Abstract:
|
This work models the time to a major nuclear war and winter leading to the deaths of 98 percent of humanity. It estimates an 11 percent chance of such an event in the next 40 years. This is based on two components: First, it assumes that there have been only two major nuclear crises worth considering: The 1962 Cuban Missile Crisis and the 1983 Soviet nuclear false alarm incident. This gives one observation of 21 years and another censored at 36 years of the distribution of time between such incidents. This leads to an MLE of 57 years for mean time between such incidents assuming a constant exponential distribution. Second, it assumes a beta distribution with parameters 8 and 10 for the probabilities that such incidents would trigger a nuclear war and winter. This beta distribution provides an 80% equal tail confidence interval of (0.3, 0.6) for such probabilities. These numbers seem consistent with the published literature on those two incidents. Details are provided in the Wikiversity article on “Time to extinction of civilization” and the vignette on “Time to nuclear Armageddon” in the development version of the Ecfun package on R-Forge.
|