Tag: fair use

Navigating Contemporary Fair Use

Shortly after the Second Circuit Court of Appeals ruled in Warhol v. Goldsmith, I wandered deep into the tall grass, PJs tucked into my boots, thinking about the fair use analysis in the difficult context of fine art. And then the Supreme Court delivered Justice Breyer’s opinion in Google v. Oracle, which, among other transgressions, broke a cardinal rule by weighing a fair use defense without first addressing the challenge to copyrightability of the work at issue. The result was a train wreck of legal contradictions in which a core copyrightability question was dropped into prong two of the fair use test like a marshmallow in a martini.

But one thing Goldsmith and Google have in common is that both opinions confirm that when it comes to the fair use doctrine, legal practitioners, the courts, and creators seem to be hiking through an ever-thickening wood without map or compass. This may be especially true with regard to distinguishing fair uses from derivative works, and I say that despite the extent to which the court in Goldsmith showed deference to derivative works—because it did so without offering much clarity on the matter. But let’s back up a bit.

I think there are a lot of moving parts in the Goldsmith case that don’t fit into a single post of ordinary length; and although the following contains several personal opinions, a few of which may diverge from my pro-copyright friends and colleagues, it is all offered as food for thought—asking more questions than presuming to have answers. Further, this post assumes a certain familiarity with Goldsmith and the related cases, but here’s the background as condensed as I can make it:

Background – Andy Warhol Foundation (AWF) v. Lynn Goldsmith

Lynn Goldsmith captured the photograph at issue in 1981, during a truncated photo session with the semi-reclusive musical artist Prince Rodgers Nelson, who was then barely known to mainstream audiences as Prince. Goldsmith contends that she made certain creative choices resulting in an image of the “vulnerable human being” behind the persona. The photograph was never published but was licensed in 1984 (unbeknownst to Goldsmith) to Vanity Fair as a “reference photo” to produce an illustration to accompany a story about musician’s rise to stardom. In fact, the photo was used by Andy Warhol make a series of silkscreens similar to those he has made of Marilyn Monroe, Mao Zedong, etc., all using photographs as original sources.

Goldsmith was not aware of the existence of the Warhol screens until 2016 when, after Prince’s death, Vanity Fair published a special edition with one of the Warhol versions on the cover. At that time, Goldsmith communicated to the Andy Warhol Foundation (AWF) that the works may infringe the copyright on her still unpublished photo, and in response, AWF filed suit seeking a declaratory judgment of non-infringement or, barring that, a finding of fair use. The district court held the Warhol screens to be fair use, primarily by following the Second Circuit ruling in Cariou v. Prince, but on appeal, the court reversed, finding the Warhol screens are not fair uses.

Warhol v. Goldsmith: Maybe Transformative & Fine Art Don’t Mix?

“Just because you can find meaning in art, doesn’t necessarily mean that was the meaning the artist necessarily intended to be there.”– Daniel Sloss, comedian, taking the piss out of a pretentious interpretation of a Jackson Pollock painting.

Although my sympathies are squarely with Lynn Goldsmith, especially because AWF sued her, I am not convinced the Second Circuit came to a sound conclusion—at least not one that can easily be called practicable for future artists. While the court did provide a fresh reminder that there is an important boundary between a transformative use tilting toward fair use and a protected derivative work, this opinion also emphasizes how nettlesome analyzing transformativeness under factor one of the fair use test remains.

Attempting to describe a transformative (potentially fair) use by means of the fallible, subjective English language is difficult enough in many circumstances. Terms like new meaning, new context, new purpose can be so ambiguous that a district court (in Brammer) held that using an “expressive” photograph for an “informational” purpose was transformative. That error demonstrates what can happen when subjective language confronts works like photographs, which are almost always both expressive and informative. Brammer was easy enough to reverse on appeal because the user copied and displayed nearly the entire image for a commercial purpose that classically requires a license. But in a fine art context, where works like reference photos are used all the time, terms like new meaning etc. may be as unhelpful or capricious as asking words to define the secondary artworks themselves.

So, although the Second Circuit provided new language affirming that there is a line between transformative and derivative, if I put myself in the shoes of the next artist seeking guidance from the outcome in Goldsmith, I am not exactly sure where the line is. Especially when the court stopped short of calling the Warhol screens “derivative works.” To put this in visual, practical terms, the artist poised to make use of a reference photograph is told that as things stand, the image on the left is fair use but the image on the right is not:

Left: Secondary use in Keinitz v. Sconnie Nation. Right: Secondary use in AWF v. Goldsmith.

Critical or aesthetic distinctions notwithstanding–and without knowing anything about the original or secondary users’ intent–what are the legal distinctions one can infer from these two images that tell the next artist where the line is? The image on the left, according to the Seventh Circuit has stripped away the protectable expression in the reference photograph, while the image on the right, according to the Second Circuit, has copied too much of the protectable expression from the photograph. And if the next artist asks her attorney for guidance, the answer probably should not be that she should only use reference photos made by authors in Illinois, Indiana, and Wisconsin.

