defense over the three eras of copyright litigation that it has become more relevant but lacks its defining efficaciousness. The de minimis doctrine is meant to quickly dispose of meddlesome lawsuits in which only trifling harm is alleged. But over the eras, the de minimis doctrine has become more complex and has lost some of its decisive efficiency. A return to the origins of the de minimis defense in copyright would reverse the outcomes of some milestone cases, as we explain below, and doing so would require only minor doctrinal adjustments producing significant benefits for litigants and courts.

In the Early Era, few cases were brought and those that were (and prevailed as de minimis) concerned fragmentary copying. They were between competitors, and the predominant legal analysis concerned lack of either market substitution or defendant’s enrichment. Copying occurred but copyright harm was absent. In the Transition Era, more cases arose involving new industries and authors of, for example, PGS works, audiovisual works, and software. Many of these cases were between competitors resembling the Early Era. But also new kinds of claims arose between authors of complementary (not competitive) works in which whole copying occurred, which was nonetheless defended as trivial. In large part, the analysis during the Transition Era remained about substitutional copying that the courts described in terms of a functional test: whether the defendant’s use has “the intent or effect of fulfilling the demand for the original.”³²⁶

In the Contemporary Era, many more cases arise between potential licensees as opposed to industry competitors. Many of the cases consist of individuals suing corporate authors (such as movie or television companies), reflecting a new strategy seeking licensing fees for uses that were previously considered either fair or not substitutional. To be sure, in many of these new kinds of cases, courts dispose of the copyright claims as de minimis with perfunctory analysis, resembling the Early Era. The difference now is that in addition to the cases of fragmentary copying, in many cases defendant copies all of plaintiff’s work (not just a small bit). In other cases from this era, however, courts commence engaging in more complex evaluations resembling fact-laden substantial similarity analyses, requiring more time and resources. New standards arise in these cases distinguishing trivial from harmful copying. In addition to quantitative measures for assessing triviality, in these new licensee-licensor cases concerning whole copying, courts determine substantiality as an approximation of replacement value (the misappropriation standard) through observability and recognizability of the authored expression. This means that while whole copying may technically occur, defendant’s use may be harmless if it is fleeting, elusive, indistinct, or otherwise so insubstantial that the law should not bother.³²⁷

De minimis copying should be resolved at the earliest possible stage and at minimal cost to achieve the doctrine’s goal. If copying is quantitatively small, such as a fragment or insignificant part, courts readily dispose of cases with the standard “de minimis non curat lex.” This is the trend throughout the eras and should continue.³²⁸ If defendant’s copying is whole, but its expressive value to the defendant at first appears debatable absent more searching aesthetic inquiry, the efficiency of the de minimis doctrine shrinks as the complexity of the court’s analysis grows with briefing and discovery demands on both parties. This lamentable trend can be reversed if courts sua sponte raise the de minimis defense at the motion to dismiss stage or litigants raise the defense early while being undeterred from preserving fair use for a later time. In this posture, courts should assess de minimis copying as it was originally intended: in terms of substitutional value for plaintiff’s whole authorial expression in the context of the defendant’s expression.³²⁹

The revised standard would focus on quantitative triviality – how much of or for how long is the plaintiff’s work appreciable to ordinary observers? – and qualitative insignificance – what function does the plaintiff’s work play in the defendant’s work as a whole? Defendant’s whole copy of plaintiff’s work should not undermine the de minimis defense. To the contrary, when the whole copy is akin to quotation, informational reference, or is otherwise substantively insignificant, transitory, or minimized in the context of the defendant’s work as a whole, the defendant’s de minimis defense should prevail. Assessment would occur before discovery and as soon after an answer/dispositive motion to dismiss is filed.³³⁰

The application of this standard is straightforward in many of our exemplary cases discussed in Part II, such as: Webb v. Powers, Matthews Conveyor Co. v. Palmer-Bee Co., Amsinck v. Columbia Pictures, VMG Salsoul v. Ciccone, and Gottlieb v. Paramount Pictures.³³¹ However, only in Gottlieb was the case dismissed early in litigation, leaving the other cases to proceed through expensive and time-consuming discovery. This diversion from the de minimis doctrine’s purpose and proper structure we think is because of the confounding negative effects of Ringgold, the changing business practices among certain industries for in-licensing when copyright law may not require it, and the evolution of the substantial similarity doctrine and fair use after 1978. In light of this empirical study, however, it is our hope that these confounding effects can be mitigated or reversed, and courts and litigants can properly assess and apply the de minimis doctrine early, capably, and with alacrity.

