I recently finished Nate Silver’s The Signal and the Noise, and frankly I’m disappointed. There were surprisingly few insights to be found in the 450-odd pages; I only highlighted two passages – an all-time low – and remarkable as I like to read with a highlighter in hand.
On the positive side the writing is clear, and the few technical discussions are easy to follow. Silver includes some interesting stories as he covers the history and reliability of predictions in finance, sports, politics, weather, gambling and economics. The general conclusion is that the predictions are not very accurate. The one optimistic tidbit is that we seem to be getting better at predicting the weather.
His main advice appears to be: Practice Bayesian Thinking [this showed up in Nassim Taleb’s Fooled by Randomness also]. I find it hard to really understand what that means in practice – and when I do look into it I tend to find myself on the side of the frequentists.
At best I’ll chalk it up to unmet high expectations – I was familiar with his outstanding reputation with predictions in Baseball and Politics. Or maybe it’s because I read this just after reading Fooled by Randomness and found that to be very thought-provoking (although not as enjoyable to read). Perhaps it’s because Silver is just so generally sensible and doesn’t throw out any controversial opinions or perspectives. Either way I find myself unsatisfied.