If you think that nuclear power has a fair amount of tail risk (very small probabilities of catastrophic accidents) then any safety considerations based on hist

If you think that nuclear power has a fair amount of tail risk (very small probabilities of catastrophic accidents) then any safety considerations based on historical data is meaningless. In that case, most of the deaths & cost that occurred in the past would come from the single most sever accident (and this seems to be the case [1]) Moreover, the worse accident (Chernobyl) is not a good guide of how bad things can go: if the probability distribution of number of deaths per accident is fat tailed, then if an accident occurs that is worse than Chernobyl it will probably be much worse. [1] http://en.wikipedia.org/wiki/Nuclear_and_radiation_accidents http://en.wikipedia.org/wiki/Nuclear_and_radiation_accidents