His theories are largely a rejection of placing too much faith in the scientific method (without rejecting it entirely, of course). The essence of his entire philosophy is that naive empiricism causes you to ignore things that happen rarely, and the more you build the assumption that rare events do not happen into your models, the more it hurts when they do happen.

This is, by definition, a hypothesis that is both obviously true and also almost impossible to test. The very problem in question is that there is a fundamental lack of data on these rare events because they happen rarely.