Welcome, new readers, to the SPN website! If you’re interested in software preservation and the legal, technical, and social work required to Save Software Together, please take a look at how you can Get Involved and consider joining SPN today! Learn more about the 2021 Hosted Emulation Services Pilot here.
Today the Supreme Court heard oral arguments (video and transcript) in the copyright dispute between Oracle and Google. The case could have huge implications because it is both a) the first case involving the fair use right that the court has heard in 25 years, and b) the first case involving software copyright that the court has heard ever. Sarah Jeong at the Verge has a great rundown of the history of the case, which has now been raging for a decade.
The basic story, to the extent that a story this epic and technical can be reduced to something basic, is that Google copied aspects of the Java code, copyright to which Oracle acquired when it purchased Sunn Microsystems, and incorporated these elements in its Android mobile phone operating system. Google has acknowledged that it used portions of the ‘declaring code’ from Java as part of Android because software developers are familiar with Java and would be better able to write new programs for an operating system that responds to Java commands. Oracle claims this copying infringed its copyrights, and wants up to $9 billion in damages. Google argues that either a) copyright doesn’t protect the code it copied, or b) fair use permits Google to reuse the code without permission.
This recap/explainer got a lot longer than I expected, so here is a TL;DR version: It seemed to me that the court was uncomfortable with this case due to its complexity and the high stakes for the software industry. They wrestled with the notion that copyright clearly protects software, and yet at least some software (or some aspects of software, or some uses of software) seem like a bad fit for copyright, and should be free for others to reuse in the way Google did in Android. There were relatively fewer questions about the substance of fair use than about the other issues in the case, but there were lots of questions about the possibility that the jury’s fair use finding should have received more deference. If I had to guess, I might predict that the court will seek shelter from the tough questions by sending the case back to the lower court with instructions to reconsider it and give the jury’s finding more deference. That outcome will give us the least clarity (as it avoids announcing clear binding rulings on either fair use or the scope of copyright in software), but it also has the least likelihood of causing disruption.
In oral argument this morning, we heard from three parties: Google, represented by Tom Goldstein; Oracle, represented by Joshua Rosenkranz; and the U.S. Solicitor General’s office, represented by Malcolm Stewart. Stewart was there because the Supreme Court had asked the Solicitor General (part of the U.S. Department of Justice) to weigh in on the case, and the SG’s brief sided with Oracle. The Court occasionally seeks the SG’s input as a neutral expert in complex cases, but I wouldn’t necessarily assume the Court will give special weight to the SG’s position. The Court has asked for the SG’s opinion in other IP cases, and it hasn’t consistently followed the SG’s recommendations.
Here are a few things I think folks in the software preservation community might be interested to know about the oral arguments, and what might drive the Court’s decision now that the arguments are over:
The Court is anxious not to upset settled expectations, especially for industry.
The amicus briefs (including ours, as well as briefs from Apple, Microsoft, and a raft of computer scientists and related groups) were chockfull of warnings about the potential harm the court could do if it went the wrong way with this case. Briefs that sided with Google warned the court that building new products using code from existing ones is a best practice in software development, and that a win for Oracle would put a deathly chill on the software industry. Chief Justice Roberts, and Justices Alito, Kavanaugh, and Sotomayor all asked Oracle and the SG variations of the question, “Can you explain how, if we rule for Oracle, we won’t be gutting the US software industry?” (Justice Breyer also expressed sympathy with the idea that Oracle would have an unfair advantage if it won, but seemed to be focused more on fairness than on the impact on industry, per se.) On the other hand, briefs siding with Oracle warned that a win for Google would nullify copyright protection for valuable software assets, devaluing those assets and reducing incentives for future creation. Justices Gorsuch, Sotomayor, Kavanaugh, and Alito asked Google to respond to this concern.
In areas where justices feel confident they understand the law and the underlying policy concerns, they’re willing to stand on principle and do things that might upset the status quo. But in technical areas like copyright and patent, where there aren’t clear partisan or other ideological battle lines and most justices are learning the law on-the-fly, you get the sense that the justices may be less confident in their own views and more deferential to industries and other stakeholders who rely on the law every day.
