One should be suspicious that my prior probability assessment is a good one if I haven't accounted for all the probability mass, but the mechanisms still work.

Which is one of the other problems I was getting at