[ExI] How dangerous is radiation?

Anders Sandberg anders at aleph.se
Sun Jul 6 10:20:19 UTC 2014

There are *lots* of potential confounders in Johns examples - the Rocky Mountains may have more radiation but might have less mold toxins. Radiologist is a fairly high prestige occupation, lowering cancer rate because of social status. Shipworkers on nuclear ships no doubt both got training and likely were a selected group. Were the people in the Taiwanese buildings average people or different? And so on. Radiation epidemiology is pretty messy, because these complications show up all over the place. 
Linear no threshold models are practical because they are simple: just try to minimize the stuff, assume large reductions are better than small reductions. But yes, people are too scared of anything with any radiation. 
Hormesis, that small radiation doses might be helpful in killing off precancerous cells, is a real thing. But you don't want to rely on it, since there could well be big individual variations in ideal exposure, easily swamped by individual exposure variation. 
When Monte Carlo-simulating insurance risk, one typically use a hazard module that generate random disasters with severity and location set by observed probabilities (and various assumptions). This random disaster is sent to the damage module, which calculates the damage a particular building takes from it: this is a random outcome determined by a "vulnerability function" which is actually a probability distribution dependent on how severe the disaster is and what kind of building it is. When you look at actual data for windstorms and many other perils it is incredibly noisy: the amount of damage is more or less evenly distributed between 0 and some upper limit that tends to curve up as a parabola until you reach saturation. I expect the same thing for radiation risk: the hazard (radiation exposure) is converted into a random outcome (health) in a fairly nonlinear way, and even people with big exposures can come out scotfree. Meanwhile, just as in insurance, the actual risk is a sum of a lot of sources/perils. So you have to combine the risks of asbestos, air pollution, lifestyle and whatnot to get a proper cancer risk - not very useful in the individual case, but averaged across a population it can help set policy. 

Anders Sandberg, Future of Humanity Institute Philosophy Faculty of Oxford University
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20140706/0da76f93/attachment.html>

More information about the extropy-chat mailing list