mattnewport comments on Expecting Short Inferential Distances