Thursday, September 15, 2011

Judge throws out Google's summary judgment motion on copyright almost entirely

Only two business days before court-ordered settlement talks between Oracle's and Google's CEOs begin, Judge William Alsup has denied a Google motion for summary judgment on Oracle's copyright infringement claims almost entirely. I uploaded the decision to Scribd. The judge wrote:

"With one exception described below, the motion is DENIED."

That sentence says it all, especially since the one exception is of very limited value on its own. On a scale from 1 to 10, Google succeeded at a level of 1 or at best 1.5 -- and failed at a level of 8.5 or 9.

While the same federal judge allowed Google (in a previous decision) to bring yet another motion related to the "damning" Lindholm email, the odds are long against such a motion even though Google received permission to file it. By comparison, today's copyright decision is far more important and impactful.

Google's original motivation and smashed hopes

Google had started this summary judgment process in early August in order to reduce this patent and copyright infringement case to just a patent case. If this initiative had succeeded, Google would have gained three key benefits:

  1. A case that would have been only about patents would have been significantly more likely to be stayed for the duration of reexaminations of the patents-in-suit at the patent office than a litigation involving patents and other claims that aren't subject to reexaminations.

  2. Google wouldn't have wanted a jury (if this case is not settled and instead goes to trial) to see the evidence of its direct copying of Oracle code. Google never denied the copying -- it just denied that it should be liable for it. But with today's decision, Google will be exposed as a reckless copycat and infringer, which can psychologically affect a jury's willingness to agree with Oracle on the presumed validity and alleged infringement of its patents-in-suit (even though those are different types of intellectual property rights).

  3. Copyright has a much longer term than patents. As a result, damages claims for future infringement can result in even more staggering amounts, though one may certainly wonder whether Android is still going to use Java in, for example, 2050 or 2060...

This is a setback for Google. In a formal sense, Google stood only to gain from this: it was an opportunity to get rid of significant parts of the problem before a possible trial. But the extent to which this summary judgment motion failed is truly remarkable. More importantly, some of what the judge writes in his 13-page decision indicates that Google is quite likely -- at least much more likely than most people thought -- to be held liable for copyright infringement at the end of this litigation. It seems that the question is only the scale on which this will happen, and that scale could be very significant.

And to make things worse, the whole question of API copyrightability could affect Google in connection with the Bionic/GPL issue I blogged about in March. Google uses large numbers of API header files from the Linux kernel and other GPL-licensed software, throwing out comments and other material and declaring the rest unprotected by copyright. Today, Judge Alsup clearly refused to follow Google's non-copyrightability theories for the most part. He stated expressly that today's order "makes no finding as to whether any other elements of the API package specifications (or their selection or arrangement) are protected or infringed". In other words, the order doesn't say that Oracle is necessarily right. But there are people out there who believe that the copyrightability of API-related material such as header files is a given, and they should realize now that this is a sufficiently complex and case-specific issue that the judge refused to throw out Oracle's copyright infringement claims.

Google can certainly afford very capable lawyers, but even they couldn't get Oracle's copright infringement assertions dismissed at this stage.

Details of the decision

In order to keep this post reasonably focused, let me refer you to my recent post on how Oracle defended its copyrightability and copyright infringement assertions in connection with a set of copied files -- many of which I presented on this blog back in January -- and, especially, its Java APIs (application programming interfaces). Here I'll just summarize Judge Alsup's decision item by item.

The judge points out that "[t]he term API is slippery". He refuses to use the term "API" on its own for anything other than "the abstract concept of an application programming interface". For the sake of greater precision, the judge consistently refers to "API documentation" and "API implementations" when discussing material that was allegedly infringed.

Copied code files

Google used a highly dubious de minimis argument to justify the copying of 12 code files, arguing that (as the judge sums it up) "its alleged copying should be compared to the entire Java platform, which Oracle registered as a single work". That would mean it's pretty easy to steal code from a large program, especially if the stolen code gets incorporate into another large program. Oracle, however, argued that "that each of the twelve code files at issue is a separate work". The judge analyzed Google's two references to case law on de minimis copyright violations and concluded that "at least on the present record, Google has not shown that the Java platform as a whole is the work to which Google's alleged copying should be compared".

