I don't think this is a bad idea, but I do think the time is not right yet. Of our current open questions very few look like the question is really open in an interesting sense. I'm not even convinced the unanswered questions are necessarily a good source for such questions.
This is actually something we have been thinking of doing on CSTheory, now that we have a community blog through which we can coordinate efforts. However, finding suitable questions seems non-trivial, as you need a few things: a question which might actually be tractable, an approach which seems worthwhile, a way of dividing the problem into bite-sized chunks, etc. It seems like we might have a few suitable questions from the unanswered section of cstheory, but by now there are well over 300 questions there.
In a sense a polymath project is only interesting if you are proving something new and interesting, where the end goal is a paper, but not so much when you are trying to figure out what was meant by the question, or know the answer is already known, or when the question is simply not interesting.
With that in mind, I would suggest we park the idea until the site has grown some more. I do like the idea, but I don't think it is really realistic at our current stage of development.
This post has been migrated from (A51.SE)