abstract. In its recent decision in Andy Warhol Foundation for the Visual Arts, Inc. v. Goldsmith, the Supreme Court held that a defendant’s use of a plaintiff’s copyrighted work would be judged “transformative”—and therefore more likely to qualify as a fair use—not simply on the basis that the defendant had altered the plaintiff’s work, but in instances where the defendant had done so in a way that gave the defendant’s new work “a further purpose or different character.” Crucially, in assessing whether changes to the work gave it the requisite different purpose or character, the Warhol Court noted that works that have the same purpose are more likely to serve as substitutes—that is, to compete. On the other hand, works with different purposes are less likely to compete. For that reason, a defendant’s work that is based on a plaintiff’s but is sufficiently different in purpose that it does not compete with it is more likely to be found transformative, and is therefore more likely, all else equal, to be a fair use.

The key question after Warhol, then, is determining whether a defendant’s work is likely to compete with a plaintiff’s work for a particular use. Unfortunately, the answer to that question often will be far from self-evident, and judges will find nothing in copyright law itself to guide them. Instead, judges are likely to fall back on unguided, unreliable intuition. Fortunately, antitrust law offers tools we can employ to better understand whether two artistic or literary works are likely to compete for a particular use. This Essay argues that antitrust market definition and substitutability methodologies lend themselves surprisingly well to the post-Warhol copyright fair-use analysis. Moreover, this Essay argues that if the federal courts turn to antitrust law for help in implementing the Warhol Court’s focus on the prospect of substitution, the fair-use inquiry may, on the whole and over time, become modestly more friendly to defendants than it was before Warhol. This is despite the fact that the Warhol Court ruled against fair use in that particular case.

Introduction

In Andy Warhol Foundation for the Visual Arts, Inc. v. Goldsmith,1 the Supreme Court considered whether the licensing of one of Andy Warhol’s Prince Series works (see the work on the left in Figure 1), for the purpose of illustrating a magazine story about the musician Prince, was a fair use of the Lynn Goldsmith photograph on which the Warhol work was based (see the photograph on the right in Figure 1).

FIGURE 1. warhol’s orange prince (left); goldsmith’s photographic portrait (right)

The Court’s review was limited to consideration of the first of four statutory fair-use factors—that is, “the purpose and character of the use, including whether such use is of a commercial nature or is for nonprofit educational purposes.”2 A central element of that inquiry is determining whether a defendant’s use of a plaintiff’s copyrighted work is “transformative.”3 At the most general level, a defendant’s use of a plaintiff’s work is transformative when the defendant has not merely copied the plaintiff’s work, but rather has altered that work to an extent that the defendant’s activity, and not just the plaintiff’s, advances copyright law’s mission of encouraging the production of new works.4

The transformativeness determination has, over the past three decades, become one of the most important drivers of the fair-use analysis, if not the most important. Indeed, Barton Beebe’s recent empirical analysis of the factors driving results in fair-use litigation reveals that in cases where a court finds that the defendant’s use is transformative, “the ratio of the odds a defendant will prevail in its fair use defense to the odds it will fail is anywhere from 86 to 91 times greater.”5 “By this measure,” Beebe concludes, “a finding of transformativeness exerts by far the greatest impact of any finding on a court’s likelihood of making an overall determination of fair use.”6

* * *

The transformativeness inquiry first took hold as a central element of fair-use analysis in the 1994 decision Campbell v. Acuff-Rose Music, Inc.7 In that case, the Supreme Court determined that the defendant’s use of portions of the plaintiff’s copyrighted song in a rap-music parody of that song was a fair use.8 The defendant’s work was transformative, the Campbell Court held, because it “adds something new, with a further purpose or different character, altering the first with new expression, meaning, or message.”9 Transformative works like the defendant’s rap parody, the Court noted, should be treated more solicitously by copyright law: “[T]he goal of copyright, to promote science and the arts, is generally furthered by the creation of transformative works.”10 As a consequence, “[s]uch works thus lie at the heart of the fair use doctrine’s guarantee of breathing space within the confines of copyright.”11

Following Campbell but prior to Warhol, transformativeness was a protean concept, its definition uncertain and changing to fit the facts of the various disputes in which the concept was deployed. In Campbell itself and in many cases that followed it, transformativeness was assessed by examining the defendant’s work, and by referring to the ways in which it copied but also altered the content of the plaintiff’s work. For example, in Blanch v. Koons,12 the Second Circuit held that postmodern artist Jeff Koons’s use of the plaintiff’s fashion photograph in his Niagara collage (see the plaintiff’s work below on the left in Figure 2 and Koons’s Niagara below on the right in Figure 2) was a transformative fair use because Koons used the photograph as “raw material” for the creation of a work with new meaning and message.13

FIGURE 2. andrea blanch fashion photograph (left); jeff koons, niagara (right)

But sometimes, in cases where the defendant’s work copied but did not alter the content of the plaintiff’s work at all, transformativeness was assessed by considering the purpose to which the defendant had put the plaintiff’s work. For example, in Google LLC v. Oracle America, Inc.,14 the Supreme Court held that Google’s exact copying of portions of the Java Application Programming Interface15 and the reuse of that code in Google’s Android operating system for mobile phones was transformative because it was done for the purpose of creating a new product.16 Google’s copying of the Java code, the Google Court found, “seeks to expand the use and usefulness of Android-based smartphones,”17 and Oracle was not, nor was it likely to be, a competitor in the smartphone market. Similarly, in Authors Guild v. Google, Inc.,18 the Second Circuit held that Google’s wholesale copying of millions of books was transformative, and ultimately fair use, because the purpose of the copying—creating the Google Books online research tool—was not to reproduce or distribute the plaintiffs’ books as expression, but rather “to make available significant information about those books, permitting a searcher to identify those that contain a word or term of interest, as well as those that do not include reference to it.”19 For that reason, as the Google Court later framed it, Google’s copying “was consistent with that creative ‘progress’ that is the basic constitutional objective of copyright itself.”20