What would this look like as applied to cases we contend are outliers and were wrongly decided? Both Bell v. Wilmot³³² and Ringgold v. BET³³³ would come out the other way and in favor of defendant’s de minimis defense. In Bell, defendant’s whole copying of plaintiff’s photograph would be de minimis because the copy was elusive and largely inaccessible to the public and because it has no expressive value to the defendant (who was buying only website addresses and not commercializing or making them publicly available). Although a whole copy of Plaintiff’s photograph was made, its near invisibility to ordinary observers is akin to a quantitative triviality. It was the equivalent of a copy stored in a file cabinet un-disseminated, relatively prevalent across the eras, or a fleeting display or performance, such as “Happy Birthday” sung by patrons and waitstaff in a restaurant.³³⁴ In Bell, the copy is valueless to the defendant – a near accident of failed due diligence upon the purchase of thousands of such websites – and thus is qualitatively insignificant to the defendant’s work as a whole. The result in Bell is wrong because copying alone is not a harm in copyright law. Actionable copying must cause harm related to the unauthorized use of the expression, such as foregone fees associated with the appreciation or apprehension of the copyrighted work. The plaintiff in Bell suffered no harm derived from defendant’s copying of his photograph except that it resided on an obscure website without his permission. Judicial resources should be minimized in such a case with a swift motion to dismiss and sanctions on appeal if affirmed. This should have especially been the case in Wilmott given the Plaintiff’s reputation as a copyright troll.³³⁵

Likewise, in Ringgold v. BET, the defendant’s use of plaintiff’s quilt poster as background art for 27 non-consecutive seconds during a half-hour sitcom should have been de minimis.³³⁶ Its use despite being a whole copy and recognizable was quantitatively trivial as fleeting and elusive for only 27 of 1,800 seconds of the show. Defendant used the quilt poster in the background of a television set to tell a convincing but fictional story about a community that would have had such a quilt poster hanging on their walls. It was an evanescent but culturally resonant use, an informational referent for a community of observers who would understand its background use as rendering more convincing the fictional story portrayed on television – like seeing the Statue of Liberty in a panoramic shot of New York City, the Charging Bull in an establishing shot for Wall Street, or Cloud Gate (“the Bean”) in the background of a scene set in Chicago’s Millenium Park.³³⁷ The television show did not compete with or serve as a substitute for the quilt or its graphic derivative in poster form. Its use was qualitatively insignificant for the television show’s expressive message as a whole except as informational and referential, and it did not frustrate plaintiff’s marketable expectation for her work. Indeed, the result in Ringgold contrives a market substitute for an attenuated new use through its novel analysis of the de minimis defense, which more closely resembled a substantial similarity evaluation that assumes market substitution as the harm to be avoided.

Reversing the results in both cases in the manner described strengthens the surprising but welcome trend in the Contemporary Era cases adapting the de minimis defense for the internet age in the context of whole copying. Whole copies of PGS works that are referential background or used in fleeting, elusive, or qualitatively insignificant ways are and should continue to be de minimis.³³⁸ These uses are the equivalent to quotations or informational referents, and courts should dismiss them early in litigation as trivial copying consistent with the fragmentary uses from the Early Era. Detailed or aesthetically complex analysis is unwarranted in such cases, confuses de minimis with substantial similarity and undermines sound copyright policy aimed at promoting knowledge and further expression through more authored works. By identifying copyright harm as competitive copying akin to market substitution for the copy, courts can efficiently dismiss trivial or de minimis copying as incommensurate with infringement.

Trivial copying arose in the mid-19^th century as a total defense to infringement. It should remain that way. Not all copying – be it partial or whole – creates copyright liability. Copyright is a misappropriation tort embedded with normative standards of “substantiality” that is a court-created heuristic for materiality and harm.³³⁹ For important reasons of court efficiency and copyright policy, the de minimis doctrine became an independent defense in the mid-19^th century. Our data shows it has laudably remained independent from fair use and the substantial similarity test – both complicated and fact-specific doctrines that spend rather than conserve judicial resources. The trend from the Early Era to the Contemporary Era grows the de minimis doctrine’s relevance but not its uniform success and efficiency, threatening to eviscerate the doctrine’s purpose. Courts and lawyers should remain vigilant and protect the de minimis doctrine as a vital defense to indispensable but trivial copying by dismissing cases early. De minimis copying is a determination that the copier’s use – be it fragmentary, partial, or whole – is immaterial to the author’s reasonable expectation for the exploitation of their authorial expression. This can be straightforwardly evaluated very early in litigation with objective metrics and does not require sophisticated aesthetic analyses, as illustrated above. As this Article shows, the pattern across the eras reflects these features of the historic and critical defense and are worth preserving to revitalize the de minimis defense for the digital age.

