Sounds reasonable. Who's gonna do an implementation? There's Greg Wilson's version, and there's an alternative by Aric Coady email@example.com that could be used as a comparison.
I'm gonna give it a try, unless somebody more qualified volunteers -- Greg's version's in nondist/sandbox/sets, right? Where's Aric's?
What should I do with the modified set.py -- submit it as a patch, or ... ?
I forget -- do you have SF commit permission? If so, feel free to add a competing version to the sandbox. Otherwise, a SF submission would be good (and post a link to python-dev when you upload it).
--Guido van Rossum (home page: http://www.python.org/%7Eguido/)