In Warhol, the Court assessed a claim for transformativeness of the first type: the Warhol Foundation asserted that Andy Warhol’s Prince Series portrait added new expression, meaning, or message to Goldsmith’s original photograph, and thus should be treated as transformative.21 The Court was no doubt wary of an approach to transformativeness that would grant fair-use immunity whenever a defendant even modestly alters the content of a copyrighted work.22 It accordingly attempted to anchor the transformativeness test to a key part of the rationale it had originally laid out in Campbell for giving transformative uses more solicitous treatment—that a transformative use was properly understood as not merely altering a copyrighted work, but doing so in a way that gave the defendant’s new work “a further purpose or different character.”23 Crucially, in assessing whether changes to the work gave it that required different purpose or character, the Warhol Court noted that works that have the same purpose are more likely to compete—that is, as the Court framed it, to serve as substitutes.24 On the other hand, works with different purposes are less likely to compete.25 For that reason, a defendant’s work that is based on a plaintiff’s but is sufficiently different in purpose that it does not compete with it is more likely to be found transformative, and is therefore more likely, all else equal, to be a fair use.26

Applying this test to the facts of the case, the Warhol Court held that licensing the Warhol portrait to illustrate magazine stories about Prince was not a fair use.27 The works were licensed for the same use; they were competing for that use and therefore shared the same purpose.28 As a consequence, the ways in which the content of the Warhol portrait differed from the Goldsmith photograph did not outweigh the commerciality of the use.29

* * *

The key question after Warhol, then, is determining whether a defendant’s work is likely to substitute for—that is, compete with—a plaintiff’s work for a particular use. Unfortunately, the answer to that question often will be far from self-evident, and judges will find nothing in copyright law itself that tells them how to answer. Instead, judges are likely to fall back on intuition—as, we shall see, the Warhol majority itself did. We shall also see that our unguided intuitions about competition are not always reliable.

Fortunately, antitrust law offers tools we can employ to better understand whether two artistic or literary works are likely to compete. This Essay argues that antitrust market definition and substitutability methodologies lend themselves surprisingly well to the post-Warhol copyright fair-use analysis. Moreover, this Essay shows that if the federal courts turn to antitrust law for help in implementing the Warhol Court’s focus on the prospect of substitution, the fair-use inquiry may, on the whole and over time, become modestly more friendly to defendants than it was before Warhol. This is despite the fact that the Warhol Court ruled against fair use in that particular case.

To understand the Warhol Court’s revised approach to the transformativeness analysis and its likely consequences for the future of fair use, we must step back and look more closely at the facts of that case and the details of the majority opinion. Part I of this Essay does that, with particular attention to the Court’s focus on the prospect of competition as a tool for assessing transformativeness. Part II describes how antitrust lawyers assess whether particular goods or services are competitors. Part III argues that courts could, and indeed should, employ antitrust tools to address the particular competition questions the Warhol Court has identified as essential to the fair-use analysis. In particular, Part III considers how to adapt methods developed to advance antitrust law’s goal of protecting competition’s price-disciplining power to the very different context of copyright law, which seeks to create conditions that allow copyright owners to charge supracompetitive prices (i.e., prices above what a competitive market would otherwise dictate). The Essay then concludes.

I. the warhol court’s introduction of competition into the transformativeness analysis

In 1981, popular-music photographer Lynn Goldsmith, on assignment from Newsweek, took a photo of Prince (shown above in Figure 1).30 Prince was then an up-and-coming musician who was just about to release the album, 1999, that would put him on the path to becoming a pop icon. Later that same year, Newsweek used the photo to illustrate a two-page spread on the musician.31

In 1984, Vanity Fair planned a lengthy profile of Prince, to be written by noted cultural critic Leon Wieseltier, then a columnist for the magazine writing under the pen name “Tristan Vox.”32 Vanity Fair commissioned an illustration for the article from Andy Warhol, well-established by then as a pop-art superstar, and the magazine negotiated with Goldsmith’s agency for a license to Goldsmith’s 1981 photograph, to serve as an “artist reference” for the illustration.33 The terms of Vanity Fair’s license from Goldsmith stated that the illustration was “to be published in Vanity Fair November 1984 issue. It can appear one time full page and one time under one quarter page. No other usage right granted.”34 Goldsmith received four hundred dollars and a source credit in the article.35 The magazine didn’t disclose to Goldsmith that it had hired Andy Warhol to do the illustration.36

Andy Warhol used the photograph, cropped down to Prince’s face, as the raw material for sixteen Prince Series works. One of those works, “Purple Prince,” ran as an illustration to the Vanity Fair article, titled Purple Fame (see Figure 3).37

FIGURE 3. warhol’s purple prince as it appeared in vanity fair

What use did Warhol make of Goldsmith’s photograph in creating his Prince Series works? We learn something on that point in Justice Kagan’s dissent, which details the artist’s process:

Warhol cropped the photo, so that Prince’s head fills the whole frame: It thus becomes disembodied, as if magically suspended in space . . . . Warhol converted the cropped photo into a higher-contrast image, incorporated into a silkscreen. That image isolated and exaggerated the darkest details of Prince’s head; it also reduced his natural, angled position, presenting him in a more face-forward way. Warhol traced, painted, and inked, as earlier described.38 He also made a second silkscreen, based on his tracings; the ink he passed through that screen left differently colored, out-of-kilter lines around Prince’s face and hair . . . . Altogether, Warhol made 14 prints and two drawings—the Prince series—in a range of unnatural, lurid hues.39

* * *

On April 21, 2016, Prince passed away at the age of fifty-seven.40 Later that month, Condé Nast, Vanity Fair’s corporate parent, reprinted a different work in Warhol’s Prince Series, Orange Prince (see the work on the left in Figure 4), as the cover for a special issue commemorating Prince’s life (see the cover on the right in Figure 4). Condé Nast paid the Warhol Foundation ten thousand dollars for a license to the work.41 Condé Nast did not seek a license from Goldsmith.42

FIGURE 4. warhol’s orange prince (left); orange prince as it appeared on the cover of the condé nast commemorative issue (right)

This second use was not covered by the original license, which limited Vanity Fair to a single use (i.e., to illustrate one magazine story). When Goldsmith learned of the 2016 Condé Nast special issue, she made her displeasure known.43 In response, the Warhol Foundation filed a complaint seeking a declaratory judgment that Orange Prince did not infringe the Goldsmith photograph and was fair use.44

