The high profile IP dispute over Google’s reimplementation of the Oracle-owned Java API in Android software is headed back to a federal appeals court after a jury handed a unanimous loss to Oracle last summer. So what’s the issue this time?
Our Story Until Now
A quick refresher: the case began following Oracle’s 2010 acquisition of Sun, which held the copyrights to Java. Oracle sued Google on various IP claims later that year. The basis for the suit was that in developing Android, Google had created its own version of the programming language called Java, a toolset that programmers use to write code. In order to enable other developers to program on Android, Google employed the same names, organization, and functionality in Android as is used in the Java API (“application programming interface”) — the protocols by which software programs communicate.
In June 2012, following a jury trial, Judge William Alsup in the Northern District of California ruled that the Java API “packages” copied by Google were not copyrightable. According to Alsup’s holding, Android may have “replicated the overall name organization and functionality of 37 [out of 166] packages in the Java API.” However, these protocols — themselves only 3% of the 37 packages — did not qualify as copyrightable since to extend that protection would provide control over “all possible implementations of the taxonomy-like command structure” used by programmers to write code. (Fun fact: Judge Alsup taught himself Java for purposes of hearing the case.).
Oracle managed to overturn this initial loss by convincing the U.S. Court of Appeals for the Federal Circuit in 2014 that the “structure, sequence, and organization” of that 3% of the 37 packages was sufficiently original to merit copyright protection. (Normally, cases in Northern California are appealed to the U.S. Court of Appeals for the Ninth Circuit, but this case was appealed to the Federal Circuit, in Washington, DC, which some argue has a pro-IP bias. This was because the Federal Circuit has near-exclusive jurisdiction over patent appeals, and Oracle had brought patent claims in its initial complaint.)
As DisCo contributor Jon Band argued, the Federal Circuit’s decision for Oracle appeared to conflict with other prominent appellate decisions on software copyright, such as Sega v. Accolade and Lotus v. Borland. This set up the possibility that the Supreme Court might intervene to resolve the difference — and in fact the Court briefly showed interest in the case before passing on it in early 2015.
The first stage of the case being resolved in Oracle’s favor, and with the Federal Circuit having determined that the code was copyrightable, the dispute was returned to Judge Alsup’s court to develop a full record for a trial on defenses to infringement, including Google’s fair use argument. That went badly for Oracle. After a lengthy second trial in May 2016, the jury returned a unanimous verdict for Google. Oracle urged the court to throw out the verdict, a proposal that Alsup “dismantled” in a strong order. Having lost its effort to overturn the jury’s verdict, Oracle has appealed again to the Federal Circuit, which bailed the company out once before.
Oracle v. Google on Appeal: The Sequel
On its second visit to the Federal Circuit, Oracle faces an uphill battle. To overturn a jury verdict, it must show that no reasonable jury could have concluded what the jury concluded here: that Google’s copying was fair use under Section 107 of the U.S. Copyright Act.
Oracle’s case turns on two general arguments: first, that the court improperly excluded some evidence; and second, that based on the existing evidence, a reasonable jury could only have decided in favor of Oracle. Various amici filed in support of Oracle last week, with one filing in support of neither party.
Oracle’s evidentiary argument claims that the district court unreasonably limited Oracle’s case to two manageable markets for Android: smartphone and tablets, rather than an open-ended trial on any product in which Android might be put to use. Oracle also argued that Google hid evidence pertaining to Android running outside of these markets, specifically, “App Runtime for Chrome” (ARC). However, as Sarah Jeong reported for Motherboard after last summer’s trial, “actually, one of Oracle’s expert witnesses devoted seventeen paragraphs to ARC in his report. Oracle knew all about these developments, and just decided not to bring it up at trial.”
Oracle’s argument against the jury verdict re-litigates the well-known four-factor fair use test. Several of Oracle’s amici do the same.
For reference, the fair use factors are (summarized):
- the purpose and character of the use;
- the nature of the copyrighted work;
- the amount and substantiality of the portion used; and
- the effect of the use upon the potential market.