It’s all about the analogies.
This Court, like most courts, is comprised of legal experts, not technology experts, and they relied heavily on analogies to pump their intuitions about the rarefied world of software code. Perhaps the most on-the-nose moment was the very first question, in which Justice Roberts proposed an analogy between Java’s structural elements and the headings and structure of a legal brief—the only creative work that every justice and even every clerk is likely to have experience making and using. Goldstein’s reply—roughly, that writing code that interoperates with Java is pretty different from writing legal briefs—didn’t seem to persuade Roberts.
That’s OK, though, because Roberts had another analogy for Oracle’s attorney, Rosenkranz: Isn’t Java just like the structure of a restaurant menu—every new restaurant needs to make its menu follow the same structure (appetizers first, then mains, then dessert) because that’s what customers (developers) expect to see? Rosenkranz’s response—that Java is a lot more complex than a menu, and there are other ways to structure software code—seemed to land more comfortably with Roberts.
Justice Breyer gets credit for perhaps the most compelling analogy of the morning, though (IMO): the QWERTY keyboard. Rosenkranz had argued repeatedly that Java’s system of declaring code is just one way to structure a programming interface, and that Google could have created or licensed its own alternative set of commands for programmers to use. In response, Breyer raised the analogy of the QWERTY keyboard, which (like Java’s declaring code) translates the user’s commands (pressing a particular key) into outputs by the machine (the corresponding letter appears on the screen). Sure, there are hundreds of alternative ways to lay out a keyboard, but once typists have learned QWERTY, isn’t everyone bound to interoperate with it or else be at an enormous disadvantage? And wouldn’t the owner of a monopoly on QWERTY therefore (unfairly) have a huge advantage by controlling such a popular interface system? And (perhaps most importantly) isn’t this why copyright doesn’t protect any “procedure, process, system, [or] method of operation” (17 USC 102(b)), because a monopoly over such functional, useful elements should require the applicant to clear the (much higher) burden of obtaining a patent?
Rosenkranz responded that Java code is unlike QWERTY because it is “expressive,” a response that didn’t seem to land. (When was the last time you curled up by the fireplace to enjoy a good book of software declaring code?) Other justices picked up on the QWERTY analogy, and it was restated and reiterated several times during the argument.
A handful of other analogies for the Java code proposed by justices or attorneys during the argument:
- The combination to a safe
- the playbook for a winning sports team
- a system of organizing a grocery store
- a prison for developers
- a system of accounting
- the French and English languages
Fair use could be the key, but maybe not in the way we expect
The justices seemed uneasy about the prospect of ruling that software code is, or certain classes of software code are, categorically beyond the realm of copyright protection under Section 102 and/or a related principle called the merger doctrine. Fair use is a more flexible way of allowing reuse without permission in the right circumstances. Indeed, the jury found that Google’s use of Java was a fair use under the flexible four-factor fair use test. The Court has two options for bringing fair use into the case. One option is to find that fair use does protect uses like Google’s. Another is to say that the federal circuit did not give enough deference to the jury’s fair use verdict.
The first option could be the most interesting, but it raises very similar practical concerns to the ones about software copyright-ability: a sufficiently clear fair use holding would have the effect of categorically permitting (or barring) substantial amounts of unlicensed copying of software, and the court seemed uncomfortable with going there.
Nevertheless, questioning probed a few key fair use issues, including transformative use. Oracle’s attorney argued that Google’s use was not “transformative,” a term of art that means use of an existing copyrighted work for a new purpose, a type of use that the Supreme Court said was “at the heart” of fair use in its last opinion on the issue, in 1994. Some justices seemed to agree, but others, like Justice Sotomayor, suggested that the use could be transformative because Java was intended for computers and Android runs on a mobile phone. Justice Kagan made the friendly suggestion that a use can still be fair without being transformative, and that Google’s purpose might still be favored by fair use. Google’s attorney agreed, adding that the word “transformative” doesn’t appear in the statute, and that copyright protection for this code should be “thin” because of its minimal creativity.