The district court sided with the Foundation, holding that Orange Prince transformed the original Goldsmith work to give it a new meaning and was therefore a fair use.45 In the district court’s view, the Goldsmith photograph portrays Prince as “not a comfortable person” and a “vulnerable human being,” while the Warhol work portrays Prince as an “iconic, larger-than-life figure.”46 The defendant’s use, according to the court, created a transformative work with a new meaning or message, and was therefore fair.47

On appeal, the Second Circuit reversed.48 The court of appeals rejected the notion that “any secondary work that adds a new aesthetic or new expression to its source material is necessarily transformative.”49 The question, instead, was “whether the secondary work’s use of its source material is in service of a fundamentally different and new artistic purpose and character.”50 Such “transformative purpose and character must, at bare minimum, comprise something more than the imposition of another artist’s style on the primary work.”51 Here, however, “the overarching purpose and function of the two works at issue . . . is identical, not merely in the broad sense that they are created as works of visual art, but also in the narrow but essential sense that they are portraits of the same person.”52

The Supreme Court granted certiorari and affirmed, albeit on narrower grounds and employing different reasoning than the Second Circuit.53 Importantly, the Court made clear that its analysis was limited to whether the Warhol Foundation’s use was transformative under the first statutory fair-use factor, which inquires into “the purpose and character of the use, including whether such use is of a commercial nature or is for nonprofit educational purposes.”54 In affirming the Second Circuit’s determination that the defendant had not made a transformative use that would tip the first factor in its favor, the Court made two key holdings.

First, the Court held that “[t]he fair-use provision, and the first statutory factor in particular, requires an analysis of the specific use of a copyrighted work that is alleged to be an infringement.”55 “The same copying may be fair,” the Court said, “when used for one purpose but not another.”56 Accordingly, the Court said, it would focus on the challenged use—the license of Orange Prince to Vanity Fair to illustrate a cover story about Prince’s life—and would not rule on whether Orange Prince itself, any of the other Warhol Prince Series works, or any other conceivable use of Orange Prince, was or was not fair use.57

Second, the Court made clear that a defendant’s use of a plaintiff’s work is not necessarily transformative just because that use contains a new meaning or message. Something more is required—the defendant’s use must have a different purpose than the plaintiff’s.58

That different purpose, the Court held, was lacking. Of particular significance to the majority, both Goldsmith’s photograph and Warhol’s Orange Prince artwork had been used to illustrate magazine stories about Prince. The particular challenged use—the Warhol Foundation’s licensing of Orange Prince to Vanity Fair—was the same as a principal use of the Goldsmith photograph. The use was therefore a potential substitute for the licensing of Goldsmith’s work.59 That fact was critical to the first-factor analysis. The Court said:

The first fair use factor . . . considers the reasons for, and nature of, the copier’s use of an original work. The central question it asks is whether the new work merely supersede[s] the objects of the original creation . . . (supplanting the original), or instead adds something new, with a further purpose or different character. In that way, the first factor relates to the problem of substitution—copyright’s bête noire. The use of an original work to achieve a purpose that is the same as, or highly similar to, that of the original work is more likely to substitute for, or supplan[t], the work.60

This statement about the role of substitution, identifying it as “copyright’s bête noire,”61 gives us an unusually revealing window into the Court’s understanding of the purpose of copyright law. Copyright’s bête noire is not copying as such. It is copying that leads to the likelihood of substitution. Copying that leads to the creation of a new work with a different purpose—one not likely to substitute for the plaintiff’s original work—is, if anything, the outcome we seek (the creation of new works), not the evil we seek to avoid (copying substituting for a copyrighted original).

The Warhol Court’s introduction of competition into the transformativeness analysis, along with its sharpening of the concept by linking it to particular uses, elevates the analysis substantially. After Warhol, the prospect of competition between the parties’ works for a particular use is now the key to assessing transformativeness. And after Warhol, as before it, the defendant’s success or failure in establishing transformativeness will in many cases be determinative of fair use.

II. antitrust’s approach to assessing competition

The question now is how courts will implement Warhol’s focus on competition. Competition is not a wholly foreign concept to copyright law—it has long been part of the analysis under the fourth statutory fair-use factor. But it is fair to say that competition is not a concept central to copyright law, because liability does not require a plaintiff to show that defendant’s infringement has a competitive effect.62 As a consequence, copyright law does not have internal tools to judge the likelihood of competition in particular cases. In contrast, antitrust law has developed and honed, over many decades of use and refinement, tools to assess competition and competitive effects.

It may be that courts follow the lead of the Warhol Court and fall back on intuition. But as the Warhol opinion itself shows, that would be a dangerous route to follow. The Warhol Court considered the question of transformativeness in the context of a particularly narrow use, licensing to magazines to illustrate stories about Prince, where the prospect of substitutability seemed obvious. Indeed, as the Court noted, both Goldsmith’s photograph of Prince and Warhol’s portrait of Prince had been licensed in the past to illustrate magazine stories about the musician.63 To the Court, that seemed about as plain a case of potential substitutability as can be imagined.64 But the truth is far more complex.

Consider that every minute of every day, in millions upon millions of transactions, both Coca-Cola and milk are sold, often in the same stores, to consumers who are thirsty.65 Does that mean that Coke and milk have the same “purpose”? Antitrust lawyers would not, in general, assess that question directly. They would approach the inquiry as the Warhol Court did (at least implicitly), by asking whether the products were likely substitutes—that is, whether they had substantial cross-elasticity of demand.66 With respect to Coke and milk, most antitrust lawyers would blanche at the idea that just because those two drinks are sold to the same kind of consumer (thirsty ones), and in the same stores, that we should conclude they are “substitutes,” or that they compete in the same market, or that, from a consumer’s perspective, they have the same “purpose.” Of course, it is possible that under some extreme conditions, even consumers who strongly prefer Coke might consider switching to milk. Imagine the price of Coke triples but the price of milk stays the same. We might see some substitution then from Coke to milk (and other drinks). Does that mean that Coke and milk are interchangeable? Or that they share the same “purpose”? Most consumers would not say so. That consumers might switch in response to a three-hundred-percent increase in price does not provide a meaningful test for assessing the prospect of substitutability. What we’re looking for is substitutability under reasonable market conditions.

