Oracle v. Google Is Over. And Rick Can Stop Blogging About It Forever.
I was in the middle of writing a blog post about The Andy Warhol Foundation for the Visual Arts, Inc. v. Goldsmith—you know, the one about Andy Warhol’s use of a photograph of Prince1The musician, not Richard Prince, the prick-ish “appropriation artist” who might make an appearance later in this post, we’ll see. to make a so-so series of silkscreen prints2”Prints” not “prince.” Also, some pencil sketches were involved. I don’t care. and whether those were a fair use. And I was planning on talking about “transformative use” and how maybe the pendulum was swinging back against an expansive interpretation of “transformativeness.” Then the Supreme Court’s ruling on Google v. Oracle was handed down, and boom, there went my Warhol post.
That’s because transformative use is here to stay, and the pendulum is not “swinging back.”
“You see that, what I’m pointing to? That, my friend, is transformative use. And I will protect it at all costs.” —Optimus Prime in Michael Bay’s Transformers: The Dark Side of Sun, just before something blows up. (Photo by Juanpfotografia via Dreamstime)
The decision is huge, in my opinion. Yes, the copyrighted work at issue is “software” 3For a certain value of “software.”, and the result might’ve been different if the work had been almost anything else, make no mistake: this case was about fair use. It is applicable to every single copyright case.
At the same time, the Supreme Court’s holding isn’t that dramatic—just big. Much of it just confirms what we suspected (and some of us feared) was the law of fair use. Transformative use, for all its flaws, is an important element of fair use, and what’s more, the “expansive” reading of transformative use is the correct one. Many of us were hoping transformative use would be pared back, made more predictable, or even jettisoned altogether. None of that happened.
This decision does, however, provide some important correctives. Fair use really is a four-factor test.4Actually, the Court made clear that it’s AT LEAST a four-factor test. I was explicit that there can be other factors. It is not, as some courts have it, transformative use plus some other stuff. Neither is any one factor always dominant, not even market effects, which at one time was the most important. Copyright is too broad and varied for that.
This corrective comes at a cost. There is no guidance about how to apply the four factors. Which factor is most important and how to weigh the factors will depend on the case. Going into a fair-use argument in litigation, lawyers won’t have a roadmap that tells them which factors the judge will think are most important.
But I think the Court does give us a roadmap. Well, OK, maybe not so much a roadmap as a quest in a sandbox. Tell a story that puts the purposes of copyright front and center, as best you can. Tell the story in terms of the factors. You don’t even need to go in order (the Court here didn’t).
Google v. Oracle answered a lot of questions. Yes, it raised a bunch more, but that’s just the way it works.
For decades, there have been open, fundamental questions about copyright law. Like: What is the purpose of copyright law? What is the role of fair use? What does it mean to “transform” an underlying work? And who decides what’s “fair,” judges or juries?
We have answers to these questions (and more!), like them or not. They are our new reality, and we’re just going to have to get use to them.
For background, I wrote this primer on the case about a year ago, when we all thought it would be decided.5The pandemic delayed thing.
The Economic Theory of Copyright Prevails
The Court begins its analysis with an obligatory quotation of the Constitutional provision empowering Congress to enact copyright laws. Then it hit us with this:
Copyright statutes and case law have made clear that copyright has practical objectives. It grants an author an exclusive right to produce his work (sometimes for a hundred years or more), not as a special reward, but in order to encourage the production of works that others might reproduce more cheaply. At the same time, copyright has negative features. Protection can raise prices to consumers. It can impose special costs, such as the cost of contacting owners to obtain reproduction permission. And the exclusive rights it awards can sometimes stand in the way of others exercising their own creative powers.
OK, fine. The Supreme Court frequently describes copyright as a kind of social bargain: we get cool stuff, and you get paid. Beyond that, Congress is mostly free to implement that bargain how it wishes.
But this is different. The social bargain is only part of it the whole picture, and it’s the positive part. The negative part is the transaction costs of getting permission (as anyone who’s ever tried to “clear” a work can tell you) and potential interference with new works.