For all the effort the Second Circuit made to clarify the distinction between Goldsmith and in its findings in Cariou—and as informative as that part of the opinion may be to the transformative test (stress on may)—reckoning Goldsmith and Cariou was not really the difficulty in this case in my view. Nevertheless, the distinction does provide a useful framework.

Distinguishing Goldsmith from Cariou

For starters, the circuit court held that the lower court applied too subjective a test by finding that the Warhol screens “can reasonably be perceived to have transformed Prince from a vulnerable, uncomfortable person to an iconic, larger-than-life figure.” To this, the reversing opinion replies:

“…the district court appears to have read Cariou as having announced such a rule, to wit, that any secondary work is necessarily transformative as a matter of law ‘[i]f looking at the works side-by-side, the secondary work has a different character, a new expression, and employs new aesthetics with [distinct] creative and communicative results. Although a literal construction of certain passages of Cariou may support that proposition, such a reading stretches the decision too far.” (citations omitted)

So, it’s like that. We know what we said, and you could definitely read it that way, but don’t. But let’s be legal laymen for a moment and look at some of the evidence side-by-side:

Left: Photos by Patrick Cariou. Right: “James Brown Disco Ball” by Richard Prince.

To the observer without any copyright knowledge whatsoever, there is an objectively perceptible difference between most (if not all) of the Richard Prince works that were held to be fair uses of Cariou’s photographs and the Andy Warhol silkscreens, now found not to be fair uses of Goldsmith’s photograph. Prince’s collages (e.g. James Brown Disco Ball), whatever one thinks of them, are not in any sense reproductions of Cariou’s expressions. They are something else entirely for which Cariou’s photos were used as raw materials in combination with other materials to create a new work. As the court states in the Goldsmith opinion:

“… in Cariou, the copyrighted works found to have been fairly used were, in most cases, juxtaposed with other photographs and “obscured and altered to the point that Cariou’s original [was] barely recognizable.” (citations omitted)

Left: “Prince” by Lynn Goldsmith. Right: One of the silkscreens made by Andy Warhol.

By contrast, Warhol’s screens are clearly made from Goldsmith’s photo, which any ordinary observer can see, but the legal question is whether they copy the protectable expression of the photograph, or merely the facts of it. Goldsmith asserts her authorship, partly by declaring that she made certain decisions in order to capture the “vulnerable, human” Prince, while AWF asserts that Warhol rendered Prince as “icon,” which is arguably the exact opposite expression. In both cases, the court discusses process, which can be problematic for two reasons: 

First, copyright protection is technically agnostic with regard to how a work is made as long as the finished product meets a minimum standard for originality, expressed in the two-prong standard in Feist that the work be independently created and that it reveals a modicum of creativity. Second, the Goldsmith opinion avers that it has sequestered intent from its consideration thus:

“Though it may well have been Goldsmith’s subjective intent to portray Prince as a “vulnerable human being” and Warhol’s to strip Prince of that humanity and instead portray him as a popular icon, whether a work is transformative cannot turn merely on the state or perceived intent of the artist or the meaning or impression that a critic –or for that matter, a judge – draws from the work. Were it otherwise, the law may well ‘recognize[e] any alteration as transformative.’” (citation omitted)

But this view is not compatible with any discussion of process in the production of art because intent and process are often intertwined until they become almost synonymous. To wit, the artist frequently says, “I wanted to express X, so I made decision Y with materials Z.” Nevertheless, stripping intent from the analysis makes sense up to a point. Intent is a highly subjective property of creation; it is apt to be amended or even invented for a complaint or defense brief; and in many cases, as with Richard Prince, an artist may reject the very notion of intent other than “to make art.”

Objectively, I do not think the distinction between Cariou and Goldsmith is unclear; it squares with what I suggested above that any ordinary observer can perceive. The Cariou holding rests on the foundation that “the secondary work itself must reasonably be perceived as embodying an entirely distinct artistic purpose, one that conveys a ‘new meaning or message’ entirely separate from its source material.”

The Richard Prince pieces that were held fair uses would seem to fit that description, but if Warhol’s screens do not meet that standard, is it because too much of Goldsmith’s expression was copied? Or is the transformativeness in the Warhol screens too subtle for this particular panel of judges to notice? Or is the language used to describe transformativeness too obtuse for this entire conversation? Because I strongly suspect those words meaning and message cause a lot of trouble when the transformative principle arrives in the salons of the fine art world.

Fair Use is Founded on “Message”

The seminal case in weighing the fair use consideration called “transformativeness” was Campbell v. Acuff-Rose (1994). There, 2 Live Crew used a substantial amount (the heart of the work) of the song “Oh, Pretty Woman” to produce a rap version that was as nasty as they wanted it to be, and the Supreme Court held the secondary work to be a fair use as a parody.[1] Seminal though it was, Campbell was a relative cakewalk as a consideration. The use conveyed parody that an ordinary listener could hear; the defendant said it was parody; and parody is a paradigmatic purpose implied in the statutory terms “comment” and “criticism” as examples of purposes amenable to a fair use defense. Moreover, the Orbison/Dees song was necessary to 2 Live Crew’s new expression. And a consideration of necessity substantially informs factors one and three.

