Why is rejection of mathematical expectation an unworkable solution?

Well, rejection's not a solution per se until you pick something justifiable to replace it with.

I'd be interested in a top-level post on the subject.