"How not to estimate the likelihood of nuclear war." This is worth passing along as I've been having similar thoughts as to how not to estimate the likelihood of nuclear war. You have the problem that there are no previous instances of global thermonuclear war, so if you try to use a "frequentist" approach, you end up dividing by zero (mentally or literally). So they talk about the Bayesian approach, but in my mind that doesn't help much as you, again, have literally zero previous instances to look to for insight, and you end up with your "priors" being based on "gut feeling" emotion. I'm glad they titled this "How not to estimate the likelihood of nuclear war" rather than "How to estimate the likelihood of nuclear war" because it doesn't say how to do that.

How not to estimate the likelihood of nuclear war

#futurology #superforecasting #nuclearwar