“I always say expected value is the probability of success times the degree of impact. That math works easily, but people are very uncomfortable in the low-probability domains,” said Vinod Khosla,”