After spending a year in Baghdad, one of the most common subjects that comes up with other political analysts is that the closer you get to the issue, the more you know that you don’t know anything. This book is about how one can take some simple numeric shortcuts to arrive at the same conclusion, and make predictions vastly more accurate than typical individuals, aggregated individuals, or even us analysts.
This book lays out a model of how to make accurate forecasts, then shows the evidence for why these forecasts work. At it’s core, it’s a book about numeracy, but does so with enough detail to be useful. It covers Brier scores, bayesian analysis, regression to the mean, and other mathematical tricks that can be applied to decidedly non-mathmatical situations. As a side-note, it casually destroys the intelligence analysis in the runup to the Iraq war and dismisses most experts as no better than dart throwing chimps.
I dig it.