Almost: copyright is indeed about form, but in the United States, it is based on _originality_ of form, not just how much work went into the form. In particular:

http://www.bitlaw.com/source/cases/copyright/feist.html

That’s the ruling that makes the US different from where the EU and UK is going.

No question that the original document, as a whole, is probably copyrightable. But my point is that if the file is a set of facts, derived works that simply use the facts wouldn’t be derived works, because the facts they use would not be copyrightable.

Of course using the facts in the work is subtlety different than using the work itself. You might argue that this is fair use.