May 03, 2012, 10:59 AM — As the software industry waits for the jury to come back with a decision in the copyright phase of the Oracle vs. Google trial, there has been increasing focus on what it might mean if the 37 Java APIs are found to be copyrightable and were indeed infringed by Google.
Traditionally, application programming interfaces (APIs) have been presumed to be non-copyrightable, because unlike other elements of a software, which involve creativity, APIs are typically comprised of facts that enable one specific task: how does my software program talk to your software program and vice versa?
In the Oracle trial, Oracle is maintaining that by copying the 37 APIs collectively, it was the same as if Google would have infringed on the structure, sequence, and organization (SSO) of a computer program, which is not the same as an API. By their very combination, Oracle is arguing, the APIs become a creative, and therefore copyrightable, element of software.
If the jury buys into Oracle's arguments, and the judge affirms them in his ruling, then this could be very bad, at least in the Ninth Circuit. APIs have never been perceived as copyrightable in industry practice, because they are essentially just a list of names and libraries with which software developers can plug into other applications. Make the lists copyrightable, and suddenly you have a very toxic situation.
(The Ninth Circuit, it should be noted, covers California and the West Coast, which is where a lot of software developers hang their collective hats.)
As you can see, a lot is hanging on this decision, more than just whether Google ripped off Oracle. It's the method of the theft (if any actually occurred) that's being tried as well.
It's an interesting distinction. In the past, court cases have found that expressions of ideas, not the idea itself, are what's copyrightable. I asked Pamela Jones, founder of Groklaw, for some help understanding this distinction, and she provided a pretty good analogy. Jones being Jones, it was a legal one, of course.
"If you watched the trial, you saw lawyers would stand up and say 'objection,' if they didn't like a question being asked by the other side," Jones wrote. "There is a list of all the possible objections you can raise."
"All lawyers know that list, although they might have to look up some of the more obscure ones to review. But if a lawyer stands up and says, 'Objection, hearsay' everyone in the room knows what it means. It's referring to the list. But in reality it's an idea. Everyone knows the definition and what kinds of uses it's appropriate for and they have that in their minds when he says, Hearsay. So hearsay is an idea. You can't copyright an idea. That's in the law.
"Saying 'hearsay' in a trial isn't to provide the definition of that type of objection. The definition isn't an idea. It's the expression of the idea. To get that expression, you'd have to look it up in the list. To implement it, you have to say it to the judge, answer any questions as to why you think it applies to the question, etc., and then the judge rules one way or another," she continued.
Jones makes a very detailed argument to get her point across:
"Now, suppose the list of objections is listed alphabetically. But you think it'd be very cool to rearrange the list some other way. Now is an item on the list copyrightable? If I rearrange the phone book starting with Z and going to A, now is it copyrightable? Oracle would say yes, presumably. That is the issue, in that copyright law says that the amount of creativity required for copyright is minimal.
"It's possible the case could go that way, and that's why everyone is pointing out how toxic this all is on Oracle's part.
"Is the alphabet now copyrightable? How about the list of objections? That's Oracle's argument, in brief. Of course APIs can be a lot more complex than a list of objections, but it's the implementation and the documentation that makes it so, more than how they may interact. Google used 37 APIs out of a list of over a hundred, so in that sense they didn't even use the complete SSO. I don't really understand how Oracle can make the claim with a straight face."
The case for copyrightability of APIs is so important for Oracle to win that Oracle consultant Florian Müller tries to make the argument that in the U.S., APIs are already copyrightable--and have in fact been so since 1989.
But in his attempt to make this claim, Müller seems to lack a full knowledge of how U.S. law works.
Müller cites the Ninth Circuit case Johnson Controls v. Phoenix Control Systems as proof of his claims. Johnson Controls was also cited quite a bit by Oracle in their case against Google, so it's little surprise that Müller would pull it out again.
But there are a few problems with using Johnson Controls. First, it was superseded by a later case Feist Publications, Inc. vs. Rural Telephone Service Company, which specifically upheld the notion that non-creative elements of software are not copyrightable. Müller seems to ignore Feist entirely, but you really can't: figuring out the nuances ofJohnson Control, Feist, and the arguments from Oracle and Google is pretty much what Judge Alsup will have to do when he determines the copyright status of APIs.
I asked Jones about this line of reasoning, and she pointed out a confirmation of the idea versus expression concept even in Müller's favorite case:
"To establish a case of copyright infringement, Johnson must prove ownership of the copyright, and copying of an expression protected by its copyright."
More directly, Jones replied:
Johnson Controls wasn't about the same type of facts that Oracle v. Google is. So far as we can tell, it was about the SSO of a computer program, not SSO of APIs. It's hard to tell because it was an appeal of a preliminary injunction, and the rules on that are pretty light. An appeal of a preliminary injunction is only possible if there is a clear mistake of law. The appeal said at that point, it couldn't rule that there had been such a mistake, but at trial it assumed the issue would come up again.
"So it didn't establish a rule for APIs. It also doesn't say what FOSSpatents seems to think it says. And it was superseded by Feist, which held that the SSO of a computer program is only copyrightable if the structure, sequence, and organization are creative."
If APIs are indeed copyrightable, as Müller maintains, then the question of their copyrightability wouldn't even be raised in the Oracle case--the judge and the lawyers would be dealing with much different arguments. But if Judge Alsup has to decide on the status of APIs and copyright, doesn't that presume the issue is, I don't know, undecided?
Jones summarizes that this is what the case is all about.
"So that is the issue before the court here in Oracle v. Google. Are the APIs creative expression or just functional code that can't be copyrighted? Are they part of the Java language? Can you use the language without the APIs? If not, in what sense was the Java language released under the GPL by Sun?"
Hopefully we'll find out soon how this will parse out. Yesterday's ruling from the EU Court of Justice on the non-copyrightability of computer functions and languages was certainly a step in the other direction. For now, nothing is certain, no matter what FUD you hear.
Read more of Brian Proffitt's Zettatag and Open for Discussion blogs and follow the latest IT news at ITworld. Drop Brian a line or follow Brian on Twitter at @TheTechScribe. For the latest IT news, analysis and how-tos, follow ITworld on Twitter and Facebook.