Judge Thomas noted that the four fair use factors in the statute are not exhaustive—courts can consider other factors, too—and asked the attorney for the US Solicitor General whether there are other factors to consider here. Given the narrow, almost robotic adherence to the four factors in lower courts, this was an invitation to innovate, but the SG offered only a restatement of the fourth factor (market effect), saying the court should give special consideration to how a fair use finding could harm the incentives for creation of new works. I wouldn’t expect any new factors to be added to the fair use calculus by this opinion.
The fair use issue that the justices seemed most keen to probe, and that they even sought additional briefing on ahead of the oral arguments, was how courts of appeal should review jury verdicts on fair use. That is, rather than reaching its own opinion on how fair use applies, should the Supreme Court (and the Federal Circuit before it) show deference to the jury’s opinion that Google’s use was fair?
Generally, for reasons rooted in the Constitution (the 7th Amendment right to a jury trial) and in good policy (a jury hears all the evidence in a carefully structured trial proceeding, while appellate courts typically have only written records, briefs, and oral arguments), courts of appeal are required to be deferential to a jury’s findings of fact. Courts of appeal can (and should!) act on their own judgment to correct errors of law—the trial court misinterpreting a statute, giving the jury erroneous instructions, and so on. But typically courts are supposed to leave a jury’s findings on questions of fact (was the defendant in town on the night of the murder, did the plaintiff lose customers as a result of the defendant’s conduct, etc.) in place unless there is very clear evidence of error.
Fair use is typically considered a mixed question of fact and law because to find fair use, the jury has to apply legal principles to factual findings. Reviewing a jury’s decision on fair use can be tricky, then, because the level of deference is mixed depending on which part of the jury’s reasoning the court is reviewing. In this case, the issue is even tougher because the jury returned a general verdict—it simply said, essentially, “Google’s use is fair use”—without issuing any more detailed findings. Appellate courts face difficulty in reviewing a general verdict because they can’t untangle the factual and legal aspects of the mixed question.
The jury heard testimony on factual questions like whether Android was a competitor to Java and whether Oracle lost profits due to Android’s use of portions of the Java code. Presumably, its findings on those issues should be left in place. But did the jury properly understand legal concepts like transformative use, and did they apply the law properly to the facts? The Federal Circuit said no, and it overturned the jury’s verdict saying that no reasonable jury could have found Google’s use to be fair. Google argued that this was improper, and that if the Federal Circuit had shown proper deference to the jury, it would have had to uphold the jury’s decision.
Direct questions on this issue came from Justices Thomas, Alito, and Gorsuch, but Google’s attorney, Tom Goldstein, raised it several times in response to questions from other justices. Perhaps the most compelling moment in the entire argument was at the very end, when Goldstein explained that Oracle’s attorney had made arguments that very day that relied on factual conclusions (was there a market for declaring code? did Android compete with Java?) the attorney simply assumed, whereas the jury had evaluated these factual claims based on evidence and expert testimony, and reached a different conclusion. For the Federal Circuit to reverse the jury’s verdict based on its own reading of the facts, and the Court to affirm based on arguments like Oracle’s, certainly does seem to undermine the jury’s role and authority as the primary judge of facts in the case.
A perhaps-ill-advised prediction
Predicting Supreme Court decisions is a dangerous game that smart lawyers avoid at all costs. But some of us can’t resist! If I had to guess at an outcome based on what I heard today, I think the court may be sorely tempted to take the easy way out and rule on the jury deference issue. That would avoid making rulings on either copyright subject matter or fair use that could upset the software industry, and wading into a complex and technical area of the law that at least some of the justices seemed wary of mucking up.
A notorious postscript
One additional thing is worth noting as we think about the Court’s consideration of this case: Justice Ruth Bader Ginsburg’s absence is likely to have a significant impact. Justice Ginsburg was a reliable vote in favor of copyright holders and the protection or expansion of their rights, and would almost certainly have sided with Oracle. For all her other virtues, the Justice was not a friend of fair use or limited copyright, and her absence makes it slightly more likely that Google may carry the day.