Thursday, March 17, 2011

Nassim Taleb has mean, mildly idiotic thing to say about the Japanese Nuclear Commission


In this short note Nassim Taleb mocks the risk assessment ability of the Japanese Nuclear Safety Commission, and in particular their stated goals
The mean value of acute fatality risk by radiation exposure resultant from an accident of a nuclear installation to individuals of the public, who live in the vicinity of the site boundary of the nuclear installation, should not exceed the probability of about 1x10^6 per year.
My emphasis. Taleb notes:
That policy was designed only 8 years ago. Their one in a million-year accident occurred about 8 year later.
Taleb's emphasis. The philosopher then regurgitates a tired old argument that convexity always leads to bias in probabilistic assessment; and that this bias is always in the same direction. This is why you can bet on long-shots all day long at the races and make lots of money (what are on Earth are you waiting for?).

It is just possible, of course, that Taleb has confused "mean" acute fatality risk with the probability of at least one fatality. The latter seems inevitable, unfortunately, given the heroic efforts of the engineers but here's hoping not everyone in the vicinity dies. The goals of the commission may well have been optimistic but it is not clear, even in retrospect, that they were hopelessly wildly optimistic; subject to fundamental limitations regarding small probabilities; or could have been better communicated without numbers. Nor is it obvious that the plant posed a bigger fatality risk to neighbours than, say, aggressive taxi drivers. What is obvious is that no-where does the commission speak about one in a million year accidents, instead quoting figures closer to one in ten thousand.

A one in 10,000 event probability is aggressive as well, from my amateur standpoint, but divide by 500 reactors across the world (to round) and you are down to 1 in 200. Go by decades and it is 1 in 20. There is plenty of room for sensible, specific debate here, once we filter out the Black Swan God-of-the-gaps fanatics, and one might start with the words of the Japanese Commission itself:
The possibility of health damage to the public by emission of radiation or release of radioactive materials accompanied with activities for utilization of nuclear energy should not meaningfully increase the risk of damage to the public’s health in daily life.
Will the coverage center on environmental, medical or nuclear experts? A few years ago you would have presumed as much but due to the rise of the Extremist cult few details seem important. The favorite now is a sorry but predictable parade and of course the media have fawned over this "generalized expert in uncertainty". We anticipate journalists exhibiting "mastery" of the Black Swan "concept" and a deluge of statistical profundity. We shall not be surprised by "abstract" points about rare events (that once seemed banal); references to the financial risk debate (which has oh so much to do with meltdowns) and a lifetime's supply of populist, pseudo-probabilistic piffle.

And this will happen with little help from Taleb, who has already said "everything he has to say" about the Japanese Earthquake. On that, at least, we can agree.