Archive for the ‘James Blodgett on Risk Calculations’ Category

James Blodgett on Risks

Saturday, May 24th, 2008

James Blodgett has a masters degree is statisitcs and leads the Mensa Special Interest Group Global Risk Reduction
(Orginally posted at LHC Concerns)

Why are physicists building a collider if it is dangerous? Why are they not worried? I have talked to a bunch of physicists. I will tell you my model.

At first, physicists had good reason to think that there was no problem. When Walter Wagner suggested in 1999 that black holes might be created, fairly simple math (even Wagner’s own math) showed that the energy required would be beyond the reach of any collider. The collider of interest at the time was the Relativistic Heavy Ion Collider at Brookhaven, which started up in 2000. Brookhaven produced a safety paper that listed several safety factors. To this day, physicists still cite that paper to show that there is no problem.

The problem was, subsequent physics papers eroded several of the safety factors. String theorists developed a new theory that showed how colliders could create black holes. The theory behind Hawking radiation, supposed to dissipate black holes, was found to have theoretical problems. And so forth. It is somewhat like a frog in water that is slowly heated. The frog will not notice the heat.

Part of the problem is that physicists are not risk analysts. They feel, correctly, that the theories that permit trouble are a small subset of all theories, so that trouble is not likely. As one physicist said to me “You do not understand probability. A probability of 0.001 means it is not going to happen.” Well, I have a master’s degree in statistics. I do understand probability. What the physicist did not understand is that 0.001 times the value of Earth gives a tremendous negative expected value. We should NOT be taking chances like that. A shuttle astronaut faces similar personal risks at every launch. If physicists where shuttle astronauts, we would applaud their confidence in their craft and their willingness to back up their confidence by risking their lives. However, a shuttle astronaut is a volunteer. A commercial airplane with a 0.001 risk would not be allowed to fly. A commercial airplane carries many passengers who did not volunteer for a risky mission. The Earth has 6.5 billion passengers, so we should hope for a risk to Earth that is less than that of a commercial airplane.

I once interviewed a physicist who had written an article for a popular publication telling us not to worry. His reasons were not clear from the article, so I asked him over lunch. It was a stimulating conversation, he seemed to be a nice guy, and he paid for lunch. His reason turned out to be that he did not believe in any kind of black hole. As he said, “When an equation goes to infinity, that is a sign that there is something wrong with the equation.” He has a point, and he might just be right. But that is not what most physicists believe. I applaud him for his theoretical creativity. I understand why he might not be personally concerned. But he doesn’t quite realize that he is betting the Earth, and telling others it is okay to bet the Earth, based on a theory that many would question.

Scientists are used to testing theories carefully. In order to publish, data need to be statistically significant, and they need to pass peer review. They tend to think like this about risk too, wanting it to be proven before they will believe it. (Consider, “The risk is only 50%, that is not proven, so we will launch.”) To address this, risk specialists have developed what they call the precautionary principle. The precautionary principle reverses the burden of proof in areas of scientific risk. Precaution does not need to prove risks to a high degree of confidence. Rather, the burden of proof is on those who propose risky experiments, to show that they are safe. The precautionary principle has been touted by many risk specialists, and has been formally accepted by the European Union, the home of CERN. Unfortunately, many scientists are not used to this type of thinking, are not aware of the precautionary principle, or if aware do not like it. Even those who accept it are sometimes reluctant to apply it. Even though the precautionary principle has been formally adopted by the European Union, EU functionaries are currently passing the buck as to who is responsible for its implementation.

The Large Hadron Collider at CERN is Europe’s moon shot. They have spent billions. It will bring them back to the preeminence in physics they enjoyed in 1910, when Einstein and Bohr and Heisenberg were all Europeans. Also, a sizable portion of the world’s physicists are working on the LHC, or on theories that will be impacted by the LHC. They are reluctant to hear that it might have a problem. If that problem is a low probability risk, that is a risk they are willing to take, or rather willing to ignore.

I should add that estimates of components of collider risk are quite subjective and quite variable. In 2004, I tried a series of Delphi questionnaires in which I asked physicists their estimates of several components of collider risk. As an example of the variability, estimates that Hawking radiation would fail ranged from 0% to 50%. The data are as follows: 0, 0, 1E-10, 0.001, 0.01, 0.01, 0.01, 0.02, 0.02, 0.07, 0.1, 0.1, 0.3, 0.35, 0.5. This was at the time that CERN was relying on Hawking radiation, before we were aware of the papers questioning its theoretical background. I guess that estimates would be higher now. (Note that even then, when Hawking radiation was relied upon by CERN, most physicists estimated a non- zero probability that it would fail.) Also, ten of these physicists were aware of my interest in the collider issue from the Delphi instructions. A few may have been shading their estimates to game the results. CERN’s Chief Scientific Officer, Jos Engelen, was quoted in the New Yorker as instructing CERN scientists to say that the risk (the overall risk of the LHC) is zero.

The model of risk in the Delphi questionnaires was [ Total risk ] = [ Risk that black holes would be created ] x [ Risk that Hawking radiation would fail ] x [ Risk that accretion would be rapid ] x [ Any other risk factor they propose ]. (These probabilities were stated as successively conditional, given that the earlier ones were true.) Ten of the physicists received these questionnaires, the others received a preliminary questionnaire. The fact that none proposed a fourth factor at least suggests that this model is plausible. I would consider the collider/cosmic ray analogy as a separate consideration that would limit the risk calculated by this model, but that should not be calculated into it.