Therefore, he refused to declare Google's alleged copying as legal under the de minimis rule. Google can try to make that claim in its presentations to the jury, but given how the judge disagrees with Google, I guess Google will have a hard time.

It seems the judge also wasn't impressed by Google's argument that some of the files are just "test files". An infringement is an infringement.

Google's only valid point: names not copyrightable

The judge agreed with Google that "the names of the Java language API files, packages, classes, and methods are not protectable as a matter of law [...] on account of the words and short phrases doctrine".

The bad part for Google -- even in this context, which is the only one in which Google achieved anything with its motion -- is that the judge explicitly separates the question of whether those names are copyrightable on their own from the copyrightability of "the selection and arrangement of elements within a work". Ultimately, if Google won on the copyrightability of selection and arrangement of elements, it wouldn't even matter whether the names are copyrightable or not.

Scènes a faire and merger doctrines

A cornerstone of Google's argument was that API-related code -- at least the one at issue in this case -- falls under two rules, the scènes-a-faire doctrine and the merger doctrine. An overly simplified way to describe them is that if something is like a necessary prerequisite, or if "the idea underlying the copyrighted work can be expressed in only one way", then the related material isn't copyrightable.

Judge Alsup concluded that Google didn't really specify which parts of the asserted API-related material it meant. If Google meant names, then the judge agrees on the non-copyrightability of names anyway. If Google meant anything else, "Google's lack of specificity is fatal", the judge writes.

He points out -- in my view, quite accurately -- that some API-related code will probably fall under one of those doctrines, but not all of it:

"[Google's] approach ignores the possibility that some method declarations (for example) may be subject to the merger doctrine or may be scenes a faire, whereas other method declarations may be creative contributions subject to copyright protection. Google has not justified the sweeping ruling it requests. Google has not even identified which categories of specification elements it deems unprotectable under these doctrines. This order declines to hold that API package specifications, or any particular category of elements they contain, are unprotectable under the scenes a faire or merger doctrines."

Methods of operation

Google argued that "APIs for a programming language" are unprotected methods of operation. In the judge's understanding, "Google states that all 'elements common to Oracle's Java language APIs and the Android APIs are unprotectable methods of operation'". And he isn't impressed:

"Google's argument that APIs are unprotectable methods of operation attacks a straw man. It is not the APIs but rather the specifications for 37 API packages that are accused. Even if Google can show that APIs are methods of operation not subject to copyright protection, that would not defeat Oracle's infringement claim concerning the accused specifications."

Those who followed the debate over the potential copyrightability of header files should take note of the part the judge emphasized. Even if one can argue that APIs in an abstract sense aren't copyrightable, specifications may very well be, depending on the specifics of each case. Today's order "does not decide whether APIs are methods of operation" (simply because Google did not tee up its argument in a way that the judge believed he could take a decision on at this stage) and explicitly "finds that the API package specifications at issue are not 'methods of operation' under 17 U.S.C. 102(b)."

Degree-of-similarity argument

Here's a passage that explains the argument Google made and the judge's primary reason for disagreeing with it:

"Google argues that '[g]iven the substantial unprotected elements in the documentation (such as the API method declarations), the ‘virtual identity’ standard applies here' [...]. This order agrees with Google that the names of the various items appearing in the disputed API package specifications are not protected by copyright. Google, however, has not shown that any other elements of the specifications are exempt from copyright protection. Because Google has not proven that a substantial portion of the specifications is unprotected, Google’s justification for applying the virtual identity standard fails."

Fair Use

Google raised the defense that its use of certain Java-related material was "Fair Use" even if one deemed that material copyrightable. In this context, Google also describes Android as (without using that particular word) a blessing for Java. The judge, however, says that "[o]n the present record, a reasonable fact finder could disagree with Google's rosy depiction of Android's impact on the Java market". He denies summary judgment on the Fair Use question since anything over which there's a dispute of fact must be put before a jury.

Today's order is one of the best news Oracle had so far in this litigation.

If you'd like to be updated on the smartphone patent disputes and other intellectual property matters I cover, please subscribe to my RSS feed (in the right-hand column) and/or follow me on Twitter @FOSSpatents and Google+.

Share with other professionals via LinkedIn: