Wasn't the idea that he integrated on n for 100 to infinity, to arrive at that probability? So the probably should be interpreted as the probability of an eventual counter-example.