Oracle v. Google, Copyright & Fair Use: One of “Billions” of Summaries

Earlier this week, the Federal Circuit Court of Appeals released its blockbuster decision in Oracle America, Inc. v. Google LLC, which held that Google’s unauthorized use of certain aspects of Oracle’s Java software was not fair use. In the past few days and in the coming weeks, nearly every lawyer who has ever had occasion to turn to Title 17 is going be writing, blogging or blabbing about this case, so you really don’t need to read this post – just log on to Twitter and absorb the opinion by osmosis.

But we are writing about it anyway. Why? There are three things that make this opinion, in our opinion, stand out from the crowd.  First, there is a ton of money at stake – we’re talking Carl Sagan numbers (“billions and billions”), which is a lot for a copyright case (there is a patent litigator in your office about to tell you that he’s currently working on multiple billion-dollar cases – tell him to shut up and let us have this moment).  Second, the opinion is notable because of its intellectual hostility towards the role of the jury in determining fair use. And third: Google actually lost a copyright case! If that’s not news in and of itself then consider that this particular opinion arguably continues a recent trend of copyright decisions favoring content owners over technical innovators.

Oh, and one other thing … this opinion is also notable because of what it doesn’t include: an analysis of Google’s most interesting fair use argument. Google had argued that its copying was in service to a greater good, that is, to make all your devices and platforms “interoperable,” thus saving the human race from another VHS vs. Betamax-style format war. If you want to find out why this argument disappeared, read on. Let’s get started.


Oracle owns copyrights in software related to Java, a computer platform that allows a programmer to write one code and have it run pretty much anywhere, including your smart phone. This software includes “application programming interfaces,” or “API packages.” These API packages (Oracle developed about 166 in all) contain elaborately organized collections of pre-written source code for certain common tasks. In other words, they are shortcuts for programmers.

The API packages have two kinds of code. First, the “declaring code” is a relatively short header that contains functional specifications and identifies a task to be performed. Second, the longer “implementing code” contains the step-by-step instructions telling the computer how to execute that task. So for example, the declaring code identifies the task “find the greater of two numbers,” and the implementing code tells the computer how to find the greater of two numbers.

The Java programming language itself is free for anyone to use but, if you want to use the APIs in a competing platform, you have to pay. Back in 2005, Google tried to license the API packages from Oracle for Google’s Android platform, but the parties couldn’t agree on terms. Google decided to create its own packages instead, and wrote millions of lines of its own implementing code. However, in order to make things easier for programmers already familiar with Java, Google copied from Oracle the declaring code contained in 37 API packages – about 10,000 lines of code in all.

Procedural Posture

Oracle sued. A jury found that Google had infringed, but was deadlocked over the issue of fair use. The Northern District of California put the verdict aside, on the ground that the API declaring codes were not subject to copyright protection in the first place. Because the case at one point had patent claims as well, Oracle’s copyright appeal went to the Federal Circuit, not the Ninth Circuit (that patent litigator is chuckling again – throw your copy of Nimmer at him). In 2014, the Federal Circuit reversed (we wrote about that opinion here), holding that the API declaring code was protected by copyright. The Federal Circuit remanded the matter for more proceedings on the issue of fair use, in particular calling out Google’s intriguing argument that its use of Oracle’s software was fair because it was engaged in the societally beneficial project of achieving “commercial interoperability” between different platforms and programs.

So back the case went to the Northern District of California. A new jury found that Google’s copying of the Oracle programming was fair use. This time, the District Court refused to throw out the jury verdict, denying Oracle’s motion for judgment as a matter of law. Oracle appealed again.

We Don’t Need No Stinking Jury!

Perhaps because the Federal Circuit was about to overturn a jury verdict that had been affirmed by the presiding judge, the Court went to great lengths to articulate its view of the proper place for a jury in a fair use determination.  The Court described fair use as a mixed question of law and fact, but one that as a practical matter is more law than fact, especially in light of its equitable roots.  The jury gets to determine only disputed “historical facts.” For example, if the parties hadn’t been able to agree whether Google copied 36 API packages or 37 API packages, that would be an appropriate thing to submit to the jury. But other than that, according to the Federal Circuit, the jury is supposed to get out of the way. It is for the Court to decide not only what legal standards govern, but to apply those legal standards to the facts.

The upshot of all this is that a jury’s findings related to fair use – other than those regarding historical fact – were “advisory” only, and the Federal Court’s standard of review on the question of fair use was more or less de novo.

The Fair Use Analysis

The Court applied the familiar four-factor fair use test set forth at 17 U.S.C. § 107, as informed by the governing case law from the Ninth Circuit (because copyright law is not within its exclusive jurisdiction, the Federal Circuit applies the law of the regional circuit in which the lower court sits).

Purpose and Character of the Use. The first factor, the purpose and character of the use, is usually divided into two elements: (1) whether the use is commercial; and (2) whether the use is transformative. The Court held that this factor weighed strongly against fair use. First, the essential commercial nature of Google’s operation was undisputed.  Google argued that this should be offset by the fact that Google engages in non-commercial activities as well but, because “Google reaps billions from exploiting Java in Android,” the Federal Circuit wasn’t buying it.

