we now have a number of different measures that appear to give different but reasonable predictions, with no obvious way to choose between them. If the probabilities we predict depend on the measure we assume, and we can assume a measure giving almost any answer we want, then we really haven’t predicted anything at all.
Measure problem
