Risk assessment for nuclear plants

Lots of questions now being asked about location of nuclear power plants near fault lines.  Today's NYT has a nice piece on the overall difficulties of risk assessment for nuclear power plants in general and the consequences of the recent earthquake/tsunami in Northern Japan.  Now we know that the Fukushima Daiichi plant was designed to withstand large quakes, but not for 9.0 on the Richter scale followed by a tsunami. 

A nuclear power plant is supposed to last a long time, so one would need to design it to withstand not just an average earthquake, but a much-worse-than-average one.  The tough question: how much worse?  Do we go to the worst quake one would expect with a 10% p-value, or 5% or 1%. 

Another tough question: who makes this decision?  Do we trust the power plant owners, or do we need regulations?  And how can we trust the regulators?  (The Securities and Exchange Commission was supposed to watch the financial markets and we know how well that worked.)