Gregory Maxwell wrote:

With text it's even worse because our public editing process make it much easier for someone to prove that our text was a derived work, where in a more traditional medium a sufficient amount of refactoring would usually manage to hide the violation.

With images I plan on just replacing all the ones with suspect copyright (i.e. everything that isn't CC* or GFDL and uploaded by the author, or with an actual letter attached that explicitly says PD or an acceptable license) over time... but I have no idea how to solve text.

Indeed, I'm rather surprised at the glib and easy way with which some people suggest "refactoring" copyright violations rather than deleting them when possible. Under normal circumstances, every revision of an article is a derivative work of previous revisions going back to the beginning. Simply rephrasing to avoid identical strings of words is not sufficient to avoid this.

To be confident that you don't have a derivative work, you would need to remove all content traceable to the copyright infringement, in much the same way that some editors rewrite an article entirely from scratch if they find it unsatisfactory. Using the infringing content as a base to work from is risky at best. When we need to preserve history that predates the violation, I can understand the dilemma that forces us to bury these problems instead of excising them, but otherwise it's a shoddy practice. Images are rather different, because the revisions can be entirely different, merely taking advantage of the same filename, and all traces of a copyright violation are easier to remove.

Exactly what qualifies as a derivative work is not always clear, and as best I'm aware there is effectively an open conflict on the issue between the rulings of two circuits of the U.S. Court of Appeals. So better guidance would only be available with a Supreme Court decision. I rather suspect that under the right circumstances, a Wikipedia article might provide them with an excellent set of facts on which to issue such a ruling. However, the expense and distraction of litigation being what it is, I'm not saying that setting legal precedent this way is something we should aspire to. Also, once faced directly with the prospect of an unfavorable outcome, people often discover that they're actually _more_ comfortable with uncertainty.

--Michael Snow