The four fair use factors are meant to be weighed interdependently, and there is a strong, magnetic interplay between factors one and three which implies a consideration of necessity. Factor one assesses the purpose and character of the use, while factor three assesses the amount of the original work used to fulfill that purpose. As Terry Hart discusses in a post written after the Second Circuit’s Cariou decision, necessity is a cornerstone of fair use jurisprudence. He writes, “… throughout the history of fair use, there has been this idea that some overriding purpose is required to privilege a use that would otherwise be infringing. And the use of the original work must be necessary to the new work.” [Emphasis added]

Although a court opinion will usually reiterate that the four factors are weighted holistically and interdependently, the question of purpose under factor one—and therefore the consideration of transformativeness—appears to take on outsize significance in many contemporary fair use cases. Indeed, scholars and practitioners tell us that factor four, which considers potential harm to the market for the original work, used to be the more dominant factor until “transformativeness” was introduced to jurisprudence in 1990.

In considering the elevated prominence of transformativeness, I believe it is reasonable and relevant to generalize that the fair use doctrine evolved in contexts in which the user’s purpose (i.e. “message”) was fairly easy to define, and this probably remains true in most cases. It is no accident that the statutory examples in Section 107, though not exhaustive, imply commonly understood forms of expression and, moreover, imply some element of commentary—in many cases, nonfiction commentary—upon the work being used.

Thus, the doctrine of necessity, as Hart describes in his post, is fairly straightforward. We do not need legal educations to understand why the historian or journalist needs to quote certain works or why the parodist needs to copy enough of an original to make the joke work. In these examples, ordinary observers usually share a common understanding of what the secondary works are saying as well as an understanding of the nature of the work used. Indeed, one rule of fair use as parody (Leibovitz v. Paramount) is that the nature of the original expression must be sufficiently understood in the consciousness of the observers in order to assert parody. My point is that purpose is typically intertwined with a broadly understood “message,” and this is the context in which we traditionally think about fair use. Thus, most of the language in fair use case law ably suits this these analyses.

But a communicable assessment of purpose can break down in the quantum universe of fine art where “message” is often eschewed or becomes subjective to a point of whimsy that some find obnoxious, but which may also be rescued in a critical sense by the principle known as art of for art’s sake. Richard Prince, whatever anyone thinks about him or his work, makes art for art’s sake. And although many other artists will call him “lazy,” I am not sure that he is always “avoiding the drudgery of working up something new” in a legal sense, at least not in the Cariou examples held to be fair uses.

Richard Prince, in this regard, is comparable to Marcel Duchamp (or Warhol for that matter), appropriating bits and pieces of creative works rather than “found objects” and treating these as raw media without caring about their original meanings or certainly the original authors. He treated Cariou’s Rastafarian photographs as “raw elements,” and as Terry Hart notes in a post arguing that Cariou was wrongly decided, finding fair use in those instances was doctrinally out of step with the Supreme Court’s instructions on transformativeness. He writes:

“… the Second Circuit disregarded the Supreme Court’s direction on transformativeness. That Prince was using Cariou’s photos as raw materials to “avoid the drudgery in working up something fresh” rather than building upon them seems readily apparent, especially given Prince’s own testimony during the trial that he “‘do[es]n’t really have a message,’ that he was not ‘trying to create anything with a new meaning or a new message,’ and that he ‘do[es]n’t have any . . . interest in [Cariou’s] original intent.’”

But does Hart’s incontrovertible observation mean that the Second Circuit invented a new fair use standard, or does it imply that the language used to describe transformativeness is hopelessly inadequate to assess potential fair uses in certain contexts? Because fine art may not be about message at all. Quite often it is not, unless somebody makes one up.

It is relevant to note that the courts tend to imbue transformativeness with subjective measurement, using terms like “slightly transformative” or “highly transformative,” and indeed anything other than a binary yes or no may be vexing for practitioners. But suffice to say, the courts do seem to agree that there is such a thing as more transformative or less transformative, which is probably a natural temptation in an analysis that purports to weigh interdependent factors.

But if that is the case, and a court cannot define what the new expression is other than “art,” then it seems the court cannot rationally assign a “more” or “less” value to transformativeness in such an instance. In Cariou, the court found transformativeness in a use without “message” by upending the Supreme Court’s instructions, finding that, “Prince’s work could be transformative even without commenting on Cariou’s work or on culture, and even without Prince’s stated intention to do so. Rather than confining our inquiry to Prince’s explanations of his artworks, we instead examine how the artworks may ‘reasonably be perceived’ in order to assess their transformative nature.” [emphasis added]

So, is Cariou one of the worst decisions in modern fair use jurisprudence—a “carve out for Richard motherfucking Prince”—as one attorney called it on Twitter? Or is it a sophisticated and nuanced interpretation of transformativeness, but which perhaps disserved the plaintiff by skewing other factors in the fair use analysis?