Note, again, that the Warhol Court was explicit in tying the determination of a work’s “purpose” to the prospect of substitution: remember the Court’s statement that “use of an original work to achieve a purpose that is the same as, or highly similar to, that of the original work” is more likely to lead to substitution—“copyright’s bête noire.”67 But without knowing more about how magazine editors would view the Goldsmith and Warhol works, we can’t rule out the possibility that the two photographs are like Coke and milk—that is, that some magazine editors would strongly prefer the Warhol portrait, and others the Goldsmith photograph, and that competition between the two works for any particular licensing opportunity is unlikely.

* * *

Antitrust assesses the substitutability of particular products or services principally as part of the task of defining relevant antitrust markets. The goal of market definition is to facilitate analysis of whether the business practice or transaction at issue in a particular antitrust dispute will affect competition. To do that, antitrust lawyers often find it helpful to form a working understanding of which products and services will count as “in competition” with one another for the purposes of legal analysis. Market definition can be thought of as a way of drawing a line between those that are “in” and those that are “out” of the sphere of competition for the purposes of antitrust analysis. A market definition is thus a simplification, for analytical purposes, of what is usually a much messier economic reality.68

The crucial insight underpinning market definition is that one product or service will tend to exercise a competitive constraint on another to the extent that purchasers of the first product or service will regard it as a substitute for the second. If Product B is a substitute for Product A, then purchasers can turn (or threaten to turn) to Product B in response to a price increase in Product A. The prospect of substitution will tend to limit the ability of Product A’s supplier to increase prices profitably.

A. The Hypothetical-Monopolist Test

Antitrust recognizes two principal ways to define a market. One is the “hypothetical-monopolist test,” and particularly the version of it known as the “SSNIP test” (a “SSNIP” is a “small but significant non-transitory increase in price”).69 This test is essentially a thought experiment. It involves starting with the product or service that we are focused on and adding substitutes one at a time, starting with the closest substitute, until a hypothetical monopolist of all included products would find it profitable to increase the price of at least one of those products or services by a significant amount (traditionally, around five percent) for a nontransitory period of time (traditionally, at least a year).70 We do this to figure out how many of these products or services must come under the control of a hypothetical monopolist before the supplier has gained the power to inflict economic harm on others.71

So, for example, imagine we want to understand the relevant antitrust market for Starbucks’s packaged coffee (see Figure 5).72 We would start by positioning Starbucks as a hypothetical monopolist for packaged coffee. We would then ask whether, in response to a five-percent price increase, enough packaged-coffee consumers might switch to packaged coffee from the product we expect would be the closest substitute—here, Peet’s—such that the SSNIP would be made unprofitable. Assume for the moment that the answer is yes. Peet’s is in the same market as Starbucks—that is, there is sufficient “demand elasticity” between packaged coffee from Starbucks and from Peet’s such that the presence in the market of Peet’s constrains the price that Starbucks can charge for its packaged coffee. If that is true, then Starbucks and Peet’s are both competitors in the market for packaged coffee.

FIGURE 5. hypothetical relevant market analysis for starbucks’s packaged coffee

The next step would be to ask the same question again, positioning a combined Starbucks/Peet’s as the hypothetical monopolist, and inquiring whether a five-percent price increase would cause enough packaged coffee consumers to switch to packaged coffee from Dunkin’ such that the SSNIP would be made unprofitable.

At some point we are going to find that a SSNIP becomes profitable, because demand for the next product is not responsive enough to a price increase in the product(s) of the hypothetical monopolist to defeat that price increase. When I teach this concept in antitrust, almost all of my students believe that the relevant market for Starbucks packaged coffee includes packaged coffee from Peet’s, but they usually are skeptical that either Dunkin’, Folgers instant, or any of the other products shown compete in that market. In other words, my students do not expect that the presence of Dunkin’ or Folgers instant disciplines the price that Starbucks charges for packaged coffee, at least at the five-percent margin that the SSNIP test considers to be meaningful in measuring competition. If that’s right, then we have defined the relevant antitrust market—that is, the products or services that are sufficiently substitutable that they are in competition—as limited to packaged coffee from Starbucks and Peet’s.

Of course, applying this test in actual antitrust disputes requires a good deal of information about the characteristics of the products or services at issue and what real purchasers will do in response to a price increase. But if it is applied correctly, the SSNIP test supplies substantial rigor to the competition analysis: it provides a methodology within which data can be gathered about substitutability, so that the contours of competition between and among products and services can be more precisely understood, rather than guessed at.

There is one wrinkle in applying antitrust tools in the copyright context that should be addressed. Antitrust’s market-definition analysis assumes that the relevant products or services are currently being supplied at the competitive price rather than a supracompetitive “monopoly” price. In antitrust it is error to accept uncritically the price charged by a firm with significant market power as the baseline for assessing the prospect of substitution. This is because the firm with market power (i.e., power over price) may be engaged in supracompetitive pricing, and whereas raising price above the monopoly price may spark substitution, a price rise above a lower, competitive, price, may not. This error has become known as the “Cellophane fallacy,” after the case73 in which the Supreme Court fell into it:

The problem arises because firms exercising market power may boost prices to a point at which further price increases are made unprofitable, constrained by the threat of demand substitution. Under such circumstances, an approach to market definition that infers buyer substitution from a small hypothetical price increase over the currently prevailing price—the usual practice in market definition analysis when the alleged harm to competition is prospective—may make buyer substitution appear to be greater than it was when price was at the noncollusive (but-for) price level, under the plaintiff’s view that the but-for price was lower.74

Of course, the antitrust assumption of competitive pricing as a baseline in market definition does not hold when the products being assessed are copyrighted works, as the entire point of copyright is to loosen competitive discipline and to create some measure of power over price for the copyright owner. For that reason, for the purposes of assessing substitution in copyright markets, the starting-point assumption of competitive pricing doesn’t have to hold for the antitrust substitution methodology to work. In the context of copyrighted works, the SSNIP test is assessing whether other products are substitutes for the copyrighted work if we move pricing some increment above the supracompetitive price that copyright seeks to create. The supracompetitive price is the baseline.75

