Extract

Forecasting the future is especially difficult when predicting rare events such as volcanic eruptions, earthquakes, nuclear weapons acquisition, or war. In this part of the book, we examine methods for making such predictions. If a predictor variable rarely changes much, and tends to persist, then it can be used to predict something that otherwise might be quite surprising. Tago and Singer (Chapter 8) use this insight to predict nuclear proliferation. For example, rivalries between sovereign states (such as the Franco-German rivalry of 1870–1945, the US–Soviet rivalry of the Cold War, or the Indo-Pakistani rivalry since their independence in 1947) tend to be very persistent, and to be strong predictors of nuclear weapons proliferation. Tago and Singer show that such a focus can successfully predict nuclear weapons proliferation in the past, and yield forecasts about likely future nuclear programs. Their forecasts, for developments two decades out, will gradually be verified, or not, with the benefit of hindsight. The Israeli destruction of what appears to have been a Syrian nuclear facility in September 2007, for instance, is indicative of a hotspot that could be predicted by Tago and Singer from the model in their chapter (based on Syria’s rivalry with Israel, etc.).