Nate Silver 2012. The Signal and the Noise. The art and science of prediction. London: Penguin. 534 pp.

Not compelling throughout, but very interesting in parts. This is a popular account of reasoning (prediction) based on Bayesian statistics. The first few chapters explore subjects where most of us have great difficulty extracting useful predictions (signal) from the vast amounts of data available (noise). Such subjects include sports statistics (baseball), weather and climate, and earthquake prediction.

The second half of the book explains and applies Bayes’ Theorem to these problems. I think it was only by reading The Signal and the Noise that I came to understand Bayesian reasoning and it’s wide applications. It is still salutory that such a simple idea works as well as it does (modifying a prior probability based on better information as it comes to hand, then repeat as often as possible, each time with slightly improved probability). It works even if the prior probability is a poorly-informed guess. And even if we don’t ever come to understand the underlying cause of the phenomenon (as with earthquakes).

One of the telling conclusions was that people’s surveyed predictions about sharemarket outlook and economic confidence generally are highly inaccurate. You are much better off doing the reverse of what “market confidence” would suggest. A Nature article on a related theme is also of interest.

[incomplete, more to come]