As noted in Part I, there are a lot of moving parts to this story that cannot be addressed in a single post; but the one thread readers should not lose is the fact that this whole dust up started because Google was the first commercial user since the launch of Java in 1995 to refuse a license agreement. Undeterred by that legal requirement, Google proceeded use Java code to develop one of the biggest commercial ventures in computing history; and if anyone out there believes they declined to enter into that license agreement in order to foster greater competition and innovation in software, then you have not been paying attention to Google’s globally anti-competitive conduct in the market to date. But in this post, I want to begin to address some of the legal arguments that have been presented in Google v. Oracle.
In the most basic terms, several of Google’s supporting amici—and therefore the story that seeps into the blogosphere—asserts that the “declaring code” Google copied in the process of developing Android is not properly a subject of copyright (i.e. cannot be protected). And even if this code is protected, Google claims that its use constituted a fair use. I shall set the fair use defense aside for a future post and address the challenge to the copyrightability in Oracle’s computer code, which turns on a single argument that what has been called the “declaring code” amounts to a system or function rather than an authored form of expression.
I personally see two significant and intertwined challenges with regard to following this case. First, because most of us have little or no foundation for recognizing the amount of “creativity” in computer code, it is a category of works that is highly dependent upon analogy and expert testimony; and second, because Google and most of its supporting amici are known to be anti-copyright ideologues, it is reasonable to ask whether the central argument being made is well-grounded in facts about the code at issue, rather than opinions that (once again) seek to undermine the nature of copyright itself.
Certainly, we have seen many arguments try and fail to holistically weaken copyright in other cases; and one hint that this game may be afoot in Oracle is the amount of logical contortionism demonstrated by certain amici on behalf of Google. And I really do mean contortionism because, for example, EFF and PublicKnowledge argued in 2017 that the question of copyrightability in the declaring code was “settled” by a completely different court finding that a system of yoga poses was not copyrightable. If that sounds like comparing bindis and bytes to the reader who is unfamiliar with copyright law, let me suggest that it may also sound that way to those familiar with copyright doctrine in this particular case. But now, I have to back up a bit.
One-hundred years before software was added to the statutory protection of copyright, the Supreme Court articulated a seminal distinction in Baker v. Selden (1879), holding that although Charles Selden’s book describing his new method of bookkeeping was of course copyrightable, the bookkeeping system itself, no matter how novel, was not a subject of copyright. In addition to its foundational role in defining the idea/expression dichotomy,* Baker also gives us the principle of the merger doctrine, which holds that when there is only one way (or just a few ways) to express something, the expression is not protected because it is said to be “merged” with the idea. (Google’s defense hinges substantially on merger. More on that in the next post.)
Jump to 2015, and Yogi Master Choudhury Bikram asserts in litigation with an American yoga studio that his system to promote health by means of twenty-six yoga poses and two breathing exercises, as described in his 1979 book, is protected by copyright. Once again, the book is clearly a subject of copyright, but Bikram’s selection and arrangement of poses and exercises to bring about a particular result are, like Selden’s system of bookkeeping, not protected by copyright. Consequently, the 2017 amicus brief filed jointly by EFF and PublicKnowledge contended that Bikram “settles” the question of copyright in the Java code, citing the Ninth Circuit opinion that, “the possibility of attaining a particular end through multiple different methods does not render the uncopyrightable a proper subject of copyright.”
That statement is true and a perfectly sound response to Bikram’s argument that he could have made different selections among poses and exercises to achieve a desired result; but even if Bikram had not asserted that particular claim of “creativity,” the doctrines established in Baker in 1879 would have been sufficient to deny copyrightability to a system of yoga poses. So, not only does Bikram fail to “settle” the question of copyrightability in Oracle’s declaring code, but asserting that it did reads as something of a detour around the only question that matters—whether there is authorship in the code, distinct from the function it performs when executed. And unlike yoga poses, computer code is expressly protected as a “literary work” under U.S. copyright law.
I will dig into more detail about the nature of the code in question in the next post about this case, along with a look at arguments presented in the briefs filed with SCOTUS in support of Google this week. But the reason I decided to highlight the Bikram detour in this narrative is that the defenses for Google seem to contain a lot of hyper-extended arguments—legal chakrasana, if you will—that begs the question as to how strong the “non-subject of copyright” argument really is. For instance, the EFF/PK brief sought to identify an apparent doctrinal split between the Ninth Circuit holding in Bikram and the Federal Circuit holding in Oracle that its code is copyrightable …
“A finding of copyrightability [by the Federal Circuit] based on the fact that ‘the author had multiple ways to express the underlying idea,’ cannot be reconciled with the Ninth Circuit’s holding that ‘the possibility of attaining a particular end through multiple different methods does not render the uncopyrightable a proper subject of copyright.’” (citations omitted).
Again, in the context of a system of yoga poses, it is unremarkable to say the existence of other yoga poses does not make the system any more copyrightable. EFF/PK tries to turn that holding inside out to mean that the creative expression of Oracle’s code isn’t copyrightable. EFF filed a brief with the Supreme Court on this case this week (separate from PK) that no longer quotes Bikram, but continues to argue Oracle’s code is not copyrightable because it isn’t creative. You or I might choose different words to express a concept, but that is more or less the soul of copyrightability – the protection of the author’s specific expressive choices.
For instance, the Ninth Circuit language in Bikram would never suffice as a reason to deny copyright in non-fiction works like news articles. Ten journalists developing their own versions of roughly the same story will be working with the same set of facts (we hope), but each will employ what we might describe as “multiple different methods”—arrangement of material, vocabulary, tone, amount of detail, editorial—to “attain a particular end.” So, copyright in journalism very much does attach principally because each author takes her own approach to communicate similar—if not the same—useful information to the reader. The same is true for programmers developing software to perform the same or similar functions.
All subjects of copyright (whimsical products of subjectivity that they are) have long been described and debated by analogy; and this is one reason why precedent language can be rather vexing when a case involving one category of works is cited in a case involving a different category—let alone comparing a non-subject like yoga poses to a statutorily defined subject like computer code. But perhaps all this “filler” has something to do with the business story behind this conflict.
To recap, hundreds of commercial developers licensed Java before Google set out to make Android—not one of them saying, “Hey, this isn’t copyrightable!” Google was about to license Java in 2005, but declined to do so because the license required a level of interoperability, which (by all appearances) would have been a barrier to Google’s interest in market dominance—a dominance it now enjoys. So, it does begin to look just a bit like Google went searching after the fact for a legal theory to justify its decision to simply bypass a creator’s copyright. Sound familiar? This is why is it worth asking how much this case is really about the copyrightability of Oracle’s declaring code in Android, rather than a challenge to the copyrightability of computer code in general—or to any protected works for that matter.