How does this work for literature? I see where "free" benefits those creating derivatives and readers, I don't see how it would be viable for authors, except, perhaps, with a ransom/patronage model.
Infrastructure-wise, existing systems that work for software (e.g., github, bitbucket) are already optimized for text-based content and would clearly work well for literature (in fact, they are already used widely for documentation accompanying software projects, which is [technical, nonfiction] literature.)
People could fork, say, a short story, rewrite parts, and submit a pull request; the original author would maintain control in this case. If the original author didn't like the change but someone else did, that person could fork the forked version (the one containing the change), add additional changes, then submit that as a pull request, etc.
It could get messy, but it also sounds like it could be really interesting, with the end result being a tree of competing stories.
Maybe somebody could submit a short story, publicize its availability, and see what happens.
-bowerbird