As in many fair use cases, Factors 1 and 4 prove to be the most contentious. In particular, many briefs focus on whether the use was “transformative”, and second, on which market was affected by the use.
Transformativeness: Since the Supreme Court’s 1994 decision in Campbell v. Acuff-Rose, many courts evaluate the first fair use factor in part by asking whether the use was “transformative.” As I’ve noted here before, transformativeness is a significant, although not essential aspect of the fair use inquiry. Oracle’s brief offers a fairly conventional argument of the first factor. Not so for its amici; several take the Campbell decision on directly. They strongly suggest that the Supreme Court’s Campbell decision screwed up copyright law. The implication seems to be that the Federal Circuit should just do away with the precedent. (That’s not how it usually works.) Even assuming the Federal Circuit has a pro-IP bent, it is doubtful that it would openly challenge a unanimous decision by the Supreme Court. This seems particularly unlikely when, as the district court observed, the question is ‘could no reasonable jury have found it fair use for Android to have re-implemented with new code 37 out of 166 Java API packages, in a new context.’
Effect on the market: The fourth fair use factor explores the effect of the use at issue on the potential market for the work. One could argue that any commercial use of a work — even a highly transformative use — could be construed as an independent market for that work, and therefore any use could arguably have a negative effect on the market for that use. For this reason, courts have traditionally weighed commercialism less when a use is highly transformative.
One amicus brief, by amici identifying themselves as IP scholars, goes so far as to devote an entire section of the brief to saying, “Fair Use Does Not Excuse Commercial Infringing Uses.” This is a head-scratcher, since most IP scholars have probably taught cases in class where fair use does exactly that. (Examples are linked to in this post, and this one). The Supreme Court’s Campbell decision itself excused a commercial infringing use, reversing an appeals court which had held the commercial use of Roy Orbison’s “Pretty Woman” couldn’t be fair.
Oracle’s brief is also notable in that it struggles mightily to recast the question of infringement as one of “plagiarism” of an artistic, literary work. As I’ve written elsewhere, most IP lawyers know that plagiarism and copyright infringement are not the same thing, although it isn’t hard to find courts casually using the terms interchangeably. Plagiarism is averted through attribution, however, not a license. And there’s nothing infringing about a licensed use of a work that doesn’t credit the initial creator. No serious observers think this case is about Sun getting credit for having developed Java, however. This is about licensing royalties, and thus, allegations of infringement, not misattribution.
Yet Oracle likely uses this pejorative deliberately. The word “plagiarism” tends to convey an element of intent, or willfulness, which conveniently muddles the separate questions of liability and remedy. In fact, Oracle is no stranger to the strategy of trying to infuse subjective values into the otherwise objective legal question of infringement: In 2014, SAP succeeded in knocking roughly 75% off an infringement judgment won by Oracle in part because Oracle had insisted on using words like “theft” and “stealing” at trial, notwithstanding a court order that doing so would be unnecessarily inflammatory and prejudicial.
After waging a Seven Years’ War over Android, it isn’t surprising that Oracle appealed this loss. The Federal Circuit has been friendly before, and so in for a dime, in for a dollar. But some aspects of the argument appear to be the equivalent of a “Hail Mary” pass in the closing seconds of the game.
These are highly fact-specific questions. They have to be, or the Federal Circuit would not have had to remand this case in the first place. If this dispute were simply a matter of “black-letter law,” the case might have been resolved in 2014. Having sent the dispute back for additional factual investigation before a jury, however, it seems unlikely that the appeals court would now set aside the jury’s conclusion and substitute its own.
In addition, the fair use doctrine has considerable strength, which has been reaffirmed in recent years through numerous cases — many covered here at DisCo. Gambling on an appeal that would narrow fair use in a context that requires distinguishing a number of other Circuit decisions may be a long shot, and even a victory sets up a possible trip to the Supreme Court.
Google’s brief and amicus briefs in support of Google will be filed later in the spring, and the Federal Circuit will hear oral argument later this year.