IV. Appendix

A. Table 1: Works

Some of the categories of copyrighted works follow the Copyright Act’s own definitions, but many of our codes distinguish within categories of copyrighted works.³⁴⁰ We include pictorial and graphic works as one category (PG), but separate photographs (PHOTO) because we hypothesize their different treatment due to their abundance in the dataset. Sculptural works have their own category, in part to separately capture useful articles and be able to discern any appreciable differences in their treatment given the Copyright Act’s special rules for their reproduction and display.³⁴¹ Among literary works, we separately code for books (BOOK) which includes correspondence but not manuals (such as technical manuals or teaching manuals). Manuals (MANUAL) are coded separately because of their frequency in the first half of the twentieth century among copyright litigants. Theatrical and film screen plays (PLAY) and computer programs (CODE) each have their own category. And we code as articles (ARTICLE) shorter writing that is not a book and includes websites and poetry. Movies (MOVIE), television shows (TV), videos (VIDEO) are also separately categorized. Videos include videos made for distribution on the internet and also potentially on television, usually for journalism purposes. We have a separate category for literary works that are list-like (LIST), which includes directories, test materials, and catalogues. Music (MUSIC) is its own category and includes both musical works and sound recordings. Videogames (VIDEOGAME) is also a separate category. None of these codes are exclusive of the other and can co-occur if the court discusses them.

B. Table 2: Form of De Minimis

We noted when the whole work was copied (even for a brief moment) as WHOLE. When defendant was alleged to copy a very small portion of plaintiff’s work, we coded that as a fragment (FRAGMENT). This included a screen shot of a video, a sentence of a book, a few lines of code in a large computer program. When the copied material was neither a fragment nor the whole work, we coded that as SOME. These are obviously coding choices subject to some interpretation, but that is how all coding works. Other coding categories (see Table 6) dig deeper into the proportion of the fragment to the whole.

C. Table 3: Specific Forms

When defendant was alleged to copy a specific part of plaintiff’s work, such as a famous first line of a book or song, we noted separately using different codes for specific forms. We break these down in the SPECIFIC FORM category. (See Table 3.) We hypothesized courts might find specific or special forms more or less relevant for the de minimis analysis. And some special parts were already deemed as such by leading cases. They include literary or visual characters (CHARACTER), which includes when something is considered the “heart” of the original work, a term that is used by courts with sufficient frequency to garner its own category.³⁴² Other specific forms worthy of separate analysis include pictorial and graphic forms, including screen shots and photographs (IMAGE/PHOTO). We also coded for copying of video, tv, and film (VIDEO), pages (PAGE), sentences (SENTENCE), and phrases (PHRASE). VIDEO is for moving images of all kinds. PAGE is for the select but whole copying of pages. PHRASE can be musical phrases, list headings, aphorisms, and visual “phrases” that resemble ideas, stock images, or themes, but not usually a full sentence or image. SENTENCE includes full sentences and usually more than one (and lines of code), and includes lede sentences. Finally, we include the specific form of ENTIRE when defendant copied the entire work, which should be partially coincident with WHOLE in Table 2, except here images and characters are considered separately. The goal with this subcategory of specific forms is to dig deeper into the kinds of fragments and special parts and forms that make up the quantitative assessments. None of these codes are exclusive of the other and can co-occur if the court discusses them.