The Court then turned to “Transformative Use.” A use is considered “transformative,” and therefore more likely to be a fair use, if it adds something new, such that the work has a different purpose or character than the original, and (importantly) does not merely supersede the original. Here, Google was using the copied declaring code for the same basic purpose as Oracle – to help developers create Java programs. Google argued that this was nevertheless transformative because it copied only 37 of the available 166 API declaring code sets, and surrounded that declaring code with original implementing code to create a “fresh” work for a “new context,” that new context being smartphones instead of desktop computers.  The Federal Circuit rejected this argument because “taking only select passages of a copyrighted work is, by itself, not transformative.” Additionally, moving something to a “new context” is not transformative if the work is still serving the same purpose (and, it should be noted, the Court was skeptical that the “smart phone” market was really a “new” context).

Nature of the Copyrighted Work. The second factor is the nature of the copyrighted work, which recognizes that some works (e.g., creative fiction) are closer to the core of intended copyright protection than others (e.g., factual compilations).  The closer a work is to that intended core protection, the less likely the use will be a fair one. The District Court opined that the jury may have found the APIs functional, not creative, and therefore the factor favored fair use.  The Federal Circuit decided to pick its battles and let this one go into the “fair use” column, but noted that the second factor is rarely “significant in the overall fair use balancing.”

Amount and Substantiality of the Portion Used. The third factor examines how much of the infringed work was used in the infringing work, and whether what was copied was qualitatively important.  Defendants can sometimes prevail on this factor by arguing that, even if they took a lot, they took no more than was necessary for the intended use. This argument works for highly transformative works such as parody, but the Court held that it is of no succor to defendants engaged in non-transformative uses. Here, Google argued that it copied only a small percentage of Java language (11,500 lines of declaring code out of nearly 3 million lines of code in the Java library), but the Court held that no reasonable jury could find that what Google copied was not qualitatively significant.

You may be wondering what happened to Google’s nifty interoperability argument, which is that Google only borrowed as much code as was necessary “to avoid confusion among Java programmers as between the Java system and the Android system.”  The Federal Circuit acknowledged that it had been intrigued by this argument back in 2014, but asserted that Google’s actual good deeds were not keeping pace with its lawyers’ clever arguments. According to the Court, it turns out that Google was not fostering interoperability; in fact, it was intentionally designing its Android system to be incompatible with other platforms using Java. So, bye-bye interoperability. Google had to drop this argument from its appellate arsenal, and factor three weighed against fair use.

Effect on the Potential Market. The fourth factor, the effect of the use upon the potential market of the copyrighted work, reflects the notion that fair use should be limited to uses that don’t usurp the original work’s market. Google argued that the parties were in different markets – smart phones versus software for desktops. But the Federal Circuit held that even if those were different markets, the evidence of actual and potential market harm was “overwhelming,” including Oracle’s previous forays into the smartphone market; its intent to enter the market again; and the fact that third parties have used the existence of the competing Android platform to drive down the price of their Java licenses from Oracle.

Balancing the factors as analyzed above, the Court had an easy time determining that the overall balance tipped against fair use.

Transformed, but not Transformative

We’re getting calls from our friends in Cambridge, Silicon Valley and other tech hot spots. They want to know if this is end the world. The answer is “no,” but also “maybe” (from your perspective). First the “no.” The Federal Circuit understood that its decision would be the subject of much hyperventilating, and therefore went out of its way to state that its opinion was confined to the case at bar and that it did “not conclude that a fair use defense could never be sustained in an action involving the copyright of computer code.” It also should be noted that that Google will almost certainly appeal, so maybe this will all amount to a hill of beans.

Now for the “maybe.” In the last few years, we’ve seen what appears to be a trickling stream of important copyright opinions that favor content owners over the creators of technical innovation designed to use content without permission. A few that come to mind:

  • American Broadcasting Companies, Inc. v. Aereo, Inc. (which we wrote about here), where the Supreme Court examined a clever antenna array intended to avoid technical infringement of the “public performance” right by instead enabling lots of unauthorized private performances. The Court declared that it would not allow technology to make an end-run around copyright law and, placing economic substance over technical form, held that the technology was an infringement of the performance right.
  • Fox News Network, LLC v. TVEyes, Inc., where the Second Circuit held that TVEyes’ use of broadcast news content to create a searchable database of news was not fair use (we wrote about the now-overturned lower court opinion here).
  • Even more eyebrow raising for many, the Southern District of New York just issued a controversial decision in Goldman v. Breitbart, which involved a bunch of news organizations using a copyrighted photo without permission. The new organizations argued that they could not be liable because they never hosted a copy of the photo on their servers; rather, they merely embedded on their websites code linking to the photos. So, even though it looked to internet users (and to the Court) like the photograph was hosted on their servers, it wasn’t. The Court held that, even though these news organizations never possessed a copy of the photograph, their role in actively causing it to appear on their customers’ screens made them liable for infringement.

These cases are not all about fair use, but the common thread among them is that the basic law of copyright cannot be undone even by the most amazing technical accomplishments. It’s tempting to assume that, because something transformed our lives, it must also be “transformative” for copyright purposes. Many of Google’s projects have indeed been held to be “transformative” (including the ones discussed here and here). But in this case, although Google used Oracle’s code in service of a host of culturally transforming products, the Federal Circuit held that it did not make a “transformative” use of Oracle’s code.

One thought on “Oracle v. Google, Copyright & Fair Use: One of “Billions” of Summaries

Leave a Reply

Your email address will not be published. Required fields are marked *