That just sounds like debt to me, and that kind of cruft I don't think reflects well on the project, as then the copyright notices would be completely inaccurate (not just mostly inaccurate). Let's make a decision and do it (either ensure copyright notices are coming in appropriately, or get rid of them).
Making them completely accurate is an impossible task. The best you can get out of humans is the "sort-of-accurate" state that we maintain now. Is that really of sufficient value to spend the cycles paying attention to them?
Honestly, generating these patches to remove them can be nearly automated, and over the course of a week or two could all be moved through. Then we can lock it down with a hacking rule, and never have to deal with it again.
Yep, the patchsets would be *large*, but trivial to produce, trivial to review, and harmless to the code. - gabriel