From Newsweek:
Google calls it the “copyright case of the decade.”
“It” is the $9 billion copyright infringement suit Oracle filed against the search giant nearly 10 years ago. Oracle brought the case in 2010 after Google incorporated 11,500 lines of Oracle’s Java code into Google’s Android platform for smartphones and tablets. Android has since become the world’s most popular operating system, running on more than 2.5 billion devices.
Google won twice at the U.S. District Court level. But each time, a federal appeals court overturned the verdict, ruling for Oracle. Now, Google is begging the Supreme Court to hear the case, and so are the 175 companies, nonprofits and individuals who have signed 15 friend-of-the-court briefs supporting Google’s plea.
. . . .
Here’s the pressing issue: How much protection do copyright laws give to application program interfaces, or APIs? That might sound arcane, but these interfaces are omnipresent in software today. They form the junctions between all the different software applications developed by various companies and independent developers that must seamlessly interact to work right.
All the apps that sit on our smartphones—like Pandora or Uber—use interfaces to communicate with our phones’ operating systems (Apple iOS for iPhones, for example). If the owner of a platform can claim, through copyright, to own those interfaces, it can limit innovation and competition, Google contends. Not only can it determine who gets to write software on its own platform, but, as we’ll see, it may even be able to prevent rival platforms from ever being written. The Harvard Journal of Law and Technology considers the case so consequential that it devoted an entire 360-page “special issue” to it last year.
“If the appeals court’s rulings stand, it’s likely to lead to entrenching dominant firms in software industries,” says Randy Stutz, an attorney with the American Antitrust Institute, which supports Google in the dispute.
Oracle, on the other hand, says the case is cut-and-dried. Its basic argument: Google negotiated to take a license for the Java code, it wasn’t able to reach terms, and then it used portions of the code anyway. (And that’s all true.) Now, it’s time to pay the piper.
. . . .
“Before Android,” Oracle’s lawyers write in their brief to the Supreme Court, “every company that wanted to use the Java platform took a commercial license…including smartphone manufacturers BlackBerry, Nokia and Danger.”
Oracle claims that, if not for Android, Oracle’s own Java software could have become a major smartphone platform. (Although Java was written by Sun Microsystems, Oracle acquired Sun in 2010, shortly before bringing this suit.) Oracle’s lawyers mock the notion that the rulings in its favor will spawn any dire consequences. Despite Google’s “sky-is-falling” arguments, they write, the software industry did not crash in the wake of May 2014 or March 2018, when the U.S. Court of Appeals for the Federal Circuit issued the two key rulings that Google seeks to reverse.
In fact, Oracle has enjoyed fervent support from its own friend-of-the-court briefs, including one from BSA, the Software Alliance, which counts companies like Adobe, Apple and IBM among its members.
Remarkably, for a case about software interfaces, the key Supreme Court precedent was decided in 1879. Obviously, that suit didn’t involve a smartphone platform, but it did define the limits of copyright and explain how a copyright differs from a patent. In that dispute, Charles Selden had authored and copyrighted a book laying out a method of bookkeeping. The book included some blank forms that could be used to implement the system. Later, W.C.M. Baker began marketing his own set of forms to implement Selden’s method that were very similar to those in Selden’s book.
Selden’s widow sued Baker for copyright infringement—and lost. Basically, Justice Joseph Bradley explained in the opinion, she was trying to use copyright to protect the ideas contained in Selden’s book. He explained that, while a patent can protect an idea, a copyright protects only expression—in this case, the particular words Selden used to describe his bookkeeping method. “The copyright…cannot give to the author an exclusive right to the methods of operation which he propounds,” the Supreme Court’s decision said. (Selden had not patented his bookkeeping method.) Since Selden had no monopoly on his method, he had no monopoly on the forms needed to carry out that method.
Congress later wrote the Court’s Baker v. Selden ruling into the federal copyright statute, specifying that a copyright cannot “extend to any idea, procedure, process, system, [or] method of operation,” even if that idea is “described” in copyrighted work.
Link to the rest at Newsweek
FWIW, I have 40+ years of experience in software development. I am still active, and also teach it. And I am definitely outside of -dsr-‘s “first approximation”.
I have absolutely no doubt that an API is a creative work consisting of a concrete representation in language of an abstract idea. As such, it meets the usual requirements to be a copyrighted work.
One need only spend the time to compare a badly written one to a well written one to realize that one of the major differences between the good one and the bad one is that the good one is expressive to the _humans_ who read it and use it. The success or failure of an API will usually depend as much upon its expressiveness to humans as upon pure technical factors.
When an API is published, it is published as an intermingled set of computer language expressions and commentary to other programmers. I don’t see how anyone could possibly argue that the commentary is not subject to copyright. But I’d argue that the expression of ideas in the computer programming language is also deserving of copyright protection.
Yes, an API is a set of instructions on how computer programs will communicate. But they are also expressions of an abstract idea on how such potential communications might be used to do something useful. The abstract idea is not, of course, copyrightable. But, as a general rule, concrete expressions of abstract ideas are.
Describe the same abstract idea to twelve different programmers and they will write twelve different APIs for that abstraction. Those APIs will differ in how things are named, in how the component ideas are divided up and inter-connected, and on the degree to which they reveal the underlying abstract ideas to or hide them from to the human readers. Is there any doubt, then, that those programmers were engaged in a creative process?
Oracle should name you as a witness. 🙂
All the doom and gloom talk ignores one basic fact: the Android coders used Sun/Oracle Java text verbatim, comments and all. This was not a clean room reverse engineering but a straight copy. If it weren’t for the massive monetary claims, the case would have ended long ago.
It’s not a particularly good case to try to set up as a major precedent unless you are inherently opposed to software copyrights.
Felix, you’ve hit the nail on the head.
We have programs that can do an automated reconstruction of an API from compiled code. Google could have done that and probably avoided the whole kerfluffle. But the resulting API would have been totally useless because no human programmer would find it readable.
There’s a more-or-less accepted practice for reverse-engineering software, that starts with such an automated reconstruction and then relies on humans reformulating the results into something readable, while following clean-room procedures to avoid polluting their work with any of the expressive content from the original. Google didn’t follow those practices, and they probably should get slapped for it.
(It pains me to side with Oracle on this. Over the years, Oracle has become the company that software developers most love to hate. They’re the IRS and DMV of the computing world, rolled into one.)
To a first approximation, every programmer will laugh at the very idea that an API is copyrightable. It’s the most factual of factual documents: a precise guide of inviolable etiquette, for one computer program to speak to another.
I believe that the majority of the exceptions work for Oracle.