You might think this is a jaundiced view of copyright, but it reflects a couple of realities. First, copyright owners have little incentive to play ball. My first reaction to many of fair-use cases is: dang it, get a license! But you can grow old and die waiting for copyright owners to get back to you, assuming you can even figure out who they are.6Ironically, in this case, Google easily could have had a license, for no money, is it were willing to license back its modifications to the code. And Google also had the option of paying for a license that would let it keep its modifications secret.
Second, there used to be another part of the social bargain: public domain. If an artist were looking for works to build upon and, ahem, “transform,” she could just pluck them from the public domain. But the massive extensions of copyright, which were meant to provide continued remuneration for ultra-popular works7If your work is still in demand after your death, you’ve really achieved something. But your quotidian work, like that “Living with Psoriasis” pamphlet in your doctor’s office, doesn’t really need that, and the extensions just gum things up unnecessarily., means that the vast majority of works won’t pass into the public domain until they have lost all cultural relevance and a lot of artistic relevance.
Expanded Transformative Use Is Here to Stay
There are a couple of things going on here. First, believe it or not, there was a question whether transformative use really is something courts should be looking at when analyzing fair use. Yes, I know, every Circuit Court of Appeals seems to have accepted “transformative use” to some extent. But the only Supreme Court decision remotely about transformative use was 1994’s Campbell v. Acuff-Rose Music (the case about a parody of “Oh! Pretty Woman”).
There, the Court spoke of “transformative use” as if it had been established in copyright jurisprudence for decades. Indeed, the Court treated it as is it were the central inquiry. But the case was about a traditional parody, which is not only merely one flavor of “transformative use” but a really strong and unusual flavor. The more general discussion about transformative use was arguably dicta.
Furthermore, in the intervening years, transformative use became unmoored. Transformative use was supposed to help us understand fair use. A lot of us started to wonder if we merely substituted one wibbly-wobbly concept for another—and whether we should ditch the concept before it becomes a monster.
Well, I have good news and bad news. The bad news is: we’re stuck with transformative use. The good news is, the Court did not treat it as some kind of overarching central inquiry. Indeed, the Court almost seems embarrassed by the term.
In the context of fair use, we have considered whether the copier’s use “adds something new, with a further purpose or different character, altering” the copyrighted work “with new expression, meaning or message.” Id., at 579. Commentators have put the matter more broadly, asking whether the copier’s use “fulfill[s] the objective of copyright law to stimulate creativity for public illumination.” Leval 1111. In answering this question, we have used the word “transformative” to describe a copying use that adds some- thing new and important. Campbell, 510 U. S., at 579.
But make no mistake, transformative use remains a central inquiry, at least for the first factor, the “purpose and character of the use.”
However, the Court accepts the expanded version of “transformative use.” You might think of the distinction I’m trying to draw here as “internal” and “external.” The song at issue in Campbell was a parody that (to some extent) commented back on the original work. That’s internal because you can see the transformative features just by examining the two works.
“External” transformative uses take that bit about “further purpose” and runs with it. There, the new work places the old work into a new context, or puts it to work for a new project. There, the main inquiry is the amount of social benefit derived from the new use. Thus, the “Google Books” project was found (by a lower court, albeit a very influential one) to be a transformative (and ultimately a fair) use because of the massive (if diffuse) social benefit.
Well, the Supreme Court isn’t going to put that particular genie back into its bottle:
To repeat, Google, through Android, provided a new collection of tasks operating in a distinct and different computing environment.
The record here demonstrates the numerous ways in which reimplementing an interface can further the development of computer programs.
What’s more, the Court rejected a purely “internal” version of transformative use. Noting that Google’s new use was both an exact copy and for almost the exact same purpose, the Court said that you can’t stop there. “Rather, in determining whether a use is ‘transformative,’ we must go further and examine the copying’s more specifically described ‘purpose[s]’ and ‘character’.” (Emphasis added.)
More than any other part of the Court’s ruling, this holding is the most bitter to those copyright holders who went to bat for an evil Silicon Valley technology company8What? You didn’t know Oracle is also an evil Silicon Valley company? in case well outside their industries. They wanted transformative use contained, if not killed outright. They got the opposite.
