1. During his talk, he mentioned that nuclear plants are designed to have a very low probability of Chernobyl-scale failure, and that the current rate of Chernobyl-scale failure given the number of nuclear plants in production is roughly 1 every 20 years.
2. He eventually concluded his talk by saying that we should double the number of nuclear plants in production around the world to remove our dependence on coal.
I asked a question at the end of his talk, given point 1. then point 2. would mean 1 Chernobyl every 10 years. He was completely dumbfounded. He had never combined his ideas with simple probability theory. One physics student present then said angrily: "Yes, but we will get better at building nuclear plants."