Because the analysis of purpose and character is historically grounded in definable forms of expression, other factors should perhaps supersede the search for transformativeness in cases where “message” cannot be clearly defined, where the defendant rejects the idea of “message,” and where we do not want judges to act as art critics and invent “messages” they perceive, which may be rejected by other observers, including the defendants themselves.[2] In such circumstances, the court has little or nothing upon which to hang a transformative analysis, and the doctrine of necessity evaporates altogether.

If secondary artworks cannot tell us what they are conveying, a court cannot rationally weigh how necessary the works used were to the new expression or how transformative the new work is. At best, the court can find that using precedent works as “raw materials” could tilt toward fair use, though in some cases, perhaps the exception is more properly de minimis use.

In fine art, where “message” does not necessarily define purpose, the fair use analysis can be frustratingly vague under factor three as well. If the user’s purpose is undefinable, the court can only guess whether the amount used was the “least amount necessary” to that purpose, which boils down to taking the user’s word for it—that he used what he needed to achieve an expression that he himself refuses to describe. And the Second Circuit in Goldsmith said the courts should be deaf and blind to such claims by the artist in general.

So, what if the courts demote the need to immediately describe transformativeness in the equation and simply affirm, for instance, that both James Brown Disco Ball and the Warhol screens are new and hold this newness to be purpose enough to proceed with the analysis? In such an instance, it seems, the analysis could begin with factor three, but with the understanding that the absence of purpose warps the “amount necessary” consideration. So, what the court is left with is reproduction. Because, in the simplest (i.e. layman-friendly) terms, it seems the Second Circuit found that James Brown Disco Ball does not reproduce Cariou’s expressions but that the Warhol silkscreens do reproduce Goldsmith’s expression. As such, should the court really proceed with a fair use analysis at all?

What is a Derivative Work of a Photograph?

Although the court sufficiently explains the distinction in reasoning between Cariou and Goldsmith, it does not compellingly explain why Warhol copied anything more than the facts of Prince from Goldsmith’s photograph, especially when the opinion stops short of finding that the Warhol screens are derivative works “as a matter of law.” Because if not, what are they? According to this court, the screens are more akin to any other print that Goldsmith might have made herself (e.g. a version where she burns and dodges to change the contrasts) from the original negative. But that begs a question not answered by this court, and one not easily answered in general: what exactly constitutes a derivative work of a photograph?

Nearly all photographs are a mix of protectable expression and unprotectable fact. A portrait in front of a plain background, no matter how expressive the face or pose may be—and no matter how much effort the photographer makes to extract the look from her subject—still contains a substantial amount of fact that is simply not protectable.

In the seminal case affirming copyright in photography (Burrow-Giles v. Sarony, 1884), Napoleon Sarony did not invent the fact of Oscar Wilde in the portrait at issue, but the Court held that the lithographic company reproduced all of Sarony’s expressive choices—background, positioning, propping, etc.—some of which required considerable effort and care in 1882. But if the Wilde photograph had been a headshot against a plain backdrop, the Supreme Court might have come to a very different conclusion—not that the lithographer had made a fair use, but rather that Sarony’s photograph contained insufficient authorship for copyright to protect.[3]

Jump to 1981, and we should by no means deny Lynn Goldsmith authorship in her photograph of Prince—doctrine has evolved, and I think correctly, to embrace more nuance since 1884. But if we follow the Second Circuit’s reasoning—i.e. that Goldsmith’s authorship is not “Prince’s vulnerability” and Warhol’s secondary authorship is not “Prince as icon”—then it seems that a consideration of what the two works share is either “Prince as fact” or “Prince as expression,” the former fair use, the latter infringing. My personal call would probably tilt toward “Prince as fact,” with each work imbued by relatively thin, but distinguishable, layers of authorship in both cases. Ergo, the Warhol screens, especially presented as a series in a multi-screen artwork, are more properly a fair use in my view.* But if they are not fair use, then it seems the court should have at least unequivocally called the screens derivative works or reproductions.

Instead, adding what seems like further confusion, the Goldsmith opinion concludes by describing the Warhol screens as “substantially similar” to the photograph, which seems like a strange finding to follow a fair use decision, especially, when considering an image with relatively thin protectable elements. Because to the ordinary observer (i.e. next creator), isn’t the court essentially calling the screens “copies” without quite saying so? And if they’re copies, the court should should end its analysis at §106 and not bother with §107 at all.