Fair Use Will Always Be Impossible to Manage
If there’s a takeaway from Google v. Oracle that I fully support, it’s that courts have been too rote in the way they apply the fair-use factors. You know what I mean. The court just goes through the factors in order, adds them up (yet somehow they always end up 4-0 one way or the other), and rules based on that score (with the first and fourth factors as tie-breakers).
No, says the Court here. Fair use isn’t just highly sensitive to the facts, but the factual situation will dictate how you go about the analysis. Some factors might dominate in some situations, others in other situations. This is especially important in this case because we’re applying factors that were laid down before there were even programmable computers.
After setting forth the four traditional fair-use factors, the Court cautions:
In applying this provision, we, like other courts, have un- derstood that the provision’s list of factors is not exhaustive (note the words “include” and “including”), that the exam- ples it sets forth do not exclude other examples (note the words “such as”), and that some factors may prove more important in some contexts than in others. … see also Leval, Toward a Fair Use Standard, 103 Harv. L. Rev 1105, 1110 (1990) (Leval) (“The factors do not represent a score card that promises victory to the winner of the majority”). In a word, we have understood the provision to set forth general principles, the application of which requires judicial balancing, depending upon relevant circumstances, including “significant changes in technology.”9Certain citations have been omitted.
I’ve argued something similar to this in previous blog posts about this case. Since computer programs are nearly impossible to “transform” (I was hoping for a primarily “internal” view of transformative use10More specifically, I thought internal transformative should be given more weight than external.), there needs to be another way to apply fair use to software. I concluded that the second (nature of underlying work) and third (amount and substantiality taken) factors should take precedence. I thought it was very important that the organization of an API library11I still think this is a more accurate way to describe the underlying work at issue than “declaring code,” which does nothing more than reflect the organization. was at the very edge of what copyright can protect.
The Court didn’t exactly do that here, but it came close. It does ascribe much more importance to those factors than courts normally do. Indeed, the Court started with the second factor (nature of underlying work), a factor so deprecated that it was falling out of fair-use analysis entirely.12What courts actually say is that the second factor can’t be “dispositive,” i.e., solely determinative of the fair-use issue. Logically, that’s 100% true, or else copyrights at the edge of protectability would be functionally unenforceable. But that doesn’t mean it isn’t important.
The upshot is that fair use is going to be even more unpredictable than before. Well, assuming that’s even possible, since fair-use decisions are notoriously “results-oriented.”13I.e., the judge decides intuitively who’s right, then works backward from there. Not only must lawyers tick all of the boxes, but they must put the factors together into a coherent narrative. For litigation, this is welcome and exciting. But for counseling, this is a nightmare because the analysis can shift under your feet depending on how you prioritize the factors.
But fair use was never going to behave. Copyright covers everything from paintings and sculptures to software and industrial designs, from architecture to coin-collector’s catalogues, from the great American novel to the last email (of any significant length) you just sent. In deciding what to protect, it makes no distinction between high art, low art, public announcements, private musings, beautiful or ugly, profound or superficial, timeless or quotidian.
Yet, fair use must apply to all of them. Trying to come up with a set of consistent rules was alway a fool’s errand. The Court just kind of recognizes that.
A final thought: Just because courts can re-prioritize the factors doesn’t mean it can just ignore one or more of the them. It can conclude that one factor just isn’t important, but it needs to say why. No more writing factors off just because “they’re not dispositive.”
The Fourth Fair-Use Factor Will Be Forever Impossible to Apply
Does anyone understand, in a principled way, the Court’s reasoning for the fourth factor (effect on actual and potential markets for underlying work)? I think the most generous interpretation is this: the (advisory) jury obviously found this factor to weigh in favor of fair use, and there’s enough evidence in the record to support that finding, so we’re just going to go with the jury on this one. (More on the jury’s role below.)
That interpretation makes sense because the fourth factor is the most fact-intensive, and finding facts is what juries are for. Oracle put on evidence about Sun’s (the original developer of Java) attempts to license Java, and Google put on evidence that Sun was never going to succeed in that endeavor. In my opinion that’s a slight edge against fair use, but Oracle chose to let a jury sit for this issue and it must suffer the consequences.