B. The Brown Shoe Test

Another way to define a market is through the appraisal of qualitative evidence about the relevant products or services, in an effort to identify the relevant similarities or differences that, in light of market practice, may affect whether those products or services are competing in the same market. This less formal approach is often associated with the Supreme Court’s Brown Shoe Co. v. United States decision.76 In that case, the Court classed evidence of price elasticity—the focus of the hypothetical-monopolist test—as just one indicator among the many that may be useful in assessing substitutability:

The outer boundaries of a product market are determined by the reasonable interchangeability of use or the cross-elasticity of demand between the product itself and substitutes for it. However, within this broad market, well-defined submarkets may exist which, in themselves, constitute product markets for antitrust purposes. The boundaries of such a submarket may be determined by examining such practical indicia as industry or public recognition of the submarket as a separate economic entity, the product’s peculiar characteristics and uses, unique production facilities, distinct customers, distinct prices, sensitivity to price changes, and specialized vendors.77

This approach has long been subordinated to the hypothetical-monopolist test but is making something of a comeback78 as enforcers, frustrated in some instances by their inability to obtain the direct pricing evidence they need to derive fully persuasive results from the hypothetical-monopolist test, favor the use of lay testimony and documents from market participants that bear on the prospect of substitution. Thus, for example, in the Federal Trade Commission’s’ successful challenge to the proposed acquisition of one smokeless-tobacco supplier, National, by another, Swedish Match, the district court held that loose-leaf tobacco and moist snuff were not in the same market based on lay evidence about substitution:

The views of Swedish Match and National competitors, statements by loose leaf distributors, and internal documents of Swedish Match and National show that price-based substitution between loose leaf and moist snuff is generally lacking. Swedish Match competitors believe that there is no switching between loose leaf and moist snuff on the basis of price . . . . Swedish Match and National internal business documents confirm that pricing has little effect on loose leaf demand.79

The less formal Brown Shoe method is likely to be most helpful when direct evidence about pricing and price elasticity (i.e., the responsiveness of demand for Product B to changes in the price of Product A) is not available. In the context of assessing substitutability of copyrighted works, this is most likely to be the case where, as in Warhol, the works are not mass-market goods (i.e., not like pop songs, movies, books, or consumer software) and transactions occur comparatively rarely and on bespoke terms.

III. applying antitrust substitutability analysis in the fair-use context

A. Types of Evidence

To see how antitrust methodologies might be applied in the fair-use context, let’s return to Warhol. Applied in the context of that case, the hypothetical-monopolist test would start with Goldsmith’s photograph and ask if a customer—in Warhol, a magazine editor choosing a photograph or other artwork to illustrate a story about Prince—would consider switching to Warhol’s work if the price of a license to Goldsmith’s was increased by five percent. The Brown Shoe test would also inquire into cross-elasticity of demand. It would examine the particular characteristics and uses of the works, compare the customers the works are licensed to and the prices the works command for licenses, and seek other qualitative data relevant to substitution.

We don’t know what either of the tests would suggest—the outcome of either inquiry is not something we can know based purely on theory or intuition. Fortunately, there are a number of well-established empirical strategies that antitrust litigants use to offer evidence about demand cross-elasticity.80

One is to examine customer reaction to historical price changes—that is, when the price of Product A went up or down, whether demand for Product B either decreased or increased, respectively. If price changes in Product A are linked to substantial changes in demand for Product B, then there is significant demand cross-elasticity between the two products, and they are likely substitutes competing in the same market. If, however, demand for Product B is unaffected or only slightly affected by changes in the price of Product B, then the two products are likely not substitutes. Of course, the inquiry also works in reverse—we can look to see whether changes in the price of Product B are associated with changes in demand for Product A.

This type of evidence is most likely to be available where the products at issue are sold or licensed widely—that is, where there is a lot of historical transaction data to analyze—and where there have been price changes in one of the products at issue but not the other. In such cases, parties arguing over substitutability can use econometric methods to provide rigorous analysis of this historical data. In the context of copyright fair-use analysis, that kind of analysis is more likely to be tractable in cases where the works at issue are marketed to consumers—for example, pop songs,81 books, motion pictures, or software widely used by consumers or businesses. Many fair-use cases involve these kinds of copyrighted works. However, in cases like Warhol that involve works that are sold or licensed only episodically—and where the works, the potential customers, and the prices paid and other terms negotiated are likely to be highly idiosyncratic—historical evidence of cross-elasticity is less likely to be helpful. As stated previously, in cases like Warhol that involve singular, seldom-traded works, we are more likely to employ the less formal Brown Shoe analysis.

Another way to investigate cross-elasticity, useful when employing either the hypothetical-monopolist or Brown Shoe inquiries, is to undertake customer interviews, which are sometimes done informally and sometimes as part of a more formal customer survey. In many cases, customer interviews can supplement evidence from historical price changes: customers can be asked whether they substituted away from Product A to Product B in response to a historical price change, and why they did or did not do so. However, customer interviews can be useful even in the absence of historical price data. Customers who have experience purchasing the kinds of products for which evidence regarding substitutability is sought can be asked to consider whether they would substitute one product for another given the kind of five-percent price increase that the hypothetical-monopolist test employs, and to give reasons for their decision.

This kind of evidence is not always reliable; self-reports regarding likely choices in a hypothetical scenario are not the same as real-world decisions. That said, in cases where the customers are experts—like Warhol, in which the customers are magazine art directors or other editorial staff who regularly make decisions about licensing illustrations—even impressions about hypotheticals may be illuminating. Focusing again on the facts of Warhol, we would want to know whether magazine editors who source illustrations consider the differences between the Goldsmith and Warhol works to be immaterial—that is, that any portrait of Prince will do to illustrate a story about Prince—or the opposite. For example, we may imagine editors testifying that they would view Goldsmith’s classical black-and-white photographic portrait of Prince and Warhol’s colorized pop-art Prince Series work as giving off fundamentally different visual impressions, such that they each would be fit to illustrate different kinds of stories about Prince, but would not, under ordinary circumstances, be substitutable for any particular story.