Derivative Works and Factor Four

It is intriguing (frustrating?) that the court declined to call Warhol’s screens derivative works given that the misapplication of a derivative works analysis can lead to poor considerations under factor four, which is arguably where the Second Circuit did the greatest disservice to Patrick Cariou. Factor four considers the potential harm to the market for the original works. And although one can sustain the finding that Richard Prince’s works make use of Cariou’s images in ways too unrecognizable to have any effect on the potential market for the original works, the court did not need to leave us with this observation about factors extrinsic to the works themselves:

Prince’s work appeals to an entirely different sort of collector than Cariou’s. Certain of the Canal Zone artworks have sold for two million or more dollars. The invitation list for a dinner that Gagosian hosted in conjunction with the opening of the Canal Zone show included a number of the wealthy and famous …. Cariou on the other hand has not actively marketed his work or sold work for significant sums, and nothing in the record suggests that anyone will not now purchase Cariou’s work, or derivative non-transformative works (whether Cariou’s own or licensed by him) as a result of the market space that Prince’s work has taken up.

That’s the kind of language which invites criticism that the Second Circuit carved out an exception for Richard Prince (i.e. that the court was distracted by his celebrity), and it is not easy to reconcile with its declaration in Goldsmith …

Finally, we feel compelled to clarify that it is entirely irrelevant to this analysis that ‘each Prince Series work is immediately recognizable as a ‘Warhol.’ Entertaining that logic would inevitably create a celebrity-plagiarist privilege; the more established the artist and the more distinct the artist’s style, the greater leeway that artist would have to pilfer the labors of others.

Agreed in principle, but damn. In order to justly consider under factor four whether a secondary work may usurp the market for the original, the court should imagine the original author’s potential means of exploiting his works. And although Cariou would be unlikely to use his photos in a manner akin to Richard Prince, photographers straddle the worlds of photojournalism and fine art all the time. As such, the court should no more have described the celebrity world occupied by Richard Prince than it should give deference to same for Andy Warhol. The light reflecting off the fame of a particular user can obscure the court’s view of a derivative work that may be advertised as a transformative fair use.

BUT (and this is one reason this case keeps me up at night), there may be instances where an artist’s “message,” like the unmistakable Warholness of the many silkscreens he made in his career, is a factor that reasonably makes a secondary work a new expression with new meaning or message. Campbell’s soup cans on a shelf are just soup. Campbell’s soup cans painted on a canvas—or any medium—are “Warhol’s soup cans.” Another artist could not hope to use that graphic element without alluding to Warhol. Like it or not, Warhol gave that mundane packaging new meaning, and that meaning cannot be entirely compartmentalized from the celebrity attached to the act of making the works.

So, to return to the start of this epic journey inward, if we place ourselves in the studio of the artist about to make use of a reference photograph for a new work, shifting her gaze back and forth between the image in Sconnie Nation and the Warhol screens, what does she conclude? I’m still not sure, but I may have to agree with one commenter on a Facebook group I help moderate that the Supreme Court just might grant cert if AWF pursues an appeal that far. Though after Google v. Oracle, I would not necessarily have high hopes for clearer direction on fair use.

Implications for New Challenges

Finally, it strikes me that a finding of fair use of works as “raw materials” in Cariou has more kinship with the opinion in Google Books than it does with assessing the purpose of a secondary work of artistic expression. There, books were used as raw materials to create a search tool, but without making whole books available as a market substitute. Similarly, in Cariou, what the court held is that Richard Prince made use of photographic elements as raw materials to create secondary artworks that did not make the original photographs available as a market substitute. In other words, perhaps Cariou could have been a more streamlined, yes or no, holding under prong one without the need to get bogged down in artsy adjectives stapled onto the already fuzzy interpretations of “transformative.”

Whether readers agree with that distillation, it seems likely that the conclusions we can draw from Cariou and Google Books will feature substantially in the brewing debate over whether creative materials may be fairly used in machine learning to teach AI to produce new creative works. But that’s another big discussion for another day.

*CORRECTION: This should have been properly stated as non-infringing because it does not copy protectable expression, not because it is fair use.

[1] Technically SCOTUS remanded, but as a shorthand, we refer to this case as though the Court made this finding.

[2] Yet, in the Goldsmith opinion, we see how difficult this is to avoid. Simply by writing descriptions of the Prince works they held to be fair uses, using words like “jarring” and “crude” to contrast with the originals they describe as “serene” and “deliberately composed,” the court, to an extent, is inventing message in order to satisfy the mandate to look for transformativeness.

[3] This does not mean the constitutional argument against protecting photographs under copyright would have prevailed in Burrow-Giles v. Sarony, only that the Court might have found nothing to protect in the image at issue.

Google v. Oracle: A Troubling Use of Fair Use

Once the die was cast (i.e. after oral arguments) in Google v. Oracle, I don’t think I was alone in feeling that if the Supreme Court held that the computer code at issue in this case was not properly a subject of copyright protection, that would be an acceptably narrow decision, even though many might disagree with it as a statutory reading.1 On the other hand, I and others felt that if the Court held Google’s use of Oracle’s code to be a fair use, that would be potentially harmful to copyright law in general, and perhaps a disservice to many parties with an interest in this case—except of course to Google.

The argument against the copyrightability of the code at issue (a.k.a. declaring code or APIs), was founded on the merger doctrine, which holds that where an expression and its idea or function are inextricably bound together, the expression cannot be protected by copyright. Here, the majority opinion, written by Justice Breyer, declined to address that consideration, instead stating that “for the sake of argument,” the Court would assume the code is protected, and then it still delivered an opinion on the copyrightability question by couching it in a fair use analysis.