The Court’s own thoughts about the fourth factor are mostly about how other factors mitigate Oracle’s evidence of market factors. I would prefer the Court save this for a conclusion where all factors are considered together (and just admit that Oracle’s evidence was actually pretty strong). The point of having factors is to make sure courts focus on a set of facts. I’m concerned that, if you don’t keep the factors somewhat siloed, you’ll let one factor start to dominate the rest, the way “transformative use” was taking over all fair use factors not too long ago.
The fourth factor has always been a hot mess. This decision isn’t going make it any less hot or any less messy, alas. It’s the most important factor precisely because it goes to the economic theory behind copyright: the incentive to profit from your creations.
For the same reason, it’s almost impossible to apply. If we focus on developed markets for the underlying work, we punish artists who just aren’t that good at business. But if we expand our analysis to potential markets, then we almost unavoidably fall into circularity (which the Court, to its credit, warns against). One can almost always imagine some hypothetical market that the copyright holder just hasn’t thought to exploit.
Fair Use Isn’t a Jury Question (and That Strengthens It)
Oddly, given how old fair use is, we’ve never really known if the ultimate question of fair use is for the judge or the jury. This is an important question for three reasons. First, juries have their biases and rooting interests that (we hope) judges don’t have. Second, jury verdicts are nearly impossible to overturn on appeal. Third, questions for the judge can be resolved early in the case because we don’t need to wait for trial to get to the jury.
The Federal Circuit, to its credit, squarely confronted this question. It held that fair use is a question for the judge. I explain the whole thing in detail here. It also dealt with the awkward fact that the fair-use issue in this case was given to a jury, and the jury found for Google. It decided the jury was just an “advisory jury,” whose findings the judge can accept or reject. It then assumed the judge accepted all of the jury’s implicit findings of “historical facts” (i.e., facts that go to what actually happened). It purported to give full credit to those jury findings (then, part way through its fair-use analysis, seemed to forget that).
The Supreme Court endorsed the Federal Circuit’s reasoning. The ultimate question of fair use is for the judge. The judge, if he or she wishes, can give certain factual questions to an advisory jury, but that’s optional.
Every fair use case is different, but overall, I think this ruling strengthens fair use. Judges are going to be much more likely to rule earlier in the case about fair use, which benefits defendants (i.e., accused infringers) more than plaintiffs. Indeed, the trend of ruling on fair use at the very beginning of the case (on a “rule 12(b)(6) motion”) can continue.
This, in turn, will encourage fair use of copyrighted works—and the pushing of the boundaries of fair use. The main issue can be resolved early in the case, saving on the expenditure of legal fees. Furthermore, copyright holders can’t take advantage of juries’ natural inclination toward plaintiffs, or any “home court advantage” juries often bring for plaintiffs.14Some folks have drawn the opposite conclusion that I’ve drawn. They read the decision as requiring use of a jury for “historical facts.” The Court’s description of the jury’s role is infelicitous, but I’m fairly certain that the jury is optional and advisory. The key is that the Court adopted the Federal Circuit’s reasoning, and the Federal Circuit was explicit that juries are optional and advisories, and the only reason we even had one in this case is that everyone involved made a mistake.
Copyright Will Always Be a Bad Fit for Software
We’re still in the “hot take” phase of reacting to Google v. Oracle, and the two most prominent hot takes in my Twitter feed may be summarized as follows:
- Fair use is nice, but APIs are too functional to be entitled to any protection. That would’ve been a cleaner decision.
- What a terrible decision. Can you even protect software at all after this decision? (But at least the decision is limited to software.)
Although these viewpoints are nearly polar opposites, they are both frustrated with the way copyright law protects software. But copyright doesn’t do for software what you think it does.
I explain why copyright and software are a bad fit, in detail, here. But all problems arise from one source: software is expression that’s primarily functional. Copyright law doesn’t protect functionality, only expression. You need patent or trade secret law for that. At the same time, software isn’t completely devoid of expression, though we really don’t know what is expressive about software. Usually, we just equate expressiveness with “choice,” but that hardly seems adequate.
To make matters more complicated, where there are only a few reasonable ways to express something functional, copyright law won’t protect the expression. That would be giving copyright-length and -breadth protection to something useful. Oddly, this means good (“elegant”) code is harder to protect than bad code. There are lots of ways to write bad code, but just a few ways to write good code, and we’re not giving you a monopoly over one of the few good ways.
