Section 102(b) reads as follows: "In no case does copyright protection for an original work of authorship extend to any idea, procedure, process, system, method of operation, concept, principle, or discovery, regardless of the form in which it is described, explained, illustrated, or embodied in such work."
The idea here is not to give anyone a monopoly via copyright over such fundamental concepts as ideas, etc. Things that are purely functional within systems and the like fall within 102(b) and cannot be copyrighted even if they are otherwise expressive. That is what Google is arguing about the 37 Java API packages in this case.
So did Huffman encoding. So did the Laplace transform. So did quicksort, red-black trees, and hell, even the Universal Turing Machine. These are all the result of many hours of creative endeavor, yet they are none of them copyrightable.
This is the same thing that makes Wolfram's claims of ownership of the existence of a proof of Rule 110 CAs as universal calculators particularly onerous.
My difficulty comes from the fact that designing a good API is hard, and I think that there clearly are creative aspects to it. Doing a good job requires having an intimate knowledge of the component being interfaced with, and how programmers are likely to use it. The mere fact that we have said before "This API sucks" implies that it is not just a mechanical process.
Basically, I want one result because I feel it's better for society. But based on first principles, that's not the conclusion I arrive at.