Although the Court did state that the decision is unique to software and does not “overturn or modify earlier cases involving fair use,” we must hope that summary can be reconciled with the part of the opinion which states, “The upshot, in our view, is that fair use can play an important role in determining the lawful scope of a computer program copyright, such as the copyright at issue here.” That is a troubling generalization, which forces the fair use doctrine (a case-by-case defense) to do what the Court otherwise could have done instead, which would have been to write an opinion limiting the statutory protection for the code at issue, and also limit its own finding. In that instance, fair use should not even be considered. Instead, the Breyer opinion asks fair use to do something it is not meant to do.

As a result, this outcome not only could disturb case law, but it also falls short of providing the market certainty many in the software business were seeking in the briefs filed on behalf of Google. For instance, the brief signed by 83 computer scientists stated, “Forcing companies that reimplement APIs to rely on fair use will not meaningfully address … anticompetitive effects. Though better than nothing, a fair use standard creates uncertainty because it depends on fact-intensive, case-by-case determinations which can result, as this case demonstrates, in lengthy and expensive litigation.”

So, rather than providing that hoped-for certainty, by either accepting or rejecting the merger argument, it is notable that the opinion instead asserts factor two (nature of the protected work) of the fair use test ahead of factor one (purpose and character of the use) in order to frame its conclusion with a lengthy discussion about what declaring code does and why the Court agrees it is distinguishable as an unprotectable type of code. The opinion states:  “It is inextricably bound to­gether with a general system, the division of computing tasks, that no one claims is a proper subject of copyright. It is inextricably bound up with the idea of organizing task into what we have called cabinets, drawers, and files, an idea that is also not copyrightable.”

That is extremely close to saying the code at issue fails for copyright protection under merger. So, why didn’t the Court just go ahead and make that finding rather than risk sowing added confusion by stuffing a pseudo copyrightability opinion into prong two of a fair use analysis? Because now, it is possible that a greater number of parties will be disserved by this outcome.

Perhaps what happens in the market should not weigh too heavily where the Court restricts its opinions to questions of law, but this opinion makes clear in its fourth factor analysis that it is terribly concerned about broad market effects. And its assumptions about the market, especially the implications beyond software, seem divorced from reason as a fair use consideration. The factor four opinion suggests the majority was unduly persuaded by the argument that Sun was unsuited as a developer to create a product like Android. “…evidence at trial demonstrated that, regardless of Android’s smartphone technology, Sun was poorly positioned to succeed in the mobile market.” That is potentially a market-devastating view, and here’s why:

First, if a party authors a work that some other entity is potentially better at exploiting, that is grounds for licensing the work, not appropriating it. Consequently, the Court’s failure to hold, in a more straightforward ruling, that the code copied was not protected fosters this more insidious interpretation of “market harm” in its fourth factor analysis.

Second, the biggest gorilla in the sandbox just got a bonus prize. After all, won’t a company like Google always have the resources and capabilities to build the next doodad faster and better than another entity? But in Google v. Oracle, to “level the playing field,” the Court just held that a startup which cannot compete with the Googles of the world, may not necessarily license its IP to the giants either, depending on how one interprets this fourth factor reasoning. Because let’s remember, Oracle ain’t exactly a startup.

Third, imagine we are looking beyond software, and this fourth factor opinion can be argued to mean that, for instance, the novelist who is “poorly positioned” to make a film adaptation of her book is subject to a similar finding in relation to a film studio appropriating her work. Granted, that’s a bit extreme, and she would, we hope, be protected by other considerations in the law. But I use the example to underscore how flawed this view of “market harm” is as a matter of principle.

All the ink spilled in this part of the opinion, lauding the value of smartphones, the quality of the Android system, etc. is an argument that only proves market harm to Oracle due to the failure by Google to obtain a license—and one that simply waves a hand at any implication of the derivative works right. “Given the costs and difficulties of producing alternative APIs with similar appeal to programmers,” the opinion states, “allowing enforcement here would make of the Sun Java API’s [sic] declaring code a lock limiting the future creativity of new programs.”

That language, which connotes a hostility to copyright in general, upends the law by failing to acknowledge that licensing works does not lock up works. We have over 200 years of evidence and jurisprudence to back up that general premise. But again, if this Court accepted the market rationale for this outcome, then it should have held the code at issue unprotectable rather than write a fourth factor “market harm” analysis that describes any rightsholder as “poorly positioned” to exploit a particular use of their works.

By transforming a copyrightability opinion into a fair use analysis, the Court seems to have fallen for the temptation to limit copyright’s protections based solely on works already developed, while failing to more expansively imagine works that may or may not be developed in the future. Aside from the potential damage done to other copyright subjects by this fair use holding (despite the Court’s caveat), the opinion does little for the next software venture, except to tell its principals that when a Google-scale behemoth appropriates some amount of their code, they may be about a decade’s worth of litigation away from finding out if there’s a remedy. And the number of new ventures that can afford that is zero.

