If we want to avoid conceding that the [______] of our knowledge is much more limited than usually supposed, there must be some difference between the probabilistic evidence we have about the lottery and evidence for regular things that we do know.