Friday, April 24, 2015

The Signal and the Noise, by Nate Silver

Staff Review by Chris Saliba

Nate Silver’s much feted book calls for a common sense approach when analysing the huge amounts of data the information age generates.

Nate Silver is a statistician and writer who has had quite a curious career. After receiving a Bachelor of Arts degree in economics he went to work as a consultant for KPMG. Obsessed with baseball statistics, he developed a system for predicting player performance. He lasted four unhappy years at KPMG before quitting for a career as an online poker player, earning some $400,000 in winnings. This line of work came to an end when legislation was introduced to rein in online poker businesses. It was probably a good thing anyway, because Silver was starting to lose substantial amounts of money. His system could only work so far; sheer uncertainty in the hands poker dealt meant it wasn’t a game he could win consistently. Around this time, due to the online poker legislation closing down sites, he took an interest in politics and started predicting election outcomes. Success in this line of prediction led him to being a much sought after commentator.

The Signal and the Noise constitutes the sum of all Nate Silver’s thinking on the subject of prediction. The book is a bit hard to pigeonhole as it freewheels over a wide range of seemingly unrelated topics. Silver’s statistical approach is applied to everything from baseball and poker to climate change and bird flu. It can make your head spin a bit as a geeky obsession with complex data is used to tackle both serious and frivolous problems. You wonder if you should be taking advice on global warming from a poker champ.

The basic conceit of the book is a pretty common sense one: we live in an age that spews out enormous amounts of data. Most of this data is simply noise. What analysts must do is look underneath all the noise for the signal, the more meaningful story the data is trying to tell us. A common mistake amongst analysts it to mistake signal for noise. Too often the temptation is to read the data to suit our own biases and prejudices. According to Nate Silver’s research, the worst offenders are political and economic forecasters. He shows that they have an absolutely disastrous record when it comes to prediction. Meteorology comes off as the most accurate predictive science, most likely because it is subjected to public opinion everyday. As for the science of climate warming, Silver gives good marks to climatologists in predicting how much the weather will increase, with a few minor caveats.

What Silver is really arguing for in The Signal and the Noise is a bit more common sense, and less ego, when it comes to prediction. He says that we can’t rely purely on data, and there comes a time when we must use our unique human skills in weighing a problem and guessing for probability. In many ways, Silver’s book is similar in theme to Daniel Kahneman’s Thinking Fast and Slow and Nassim Nicholas Taleb’s Antifragile. It cautions against hubris and advises there are stark limits to what we can predict. The prose is a bit dull, however, and the organisation messy. In fact, the book doesn’t really hold together as a philosophical whole. It’s more like a series of riffs by an extremely clever young man. It probably could have been a lot shorter and more focused.

Despite these caveats, The Signal and the Noise offers a lot of food for thought. Maybe all economists and political commentators should be made to read it. The section in which Silver interviews a highly ranked bank economist who confesses that no one, in all truth, has a clue about predicting the future of the economy due to its complexity, is a great moment of truth.

The Signal and the Noise, by Nate Silver. Published by Penguin. ISBN: 9780141975658 RRP: $22.99

To sign up for our monthly newsletter, featuring new releases, book reviews and favourite articles from around the web, click here.