Book review: The Signal and the Noise by Nate Silver

The Signal And The Noise

by Nate Silver

Penguin, £8.99

The Signal And The Noise was published in America just before that election and became a bestseller overnight, but Silver was already an established statistician, having successfully used mathematical analysis in baseball and gambling.

This book is an astute and modest look at “the art and science of prediction”, the author showing intelligence and awareness of the limits of statistical analysis.

The first half looks at several case studies of bad prediction, taking in the global financial collapse, politics, baseball and more complex systems such as weather. According to Silver, the key to these bad predictions is human arrogance about how much we can know, combined with ignorance of what the numbers do or don’t tell us. Misinterpretation is a repeated failing of professional pundits and commentators, and Silver does a good job of sorting the wheat from the chaff.

So how can we improve predictions? Silver suggests rigorous application of Bayes’ Theorem, which gives a formula for predicting conditional probability – that is, the chances that something will happen given that something else has already happened. It’s obviously a tad more complex when you get deep into applying it to real life situations, but Silver convincingly argues that a proper understanding of its implications is the way forward.

Silver applies Bayes’ Theorem to simple models first – sports and games with well-defined rules – then moves on to tackling dynamic environments such as global warming and terrorism.

Throughout it all, Silver is thorough and careful, as any good statistician should be, making this an expert study of what we can and can’t predict in the future.

Doug Johnstone

• Nate Silver is at the Edinburgh International Book Festival on 13 August