24 Following
Brick

Brick

The Signal and the Noise: Why So Many Predictions Fail - But Some Don't - Nate Silver An excellent read, with very detailed insights into areas in which Silver has a wealth of personal experience, including limit Texas Hold'em, baseball statistics used in forecasting player performance, and political polling, combined with insights derived from extensive interviews with top practitioners in many, many other diverse areas. Although I worked my way through more details on professional gambling than I would have thought I could handle, similarly baseball performance forecasting, the details left me feeling that for the first time I had an insight into how these areas work for the practitioners.

Finally, I enjoyed Nate Silver's discussion of the impact of using Bayesian methods for forecasting, and although the critiques of his presentation may be on target with regard to formal mathematical exposition, Nate shows how these methods have risen to the challenge of providing insight and guidance to decision making under uncertainty, where Frequentist practitioners have so often explained that they could give no opinion in the absence of data, i.e., "I just can not venture an opinion, when we have not yet seen a statistical sample of this phenomenon" kind of sad lack of advice to counter-terror efforts, earthquake preparedness, with remarkably poor results in Fukishima and on 9/11.