D. Table 4: De Mininis Legal Standard

We coded all fragmented literal similarity analyses as quantitative (T) in the context of a de minimis case. By contrast, we considered an invocation of the “ordinary observer” test in the context of what appeared to be a de minimis analysis as a qualitative analysis and coded that as SS. Discerning these tests was important, we thought, as we were interested to learn if the rise of the fragmented literal similarity test (or the ambiguities inherent in the substantial similarity test) correlated with other patterns in de minimis cases.³⁴³ When the court’s assessment was primarily qualitative without an articulated standard or quantitative measure, we coded the case as L. When the de minimis standard was more general without any quantitative or qualitative metrics, or any other metric except a recitation of the principle that “de minimis non curat lex,” we coded that as DM. Courts also sometimes apply a standard that resembles the market harm analysis from fair use when discussing the de minimis defense.³⁴⁴ This occurred in early cases, but sometime also arose in later cases referring to early ones. We coded that as LABOR VALUE, because it was frequently discussed in terms of the investment in hard work and the return on that investment as an original goal of copyright law. Relatedly, courts sometimes consider inaccessibility of the defendant’s unauthorized copy as part of its de minimis defense, which we understood to be a determination that plaintiff was not being harmed by the unauthorized copy. We also included this in the LABOR VALUE category. Both of these standards – DM and LABOR VALUE – are separate from a court’s qualitative analysis, which more resembles the “substantial similarity” test varieties (of L and SS). Since at least Ringgold, “observability” or “recognizability” has become a factor in the legal standard of infringement, in some cases negating a de minimis defense. We included those cases in SS. This includes when courts describe the taking as “meager and fragmentary” such that an audience would not “recognize” the plaintiff’s work. This code often co-occurred with one of the other categories, but its mention in the cases deserved independent analysis because of the doctrinal shift in the de minimis doctrine as described in Part II, which we are attempting to study with some empirical precision. When no standard was announced, not even recitation of the statement “de minimis non curat lex,” we coded that as NS. None of these codes are exclusive of the other and can co-occur if the court discusses them.

E. Table 5: Quantitative Metrics

We further disaggregated the quantitative and qualitative categories into the following subcodes. When the court specifically considers quantitative measures (such as 27 seconds in a television show, or 3 pages of a book) or otherwise described the amount copied in percentage terms, we coded that in terms of the percentage and as AMOUNT. When the court described the taking as “trivial,” “small,” “minimal” or “insignificant,” but did not otherwise use numbers in its assessment, we coded that quantitative metric as FEW/SOME. This category overlaps often with the “form of de minimis” code “FRAGMENT” but not always, because some fragments repeat, become quantitatively and qualitatively significant, and did not merit (in the court’s view) the adjective “trivial” or “insignificant.” We thus also coded for REPETITION, which would co-occur with these other codes, to indicate the court considered the repetition of fractional taking in its analysis, such as when a musical phrase is repeated in a new work. Whole copies were coded as WHOLE. WHOLE co-occurred with a fractional taking (AMOUNT) when the whole work was copied but was displayed or performed for a short time, such as in the background of a film, and the court discussed the time increments. If no amount was discussed, it was coded NS. None of these codes are exclusive of the other and can co-occur if the court discusses them.

F. Table 6: Qualitative Metrics

The qualitative metrics include some of the previously designated codes, including PROTECTABILITY (which includes discussion of accessibility) and RECOGNIZABILITY. We also coded for analyses of the relevance of the part taken (RELEVANCE) when the court considered the expressive significance of the part copied.³⁴⁵ When the court assessed defendant’s alterations to the work copied, we coded that as ALTERATIONS. We coded assessment of the plaintiff’s work pervading the defendant’s copy within RECOGNIZABILITY. This includes discussion of the copying of aesthetic and semantic themes, and it is the qualitative analogue to the quantitative REPETITION. When the court discusses whether a defendant’s work focuses on the part taken from the plaintiff in terms of aesthetic form (e.g., discussing how the camera centers an image in the background, or when the image is blurry), we included that within RECOGNIZABILITY as well. NS is used when no qualitative metric is discussed. None of these codes are exclusive of the other and can co-occur if the court discusses them.

G. Table 7: Industries

SOFTWARE indicates a company that sells computer programs and software products. MANUFACTURER indicates a seller of durable goods, including devices embedded with software such as medical devices. MEDIA indicates all traditional media (news, film, video, print publishers, website publishers) and includes advertising. This category aims for readership and audiences that consumer their content. It does not include music publishers, however, which is its own category (MUSIC PUBLISHER). Musical artists when suing in their individual capacity are coded as INDIVIDUAL, as are all other parties suing in their own capacity. HARDWARE indicates a seller of computer hardware and network systems. EDUCATION/CONSULTING includes educational institutions and those that consult for educational institutions (including authored materials for that purpose). DATA SERVICES indicates companies that sell or provide data, such as phone book or directory companies. But catalogues that are peripheral to a company’s main revenue – such as a manufacturing catalogue – is included as MANUFACTURER. FINANCIAL indicates a company that provides financial services, including financial information services. ARCHITECTURE indicates architectural firms and contractors. These categories can co-occur, such as SOFTWARE and HARDWARE, and multiple party plaintiffs will also have multiple listings of identities.