I think it's closer to a recipe book. There are only so many ways you can make an apple pie and although you wouldn't steal the exact wording from another book (unless you're an idiot) there's nothing wrong with reading a book, adding a bit of cinnamon, juggling the quantities, describing the pastry rolling out a bit better but in your own words, etc.
What Google copied was the necessary parts of the process. What Oracle are claiming is that only they can tell you how to cook an apple pie, and every apple pie recipe is somehow based on theirs, even when someone has entered into an "apple pie making industry". Collect recipe books and, pretty much, they will all have the same recipes in them, and the recipes will be very similar. The expression is in the details, in the actual description, and even in the collection and order of the recipes inside the book. It's NOT in the process of how to make an apple pie that is edible and recognisable to others as an apple pie.
Oracle really scrape the barrel, like SCO did. No sensible lawyer in the world would try making that argument after the judge ruled the code uncopyrightable - that's almost a direct attempt to say "we still think it's copyrightable" by even making the comparison, without actually expressing that directly. The judge isn't going to like that, and the more publicity the analogy gets, the more damage done to Google on a false analogy.
Makes you wonder who's behind it really, again. Just as the SCO thing gets settled (by driving themselves into bankruptcy deliberately to pursue a lawsuit totally without merit), suddenly there's a new threat to the copyrightability of code, on a Linux-based operating system, using open technologies, that compete in areas that a certain company is trying to push into.
How hard is it? You can't copyright a process. You can only copyright an expression of that process. If someone copies a Haynes manual, they would have grounds to sue them. But if someone else writes a book about the correct way to dismantle a Ford Mondeo, which uses none of their pictures and none of their text but might have similar (trivial) chapter headings ("The Exhaust System", etc.) then they don't.
The judge already decided this issue in terms of what's REQUIRED for interoperability (the "chapter headings"), what's uncopyrightable (the "chapter headings", and the trivial bits that are necessary to make them work properly which in code terms means quite a lot of code that has no other sensible way they can be written - e.g. returning certain structures means filling those structures, etc.), and whether Google did anything wrong. Annoying them by putting out a false analogy that suggests they were wrong, without formally challenging his individual points, is only going to end in tears for Oracle.
Damn, I'm glad they shoved OpenOffice out of their way because they didn't want to touch it. Probably the only saving grave to come out of Oracle in decades.