The Signal and the Noise by Nate Silver

I picked up The Signal and The Noise from the dreaded 'bestseller'/management section in my local bookshop for no better reason than that I had heard Nate Silver 's name after the last couple of US elections. Silver was using Bayesian statistics to predict election outcomes quite successfully, and did a good job of showing how completely ridiculous the TV pundits tend to be. I didn't know what to expect, but though why not. I like signals, I am fascinated by noise, and Bayesian statistics is always fun.

Well, The Signal and the Noise was kind of interesting to read. Silver makes a case for statistical reasoning by applying it to many different situations. From Poker (what got him into statistics), via election forecasting ( Silver's claim to fame), to climate change (an important topic). The chapters are built around individual applications with Silver explaining the challenge, interviewing experts, and how to apply statistical modelling.

Nate Silver makes a strong case for Bayesian modelling, if at times perhaps a bit too direct. This is a very much 'what works' approach, and the tone can sometimes be dismissive. On the other hand he is used to being right Now, at the same time I got the feeling that he is trying very hard to be politically neutral and not draw conclusions from his models. That felt more like a strategic choice in the US' very polarized political situation. It is like the editor was afraid that it stepping on toes might hurt the sales. A bit of a shame. That is perhaps my only real critique. Not a bad read over all as I recall, and I think it can get a few people interested in modeling which is always good.