They are "safe" in the sense that the probability of a catastrophe is 1 in 10,000 years (that is the statistic).
But they are exponential orders of magnitude more deadly, with the consequences thereof lasting for tens of thousands of years, when that 1/10,000th year comes around.
With 500 nuclear power plants operating, we can expect a serious nuclear catastrophe every 20 years.
These numbers are based just on nuclear power plants and exclude nuclear submarines, warheads etc.
Deaths per TWH is no real measure of safety at all when it comes to nuclear. Accidents affect not only human beings but the environment and indeed entire continents and the ecosystems thereof. Even today when they hunt wild boar in Germany many of the boar are found to have been irradiated as a result of Chernobyl, and hazardous for consumption. These events cross borders. It would be foolish to attempt to comprehend their relative significance in terms of death counts.
Applying deaths per TWH to nuclear is naive. It's an abstraction that in no way reflects the underlying reality, and which leads the general populace into mad ambivalence. It's the kind of rhetoric you hear on Animal Farm. This idea that some deaths are better than others. In fact, it's a question of asymmetric risk with outsized consequences. If you learn anything from Ben Graham or Warren Buffett, it's that that's not the kind of game you want to be playing, unless you're really smart like LTCM. You don't want to have anything to do with it, especially not if there is human mental apparatus, governments, and contractors involved. There are too many variables you cannot control, and they are not captured by deaths per TWH.
The second thing to learn from Graham and Doddsville, when dealing with asymmetric odds is that history is no reliable guide to the future. What happened last year or the year before, is no indication of what may happen next year or the manner in which we can expect it to happen. That nuclear catastrophes have historically not been catastrophic (according to the author), is no reason for the author to believe that they may continue to remain so.
Yet the Washington Post article has based its argument on the premise that you can even measure deaths per TWH. You cannot count deaths directly and indirectly related to a nuclear accident. We have learned that much from the history of these accidents, and it would be ignorant and insensitive to claim otherwise. Take for instance Chernobyl, does the author truly believe that only 50 people have already died as a direct and indirect result? Even the most conservative numbers reported (deaths actually incurred) are higher, and show significant variance between sources:
"Estimates of the number of deaths potentially resulting from the accident vary enormously; the World Health Organization (WHO) suggest it could reach 4,000 while a Greenpeace report puts this figure at 200,000 or more. A UNSCEAR report places the total deaths from radiation at 64 as of 2008." - http://en.wikipedia.org/wiki/Chernobyl_disaster
From those to whom much as been given, much will be expected.
They are "safe" in the sense that the probability of a catastrophe is 1 in 10,000 years (that is the statistic).
But they are exponential orders of magnitude more deadly, with the consequences thereof lasting for tens of thousands of years, when that 1/10,000th year comes around.
With 500 nuclear power plants operating, we can expect a serious nuclear catastrophe every 20 years.
These numbers are based just on nuclear power plants and exclude nuclear submarines, warheads etc.