I don't see why. There's plenty of creative discretion involved in the writing of an API. I would have expected this to be a fair use issue rather than a copyrightability issue (though IANAL).
"Fair use" is, by definition, an issue involved in copyright. But I see your point- if APIs are copyrightable, then uses of that API may be "fair use." But then someone's got to drag all the potential uses through the courts to determine what's fair use and what's not (because apparently fair use can only be determined by the courts...) I would surmise that, in this dystopian fantasy world (and I hope it remains such), 'interoperability' would be the de facto fair use and any other use of an API would legally require a license. What a nightmare.
Sure, I'm just addressing your fair use comment with my own input.
"...non-interoperable reimplementation of the API"
I do not understand this phrase. If you implement an API that's exactly the same as the API in a library that I publish, the underlying implementation is irrelevant to your clients because your software is now an interoperable replacement for mine. If you implement some API that's not the same as my API (and thus not an interoperable replacement), you can't have infringed on my theoretical API copyright.
Whether or not we limit he discussion to this case, I don't think there's any reason to fear that using an API would be a copyright issue (though again IANAL). The whole purpose of an API is for people to write software against it. I think that people misconstrue what this case is really about and start talking in hyperbolic terms (not that you're doing so, but in the popular press, blogs, etc.).
At the risk of mischaracterizing Oracle's argument, the basic idea is that Google wanted developers to have a familiar API and wanted to reuse some of the tooling built for Java, but there is no way to get a compiled Dalvik program to run on a JVM, and likewise compiled Java software can't be run on Android. So Google piggybacked on the hard work Sun put into designing a coherent (and quite large) API, but didn't really do so in an interoperable way. In some ways the case is reminiscent of the Sun/Microsoft quarrel about Microsoft's extensions to Java (though that also involved trademark issues, as I recall), though many commenters who supported Sun in that case now seem to support Google in this one.
Because it's not an act of creativity to decide that a name or address is needed (for example), and there's very little creativity in deciding to ask for the first name first and the last name second.