Under basic, plain-vanilla copyright law, computer programs are potentially copyrightable.15For strategic reasons, Google didn’t put on very much evidence about the protectability of the organization of the Java API library, so when the issue became front and center on appeal, it didn’t have much to fall back on in the record. Also under basic, plain-vanilla copyright law, computer programs are on the edge of what’s copyrightable and thus vulnerable to fair use. That’s exactly where the Supreme Court came down in Google v. Oracle.
Congress decided to make computer programs protectable by copyright in 197816Effective 1980. Arguably, this changed nothing because, arguably, computer programs were already protectable by copyright.. As far as I can tell, Congress never gave protecting software another thought. It’s as though Congress thought it solved the problem forever, even though few industries have changed more since 1978 than software development.
Software needs its own protection regime. Until then, everyone is going to be unhappy with how it’s protected.
A Final Thought
Fair use has expanded because copyright has expanded, and because Congress won’t do its job. Copyright was once concerned solely with literal or near-literal copying, but “substantial similarity” came along and expanded what a copyright can protect—the the point where we get decisions like “Blurred Lines.”17And think about it: we’ve only recently have had any coherence brought to the law of substantial similarity. And once, copyright lasted just a decade or two. Now it lasts (typically) for over a 100 years. And once, copyright protected only a certain core set of works, like books and maps. Now it protects photographs, architectural works—and computer programs.
What the Supreme Court make clear in Google v. Oracle is that fair use is fundamental to copyright law as a whole. It is not an exception but an equal partner, not an intrusion into copyright but the other side of it.
Thanks for reading!
Footnotes
| ↑1 | The musician, not Richard Prince, the prick-ish “appropriation artist” who might make an appearance later in this post, we’ll see. |
|---|---|
| ↑2 | ”Prints” not “prince.” Also, some pencil sketches were involved. I don’t care. |
| ↑3 | For a certain value of “software.” |
| ↑4 | Actually, the Court made clear that it’s AT LEAST a four-factor test. I was explicit that there can be other factors. |
| ↑5 | The pandemic delayed thing. |
| ↑6 | Ironically, in this case, Google easily could have had a license, for no money, is it were willing to license back its modifications to the code. And Google also had the option of paying for a license that would let it keep its modifications secret. |
| ↑7 | If your work is still in demand after your death, you’ve really achieved something. But your quotidian work, like that “Living with Psoriasis” pamphlet in your doctor’s office, doesn’t really need that, and the extensions just gum things up unnecessarily. |
| ↑8 | What? You didn’t know Oracle is also an evil Silicon Valley company? |
| ↑9 | Certain citations have been omitted. |
| ↑10 | More specifically, I thought internal transformative should be given more weight than external. |
| ↑11 | I still think this is a more accurate way to describe the underlying work at issue than “declaring code,” which does nothing more than reflect the organization. |
| ↑12 | What courts actually say is that the second factor can’t be “dispositive,” i.e., solely determinative of the fair-use issue. Logically, that’s 100% true, or else copyrights at the edge of protectability would be functionally unenforceable. But that doesn’t mean it isn’t important. |
| ↑13 | I.e., the judge decides intuitively who’s right, then works backward from there. |
| ↑14 | Some folks have drawn the opposite conclusion that I’ve drawn. They read the decision as requiring use of a jury for “historical facts.” The Court’s description of the jury’s role is infelicitous, but I’m fairly certain that the jury is optional and advisory. The key is that the Court adopted the Federal Circuit’s reasoning, and the Federal Circuit was explicit that juries are optional and advisories, and the only reason we even had one in this case is that everyone involved made a mistake. |
| ↑15 | For strategic reasons, Google didn’t put on very much evidence about the protectability of the organization of the Java API library, so when the issue became front and center on appeal, it didn’t have much to fall back on in the record. |
| ↑16 | Effective 1980. Arguably, this changed nothing because, arguably, computer programs were already protectable by copyright. |
| ↑17 | And think about it: we’ve only recently have had any coherence brought to the law of substantial similarity. |