  1. I wrote against the merger argument here and here.

Google v. Oracle XIII:  SCOTUS Should Be Skeptical of the “Sky Will Fall” Argument

I realize the Court will rule anytime now, and that I may be gilding the proverbial lily here; but I drafted this post in early January, and then the world got a little crazy and distracting. Anyway, FWIW, below is my last observation about Google v. Oracle. At least until after the decision. 🙂

In Google v. Oracle, the Supreme Court will render opinions on two legal arguments, either of which could have profound effects for different interests. The Court’s opinion on the copyrightability of Oracle’s “declaring code” will, in one way or another, be felt throughout the software industry, while the Court’s opinion on fair use will affect the entire ecosystem of creators in every category of copyrighted works.

As discussed in other posts, the Supreme Court should reject Google’s attempt to hyperextend the purpose and character of fair use, and in doing so, it should unanimously decline to transform modern copyright doctrine from the bench. In earlier posts, I discussed why Google’s claim that the code at issue should not be held uncopyrightable under the “merger doctrine” (§102(b)), which would have to affirm that the code at issue is a method rather than a form of expression. Nevertheless, the Court may feel hesitant to “upend the software industry,” if it is persuaded that finding copyrightability in Oracle’s code might have this result.

The most compelling argument in this regard is presented in the amicus brief filed by eighty-three computer scientists, which includes some of the most renowned names in software development over the last half century. It is hardly sensible for most of us—and certainly not for me—to debate that industry’s conduct with the likes of Steve Wozniack et al. If these experts say that “reimplementation” of software interfaces (APIs) is standard practice that the software industry has relied upon for decades, that statement must be given both deference and weight.

At the same time, we must keep in mind that “reimplementation” is not barred by copyright—that in fact much of the “open source” copying in that industry is bound by various conditions, which are defined by licensing agreements that are only enforceable under copyright law. In that regard, Java is a classic example of code that offers different tiers of licensing where, for instance, the educator may access all of Java for free, while the commercial user is subject to fees and other conditions. There is nothing remarkable or inherently stifling about these distinctions.

More specifically, as a question of law, even if we accept the computer scientists’ broad description of industry-wide reimplementation as fact, it tells us nothing about whether there is sufficient creativity in Oracle’s declaring code to qualify for copyright protection. In reviewing the various briefs filed by experts on both sides of this case, it seems clear that some declaring code is quite simple, and some is very complex—and creativity, presumably, expands with complexity. Further, there does not appear to be much if any quarrel with the premise that declaring code can be highly creative—easily creative enough for copyright to attach—and if that is correct, that should be the ballgame as a legal matter, regardless of industry practice and expectations. And Google has conceded that Oracle’s declaring code is creative.

This does not mean, however, that the Court will be wholly unsympathetic to the “standard practice” argument, or eager to disturb an entire industry if they believe this could be a consequence of its decision. So, let’s consider the argument a bit further, assuming the computer scientist amici are absolutely right on key facts, but perhaps a shade over-saturated in coloring their picture of the broader landscape relative to Google v. Oracle. For instance, I would pay attention to language in the brief that makes statements like, “Android is the most popular [mobile OS] in the world,” which is presented more than once in defense of Google’s reimplementation of the Java APIs to ultimately “transform” the mobile market.

That sentence caught my attention because the word popular implies consumer choice, which is in fact very limited in the mobile market. If the consumer is a dedicated Apple user, those phones are quite expensive. Alternatively, if the consumer needs a more affordable mobile device, she can choose among different phones that are nearly all running on one OS called Android.* And Android was not made freely and widely available as a gesture of Google’s largesse, or for the purpose of fostering competition of any kind.

While Google seeks to frame its free mobile OS as both generous and revolutionary, consumers have largely come to understand the digital-age axiom that if you’re not the customer, you’re the product. Google no more gives away Android “for free” than it does any of its other platforms. Consumers and various government agencies investigating antitrust practices fully recognize that the price of “free” has been to allow companies like Google to accumulate and manipulate data that is then used to alter consumer behavior, stifle small business in various markets, generate advertising revenue from the exploitation of often-questionable content, and, above all, to solidify their own market dominance.

This is not to say that if Google had licensed the Java code at issue, it would not still be the leading supplier in the mobile market—but that’s part of my point. The reason I homed in on this fallacy of Android’s “popularity” is that it informs a response to the claim in the computer scientists’ brief, which argues that “Uncopyrightable software interfaces address network effect barriers by enabling startups to plug into existing systems and innovate through cumulative improvements.” [Emphasis added]

While that sounds plausible as a generalization, in this particular case, the Court should be mindful that the forces buttressing Android’s market position—especially the network effects—render Google nearly immune to competition from startups. And these forces have little to do with copyright one way or another.

