I remember going to the Centre for Alternative Technology a year or so ago. There was a display about energy technologies.Energy technologies were assessed according to diffferent criteria from green (meaning good) through amber and red (bad) to black (very bad). Nuclear energy scored quite highly on all the criteria except one: risk where it got black (very bad).
Now of course risks from nuclear energy are of various types, including risks that are quite well specified because we have good knowledge of them. I'll split the safety issues in two: those associated with the direct design of the reactor and operation under normal circumstances and those through other matters. Here I will deal with only the risks of the direct operation of the reactor.
I am surprised at how little design features of nuclear reactors are discussed. Unlike the Soviet RBMK design used at Chernobyl, nuclear reactors built in the West over the last four decades typically enjoy a 'negative void coefficient', meaning that power would fall in the event of a void in the coolant, due for example, to a loss of pressure, a negative feedback loop which leads to the reactor shutting down. Some new reactor designs (e.g. the Westinghouse AP 1000) also employ 'passive safety' features, meaning that failure of safety systems leads to reactor shut-down through 'passive' processes such as gravity, rather than relying on the defence-in-depth of multiple parallel safety systems. Redundantly, Western nuclear power stations are also typically built with a concrete containment dome.
So in this case, nuclear reactors have a very low risk. How low? Well risk assessment can only by definition assess those risks that can be assesed, but nuclear reactor designs should blow up only one in every 10 million reactor-years, and even then they should be fully contained. So there is a 1/10000000 chance of a contained nuclear reactor accident. This appears absurdly low: even if the world built 1000 nuclear reactors, we would only expect a contained accident every 10,000 years or so. If only climate change had such low risks of catastrophic outcomes!
Of course there may be other risks and environmental damages associated with nuclear power, and I'll try to deal with all these some other time.
I have also been thinking about the banking system and wondering if similar design features should be applied to it! Can we get 'passive safety' into the financial system. Might be worth a try!