Copyright (in the US, at least) is a “natural” right. It exists, as soon as the work exists, and registration is not required. It needs to be explicitly transferred and/or assigned. There is no need to "defend" it.
There have been a lot of challenges, based on “derivative copies,” but these generally seem to be from where someone uses a photo done by someone else, in their own creative work (like the Obama "Change" poster). Some of these challenges succeed, some do not. I would assume someone extending these weights might be considered a “derivative.”
Copyright law is odd, and there’s a lot of “fuzziness,” especially with creative works.
I’ve always been a bit skeptical of applying copyright to compiled and opaque binaries, but I guess I’m in the minority.
>I’ve always been a bit skeptical of applying copyright to compiled and opaque binaries, but I guess I’m in the minority.
You're not in the minority - copyright requiring human authorship and creativity is standing law. A compiled binary cannot be copyrighted on it's own[0]; the copyright flows from the source to the binaries via derivative works. We just don't normally think about this because until very recently all binaries were derivatives of copyrightable human creative expression. Applying a mechanical process to a creative work doesn't make it non-creative, after all.
[0] Hand-assembled binaries would be considered "source code" in this case.
There have been a lot of challenges, based on “derivative copies,” but these generally seem to be from where someone uses a photo done by someone else, in their own creative work (like the Obama "Change" poster). Some of these challenges succeed, some do not. I would assume someone extending these weights might be considered a “derivative.”
Copyright law is odd, and there’s a lot of “fuzziness,” especially with creative works.
I’ve always been a bit skeptical of applying copyright to compiled and opaque binaries, but I guess I’m in the minority.