Android is a poor context in which to discuss “addressing network effect barriers.” Google’s market-share and wealth makes the company the very definition of a “network effect barrier.” As such, it seems equally possible that copyright (i.e. a mandate to license the code) is the only protection that a prospective startup has while attempting to thrive in a market presently conquered by the Googles, Amazons, and Facebooks. So, while a startup may get off the ground by copying some aspects of an already-dominant platform, the weakness Google now asks the Court write into copyright law would allow Google to turn around and copy the innovative aspects created by the startup, thereby crowding the startup out of business.

So, when the computer scientists’ brief describes competition in the market, it seems that it is often alluding to intramural competition on a technological playing field owned by one or a few prevailing companies. For instance, there may be competition among developers writing apps for the Android platform, but there is no startup, at least not in the American market, that can feasibly challenge Android for a piece of its share in mobile. And if such a startup were to emerge, it seems farfetched to allege that licensing declaring code, for instance, would be the barrier to stifle that prospective venture. Instead, it seems more likely that the barriers to that potential competitor are much more potent market forces that have little to do with copyright law in general, and nothing to do with the copyright questions presented in this case.

Are the Generalizations Instructive?

Quite possibly, the most intriguing segment of the computer scientists’ brief is where it describes how many developers, including Sun Microsystems itself in the development of Java, have reimplemented software interfaces in the process of bringing their products to market. This section presents a very clear portrait of standard industry practice, but it also reprises those two bugaboo questions I’ve asked before: 1) If unlicensed reimplementation has been so standard for so long, why did other commercial developers license Java declaring code for various purposes?; and 2) Why did Google itself almost enter into a license with Oracle that it only declined due to interoperability conditions with which it did not wish to comply?

Looking at this narrative as an outsider and giving all parties in the computer expert world their due respect, it is hard not to feel that, amid the generalizations about industry practice and innovation, some details are missing that are intrinsic to this case. Either declaring code is never the subject of copyright OR it is always the subject of copyright, OR some declaring code is properly protected while other declaring code is properly not protected. This latter conclusion would depend upon the amount of originality in the work, just like every other copyright category. And again, there seems to be consensus among all software experts that some declaring code can be highly creative, or as Deputy Solicitor General Malcolm Stewart described at oral arguments:

 …the briefs talk about the practice of copying interfaces or APIs, but those terms are very vague and potentially expansive. And a lot of things that might be called interfaces would be segments of code that are so short that they don’t exhibit necessary creativity, segments of code that are necessary to preserve interoperability. It may be that in particular circumstances, particular interfaces can be copied without authorization, but that’s not a basis for a general rule.

In other words, broad statements about industry practice, no matter how many names sign an amicus brief, can obfuscate the salient details in this case, as well as countless other scenarios in the software universe where reimplementation is ably supported by licensing agreements. This begs one of the real questions at issue, which is who benefits most from the bright-line rule the Court is being asked to make on the copyrightability of computer code—the independent software developer or the entrenched giant? While Google’s computer industry amici ask the Court to imagine how StartupXYZ benefits from copying GiantXYZ’s code, it also asks the Court to ignore the inverse scenario when GiantXYZ copies StartupXYZ’s code. It is easy to forget this when neither party in this lawsuit is a startup, but it is a question that should not be lost in a river of generalities.

Computer Scientist Brief Says Fair Use is Not Enough

Interestingly, the computer scientists’ brief asserts that a finding of fair use for Google would be of insufficient value to the software industry overall because this “would create uncertainty” in the trade. Naturally, a holding that declaring code is simply never protected is far more certain than a narrow finding of fair use in this one case, which would not preempt future litigation over copying the same kind of code. Thus, the computer scientists’ brief confirms that a finding of fair use would only help Google while asserting it would do little for the industry as a whole.

That’s just as well since finding fair use in Google v. Oracle would, I believe, be an error of law that would be holistically detrimental to creators in all industries. The fact that the defendant in this case happens to be directly responsible for evangelizing an extremely broad fair use doctrine, while reaping the financial benefits of widespread online infringement (e.g. on YouTube), is at least an aggravating factor, if not a dispositive one.

Returning to the questionable proposition that “uncopyrightable APIs” necessarily spawn competition and innovation, it is very hard to ignore the background narrative in which mass copyright infringement has been integral to Google’s acquisition of market share in various lines of business, thus producing the mother of all “network effects” such that parent company Alphabet—along with Facebook, Amazon, and Apple—is facing antitrust investigations in multiple countries. Simply put, words like competition are incompatible with Google’s conduct throughout the industry, and its monopolistic presence should at least color how the Court interprets the “standard practice” argument presented in this case.

If the Supreme Court can justly hold, as a matter of law, that the declaring code at issue is uncopyrightable under §102(b), then this is the only basis on which it should arrive at that finding. As for the broader implications for technological innovation, while it is certainly difficult to dismiss an august body of computer scientists, it is equally tough to reconcile the ways in which Android so dramatically belies their premise. Speaking as a consumer who feels pretty damned locked into very limited choices in mobile, I am simply not seeing the benefits of unlicensed reimplementation in this particular example.

*Though Microsoft is a player in mobile, it presently has a very small foothold.