Best case: judge concludes that the 37 APA packages are purely functional and hence not protected by copyright. Ninth Circuit precedents (in my view) preclude him from doing this as a categorical imperative ("all APIs are inherently unprotectible by copyright") but nothing prevents the judge from saying, in effect, that the experts are in essential agreement that the "structure, sequence, and organization" of these APIs plays a purely functional role (to facilitate calls to an underlying black box that has the implementing code) and is therefore merely a method or system of operation under 17 U.S.C. sec. 102(b) (meaning, Congress has denied copyright protection to such elements of a software program even if it otherwise constitutes a form of original expression). Such a ruling would be entirely consistent with Lotus v. Borland (a First Circuit case affirmed by a divided Supreme Court) but would depend heavily on the judge's assessment of the factual evidence (i.e., what the experts testified about how the APIs work within the Java context) and not on categorical pronouncements about APIs generally. In my view, the Ninth Circuit precedents would block a generalized ruling but not a fact-specific one. That is why the judge had to kick down the road so often. He had to wait until all the evidence was in to make a final assessment. Because the ruling would be fact-specific, it would not be an all-out victory for interoperability but it would still be very strong.

If the judge cannot reach such factual conclusions from the evidence, then he is left with concluding that the SSO of the APIs as represented in Java do have expressive elements and are hence protected by copyright. This would mean a retrial with instructions given to the jury much as before and with a heavy fight over fair use.

Of course, in the end, who really knows. It is clear that the judge has done extensive homework on this and that is perhaps the best sign that the decision will be well thought out. Only time will tell.