B. Setting the Margin

Perhaps the most important question that must be addressed in applying ’antitrust methodologies in fair-use cases involves the margin we should use to assess competition. Should we apply the five-percent margin that stands for antitrust-relevant substitutability in copyright fair-use cases, or is some other margin appropriate? A dozen years ago, Mark A. Lemley and Mark P. McKenna noted that if applied in the intellectual-property context, the standard antitrust approach to substitutability would produce some surprising results:

Does Pepsi compete with Coke? It seems a straightforward question; perhaps you think it has a straightforward answer. Sure they compete, you might say; they are both colas, they are used for the same purpose, they are sold next to each other in the grocery store, and they cost about the same. In fact, however, the answer is far from clear. Whether you think Coke competes with Pepsi may depend on whether you drink one of them. If you do, ask yourself this question: how big a price increase would it take before you would switch from one to the other? A penny? Five cents? Ten cents? What if the price doubled? Maybe you wouldn’t switch at any price. If you are like a significant number of consumers, it would take a pretty dramatic change in price to get you to give up your preferred brand.

For antitrust, this insensitivity to price has a simple—and shocking—implication: Coke and Pepsi don’t compete in the same market.82

It is far from obvious what to do about this—or, indeed, whether the observation is a problem at all for the use of antitrust substitutability methodologies in the fair-use analysis. The five-percent threshold is itself somewhat arbitrary; it is not a rule but rather an approximation,83 subject to adjustment in individual cases, that is useful for assessing substitution. Applying the SSNIP test as a baseline for determining transformativeness in the fair-use analysis will not destabilize intellectual-property law generally. Nor will it provoke inappropriate use of antitrust law to curtail the market power that intellectual property seeks to foster. Using antitrust methodologies to discipline the competition analysis that Warhol requires is a discrete intervention.

Considering that the supracompetitive prices that copyright fosters can serve as the baseline in the substitutability analysis,84 the SSNIP test as applied in the copyright fair-use context is measuring the prospect of substitutability at a five-percent margin above whatever supracompetitive price the copyright owner is able to charge. In this way, the SSNIP test adapts more or less automatically to the copyright context. Still, it is true that antitrust’s concern with curbing anticompetitive conduct by firms with pricing power is in fact the inverse of copyright’s policy goal, which is to protect pricing power against certain kinds of competition—centrally, competition from unauthorized copies. One might argue that when applying antitrust’s methodologies for assessing substitution in copyright fair-use cases, and, in particular, the hypothetical-monopolist test, the margin used to assess substitution should be more tolerant of pricing power than the five-percent margin that antitrust typically employs.

At the theoretical level, adjusting the margin to fit with copyright’s policy goals would mean that the test should consider the likelihood of substitution by reference to a margin calculated to permit the copyright owner to recover not just the marginal cost of producing any given copy of the work, but the fixed costs of producing the first copy of the work. Because unrestrained competition will tend (at least in theory)85 to drive the price of artistic and literary works down to the marginal cost of producing a copy, it is principally the copyright owner’s ability to recover the fixed costs of creation that is threatened by competition from copies. This element of cost, which we can refer to as “first-copy costs,” includes both the cost of producing the first copy of the work and a return, whether on the capital invested or the opportunity cost sustained, that reflects the risk of the creative enterprise.

First-copy costs vary enormously across different types of copyrighted works. Many copyrighted works—including many photographs, songs, poems, paintings, and even novels—have relatively low fixed costs of creation. Others, such as motion pictures, have relatively large first-copy costs. The margin along which substitution is assessed in particular fair-use cases should be adjusted to track these costs, just as the margin in antitrust cases can be adjusted to fit the realities of particular product markets.86 For a work with higher first-copy costs, the copyright owner must be able to sustain pricing that is a greater increment above the marginal cost of producing the next copy, relative to a work with lower first-copy costs. This would mean that the margin should be larger for works with higher first-copy costs: to protect the copyright owner’s incentives to create, works that are substitutable only at a larger margin should be considered to be competitors. The adjustment need not be precise—like the SSNIP used in antitrust analysis, the margin used to assess substitutability in the copyright fair-use context will produce useful information about the likelihood of substitutability if it is roughly proportionate to the fixed costs of creativity in a particular copyright market, on average.

None of these complexities are insuperable. In many cases, parties will be able to produce evidence about the fixed costs of creation sufficient to adjust properly the margin for assessing the prospect of substitution. But in a practical sense, the added complexity of a rigorous analysis of substitution is not a serious argument against the adaptation of antitrust tools to assess the prospect of substitution in the fair-use analysis. The Supreme Court has identified fair use as an affirmative defense.87 That means that defendants bear the burden of establishing fair use—and, in cases where transformativeness is at issue, would consequently bear the burden of establishing that a lack of substitutability points toward transformativeness. If defendants prefer to invite courts to rely on their intuition rather than offering evidence regarding substitutability, that is their choice. But in cases where intuition does not clearly point in the direction of transformativeness, defendants can either shoulder the burden of offering empirically based arguments or trust their luck and allow a court or a jury to decide the issue using its unguided intuition. The result in Warhol illustrates the risks of the latter option.

C. Noninfringing Works and Substitution

Sometimes, courts will need to assess substitutability in cases where there are noninfringing works that may be substitutable for the plaintiff’s work. In such cases, the defendant’s allegedly infringing work will not be the only source of potential competition.

The facts of Warhol illustrate this issue. It has sometimes been said that Prince disliked being photographed,88 or that, at best, he “had a love/hate relationship with the camera.”89 And yet, Prince posed for a large number of photographic portraits. Goldsmith’s copyright in her photograph of Prince gives her no exclusive right to Prince’s portrait in general—other photographers remained free to photograph Prince, and painters to paint him.90 And some of these noninfringing portraits of Prince will be available for licensing to illustrate magazine stories about Prince, potentially in competition with the Goldsmith work.

This problem generalizes: there will be a number of fair-use cases in which noninfringing works may potentially substitute for the plaintiff’s work. How should the substitutability analysis take account of potential noninfringing competitors when assessing whether the defendant’s work is a likely substitute?

Antitrust’s unilateral-effects analysis offers a way to think about this situation. In the context of antitrust analysis of corporate mergers, “unilateral” anticompetitive effects arise when, as a result of the loss of head-to-head competition between the two parties to a horizontal merger, the merged firm can unilaterally raise its prices, lower quality or output, or otherwise change its behavior in harmful ways as a result of the lost competitive pressure.91 This kind of harm is a central concern when the merging parties are particularly close competitors in a market with differentiated products or services before the transaction.92

Like antitrust’s methodology for investigating the prospect of substitution, unilateral-effects analysis can be adapted to the copyright context, specifically to address the situation in which potential substitutes for a plaintiff’s work include noninfringing works. The question would be whether the defendant’s work is a substantially closer substitute than the relevant noninfringing works for the particular use at issue, such that it is substantially more likely than those noninfringing substitutes to discipline the supracompetitive price that the plaintiff copyright owner can charge for the relevant use of its work. If there are a substantial number of potentially substitutable noninfringing works, and the defendant’s work is not a particularly close competitor to the plaintiff’s work, then there is little reason to believe that the presence in the market of the defendant’s work is likely to make any difference in the price the plaintiff is able to command for licensing its work—there are sufficient noninfringing substitutes already in the market to discipline the plaintiff’s price.

D. “Filtering” to Guard Against False Positives

A defendant’s work that has copied sufficient material protected by copyright to trigger liability may or may not also be an effective substitute for the plaintiff’s work. Crucially, however, even if a defendant’s infringing work is in fact a substitute for a plaintiff’s copyrighted work, that does not mean that the infringing content makes it a substitute. It may be that noninfringing elements of the infringing work are the source of the substitutability.

Once again, the Warhol case illustrates this possibility. Lynn Goldsmith’s copyright does not give her exclusive rights in the appearance of Prince’s face. Although neither the Supreme Court nor either of the lower courts in the Warhol litigation fully specified the protected elements of Goldsmith’s photograph, at most those protected elements would include Goldsmith’s choices about lighting, posing, some of the makeup choices, and perhaps the arrangement of Prince’s hair. Given that much of the photograph’s content are facts (i.e., the outlines of Prince’s face) that are unprotected by copyright, and given that the copyright in the Goldsmith photograph is therefore relatively thin, it is entirely plausible that if Warhol’s portrait is substitutable for Goldsmith’s photograph, it is not because of the appeal of Goldsmith’s discrete creative contributions (not all of which were reproduced in full in the Warhol portrait) but because the subject of both works is Prince.

A similar problem arises when a defendant’s work copies protected content from a plaintiff’s work but also adopts the same artistic style as the plaintiff’s work. Artistic style is, at a general level, not copyrightable.93 Again, it would be possible that, if the defendant’s work is substitutable for the plaintiff’s, it is not because of the protected content reproduced in defendant’s work, but because of the defendant’s lawful adoption of the same general artistic style.

The question is what, if anything, a court should do to distinguish between protected and unprotected elements when applying antitrust substitutability tests in a fair-use case. One might argue that there is no need to distinguish, because the assessment of substitutability is directed toward determining whether the defendant’s work is transformative, and transformativeness is a quality that arises out of elements that the defendant has added to its work. On this account, what we’re looking for in the substitutability analysis are differences, not similarities, whether infringing or not.

In my view, that argument is not persuasive, both because it misunderstands the problem and because an effective fix is readily available. First, with respect to the nature of the problem, transformativeness can arise either from addition or subtraction—in other words, from elements that the defendant adds to its work, or elements that the defendant removes. In the context of a photographic portrait, we can imagine a defendant transforming a work by removing some or all of the original work’s copyrightable elements—for example, by removing creative lighting elements from a photograph.94 It may be that what’s left are the uncopyrightable outlines of a face, and yet the works may still be substitutable if purchasers are interested simply in the person portrayed, rather than in the particular copyrightable elements of the plaintiff’s particular portrayal.

Again, courts may be tempted to fall back on their intuition in deciding whether a defendant’s work may be substitutable because it copies specific copyrighted elements from the plaintiff’s work, or because of similarities that arise from the copying of uncopyrighted elements. Courts may think, based on the same intuition about the prospect of substitutability in general, that they understand the source of substitutability. But as with the prospect of substitutability in general, if courts take the more rigorous approach outlined here, they have opportunities to improve upon their intuition regarding precisely why two works may, or may not, be substitutes.

If examining historical price data, courts can compare cross-elasticity between the plaintiff’s work and the defendant’s with cross-elasticity between the plaintiff’s work and noninfringing potential alternatives. If the comparison suggests that the plaintiff’s work has greater cross-elasticity with the defendant’s than with noninfringing alternatives, that is some evidence—albeit in many cases not conclusive—that the substitutability of the defendant’s work can be assigned to its copying of protected material. A simpler, and perhaps more helpful, approach may be to rely more heavily on qualitative evidence—in particular, customer interviews—and simply to ask, when customers report that they consider a defendant’s work to be a substitute for the plaintiff’s, why they think that. If the answers tend to focus on similarities that arise from the defendant’s copying of uncopyrightable elements, then that tends to reduce the significance of the substitutability finding. Of course, that does not necessarily mean that the defendant’s work is transformative. But it does discount the possibility that defendant’s copying has summoned copyright’s “bête noire.”

E. Antitrust Methodologies and the Fourth Statutory Factor

Although Warhol innovated in introducing competition into courts’ analysis of the first statutory fair-use factor, competition has been part of the fair-use analysis since its inception. It is the focus of the fourth statutory factor, which inquires into “the effect of the use upon the potential market for or value of the [plaintiff’s] copyrighted work.”95 However, although the prospect of substitution has long been a consideration in the fair-use analysis, copyright law has never developed a methodology for assessing, in the fourth-factor analysis or anywhere else, the likelihood of substitution in a systematic or rigorous way.

The same antitrust methodologies that can be used to inform that substitutability analysis under the first statutory factor are relevant to the fourth. But after Warhol, there is a new relationship between the two factors. In her opinion for the Warhol majority, Justice Sotomayor articulates what she believes to be the distinction between the first and fourth factors: “While the first factor considers whether and to what extent an original work and secondary use have substitutable purposes, the fourth factor focuses on actual or potential market substitution.”96

Justice Sotomayor’s attempt to keep the first and fourth factors distinct is analytically weak: works that have “substitutable purposes” are likely to be substitutable in the eyes of consumers, and it is the prospect of actual market substitution, not some abstract similarity of purpose, that Justice Sotomayor elsewhere in her opinion links to the object of the first-factor analysis. Indeed, Justice Gorsuch, summarizing in his concurrence the majority’s principal holding, does not distinguish between substitution of purpose and commercial substitution:

[H]ere, the undisputed facts reveal that the Foundation sought to use its image as a commercial substitute for Ms. Goldsmith’s photograph. Of course, competitive products often differ in material respects and a buyer may find these differences reason to prefer one offering over another. But under the first fair-use factor the salient point is that the purpose and character of the Foundation’s use involved competition with Ms. Goldsmith’s image.97

As a consequence, in the subset of cases where transformativeness is at issue, the first and fourth factors have largely merged. In these cases, competitive analysis done to assess transformativeness under the first statutory factor is likely to drive the outcome of the fourth-factor analysis. However, when the defendant does not claim a transformative use, substitution will not be an element of the first-factor analysis, and will be undertaken, instead, in the courts’ consideration of the fourth statutory factor.

F. The Limitations of Antitrust Methodologies in the Fair-Use Analysis

Finally, it must be acknowledged that there are certain kinds of fair-use cases where the use of antitrust methodologies might suggest that a work is transformative, but where a transformativeness finding is nonetheless not likely to drive a favorable result for the defendant. An example would be an unauthorized translation of an English-language novel into a language, such as Zulu, spoken by a relatively small number of people. One can readily see that the vast majority of potential consumers would not consider the unauthorized Zulu translation to be a substitute for the English-language version. And we can sharpen the hypothetical further by positing that the authorized publisher had no plans whatsoever to offer an authorized Zulu translation or to license one—and so the defendant’s use creates no prospect of substitution in a “derivative” market. And yet it is unlikely that a court would rule that the unauthorized Zulu translation is a transformative fair use.

The explanation for this can be found not in a volume of logic but in a page—or less—of legal realism. The Copyright Act gives authors exclusive rights to prepare so-called “derivative works”98—that is, works based on a preexisting copyrighted work that alter the content of that work in ways that add new originality and create, in effect, a new, separately copyrightable work.99 Measured according to the competition criterion laid out in Warhol, our hypothetical Zulu translation may in fact be “transformative,” but it is nonetheless likely outside the scope of fair use. This is because translations are a category of work specifically mentioned in the definition of derivative works, and courts are likely to treat translations and other paradigmatic derivative works as presumptively within the scope of the copyright rights in the original works on which they are based. The same is true of the other types of works specified in the statutory definition of derivative work: a motion-picture version of a novel, for example, may be highly transformative in fact, and may not, in the view of consumers, serve as a substitute for the original novel. Indeed, it is possible that a particular motion-picture adaptation, if it is different enough from the underlying novel, may not even substitute for the plaintiff’s granting of a license to create a different motion picture—one, perhaps, that is more faithful to the underlying book. The motion-picture adaptation of a novel is nonetheless likely to be treated in almost all cases as a derivative work outside the scope of fair use.

In the Warhol case, Goldsmith attempted to make a version of this categorical argument for a type of work—a painting based on a photograph—that is not specifically listed in the definition of derivative work but rather is covered by the definition’s “catch-all” provision: that is, the part of the definition that would include as a derivative work “any other form in which a work may be recast, transformed, or adapted.”100 Goldsmith argued that because Warhol’s portrait was a recasting of her photograph, it was a derivative work, definitionally within her exclusive rights under the Copyright Act and outside the scope of fair use.101 The Warhol Court carefully stayed away from this argument, ruling only that “the degree of transformation required to make ‘transformative’ use of an original must go beyond that required to qualify as a derivative.”102 That conclusion is unsurprising: the “degree of transformation” required to create a derivative work is exceedingly low—the derivative work’s author need contribute nothing more than a “nontrivial, distinguishable variation.”103 In contrast, under the criterion set out in Warhol, a finding of transformativeness requires a variation sufficient to give the new work a different purpose—a distance that is measured by whether the new work can substitute for the original on which it is based.

The Warhol Court’s understanding of the relationship between derivative works and fair use is unlikely to please anyone laboring under the illusion that the Copyright Act can provide internal consistency in every case on the scope of a copyright owner’s exclusive rights. But for those who approach the law with more realistic expectations, the Warhol Court has laid out a defensible accommodation of two fundamentally antagonistic congressional commands.

Conclusion

There is a fair prospect that the Warhol Court’s focus on competition in determining whether a work is transformative under the first statutory fair-use factor could, if pursued with rigor in future cases, help to make the fair-use analysis both more predictable and more consistent with copyright’s principal goal (at least in cases where transformativeness is at issue). That goal is not to suppress copying as such, but to suppress potentially substitutive copying. The Warhol Court’s approach may also help to nudge the fair-use analysis in a direction that is modestly friendlier to defendants. Once competition analysis is situated in the context of specific uses, as the Court has done, the antitrust tools that we have to analyze competition between artistic and literary works are likely to suggest, in many cases, that copying by defendants that is not exact or nearly so will not meaningfully constrain the price charged for the original work. The fair-use analysis should treat these nonsubstitutive uses more solicitously, as their production advances copyright’s principal goal of encouraging the production of new works.

Murray and Kathleen Bring Professor of Law, New York University School of Law, and Co-Director, Engelberg Center on Innovation Law and Policy. The author thanks Daniel Francis, Stefan Bechtold, Chris Buccafusco, Pam Samuelson, Mark Lemley, Mark McKenna, Jennifer Rothman, Tony Reese, Lydia Loren, Daniel Gervais, Molly van Houweling, and participants at the 2024 Intellectual Property Scholars Conference at the UC Berkeley School of Law for helpful comments and conversations. The author also thanks Jackson Livingstone for excellent research assistance, and the Filomen D’Agostino and Max E. Greenberg Research Fund for grants that supported this work.