"History suggests that nuclear power rarely kills and causes little illness. That’s also the conclusion engineers reach when they model scenarios for thousands of potential accidents."
Auditing nuclear power's extremely short history is not a valid method of assessing risk. In poker they call this 'results oriented thinking'. Especially when discussing black swan events, history is almost by definition not a very useful basis for predictions.
"The history is not 'extremely short'. Nuclear has been in operation for decades."
In the context of highly unlikely events and the mathematics of probability, a period of decades is an extremely small sample. Consider that some of the types of events we are discussing may only happen once in a century. Clearly in this case several decades is not enough to fully gauge what the true risks of nuclear are to humans in the long term. One worst case scenario disaster would tip the scales quite heavily.
Auditing nuclear power's extremely short history is not a valid method of assessing risk. In poker they call this 'results oriented thinking'. Especially when discussing black swan events, history is almost by definition not a very useful basis for predictions.