I called this the "patch hole" in the law over 20 years ago, in analogy to the Unix patch tool, and I still don't think I've seen a legal system cleanly grapple with what it means to be able to distribute modifications on their own. It is a qualitatively different process and result than having to distribute modified versions directly. It rather directly faces a deontology vs. consequentialism question... if two different processes (distributing changed contents directly vs. distributing patches to an end user and providing combination software) produce the exact same end results, on what basis will the law decide that one is OK and the other is not, since allowing people to distribute change lists provides an effective 100% workaround to the existing illegality of distributing changed works?
I'm not trying to offer an answer here, I'm merely trying to establish that talking about "books" isn't going to give a good grasp on this situation. The metaphor simply doesn't work. It's far too simple to capture too many relevant issues.
You're going about the question backwards.
If we were basing things on reasonable consequentialism, then downloading a modified version of something you already legally possess would not be a copyright violation.
However, something that it's own generic work or not copyrightable (like a block list) would then not be considered a derivative work.
Patching pirated work is still piracy, but patching work you have purchased legitimately (or were given by the copyright holder, such as a web page) is not piracy.
That's a fair point but copyright law was designed with "books" in mind so everything is ultimately related to that. Computers have really put a wrench in what was otherwise a pretty simple system.