Using near misses to gauge risk

[campbell]

http://online.wsj.com/article/SB10001424052748704312104575298820977456114.html?m od=djemnumbers_t

Near misses can be studied by statisticians to estimate the probability of an event that hasn't occurred before. Estimating the probability of unlikely disasters has become standard practice for nuclear and space regulators. Such an exercise, experts say, could help companies involved in deep-sea drilling evaluate risks and possibly prevent catastrophes like the Gulf oil spill.


http://si.wsj.net/public/resources/images/NA-BG445A_NUMBG_NS_20100611232002.jpg

[E]

I like that concept. It also address one of the problems in your Posner article: not getting credit for avoided disasters.

I guess it takes a framework that would let you credibly say that Event Y would have happened if we had not taken Action X.

Then, you could count Event Y, which never happened, in your estimate of the probability of it happening again.

Or something like that.

[campbell]

Yes. Just because something is difficult to quantify/evaluate doesn't mean it shouldn't be done.

For operational risks, this sort of approach is helpful.

[Heathen]

Yes. Just because something is difficult to quantify/evaluate doesn't mean it shouldn't be done.

For operational risks, this sort of approach is helpful.

Is this like the time I almost married a stripper?