Oct. 6 (Bloomberg) -- Humility and punditry are two characteristics not often found in the same person -- or the same book. But they co-exist more or less happily in “The Signal and the Noise,” Nate Silver’s engaging if occasionally shambling new book on the fine art of prediction.
And it is an art, as Silver -- the proprietor of the New York Times’s FiveThirtyEight political blog -- makes abundantly clear. When it comes to successful forecasting, he shows that knowing what to ignore is at least as important as knowing what to consider.
Just as vital, he argues, is being willing to embrace rather than ignore the uncertainties inherent in trying to foretell the future. And if that means the resulting prediction is a little less sure of itself than we’re used to hearing from our expert class, well, so be it.
Silver first came to prominence among a subset of baseball statheads with Pecota, a system for projecting players’ performance (the name being a retroactive acronym constructed to recognize Bill Pecota, a marginal infielder who used to torment Silver’s favored Detroit Tigers).
Pecota had some notable successes, including tagging Boston Red Sox prospect Dustin Pedroia as a future star when few scouts would have agreed. But it had some notable misses too -- and Silver, to his credit, not only talks about them but draws the larger lesson:
“The key to making a good forecast,” he says, “is not in limiting yourself to quantitative information. Rather, it’s having a good process for weighing the information appropriately.”
That task is made infinitely more difficult by the vast amounts of information available. The trick is in separating the knowable from the unknowable and the relevant from the irrelevant, and in recognizing the difference between correlation and causation.
Ice-cream sales and forest fires both peak in the summer, he notes. “But there is no causation; you don’t light a patch of the Montana brush on fire when you buy a pint of Haagen-Dazs.”
Indeed, Silver finds common threads between our predictive failures when it comes to natural disasters and man-made ones.
In a chapter called “How to Drown in Three Feet of Water,” he skips blithely from the failings of economic forecasters -- most of whom not only failed to predict the recessions of 1990, 2001 and 2007, but didn’t even recognize them after they had already begun -- to the 1997 Red River floods in North Dakota, then back again.
Those intellectual leaps are some of the most interesting things about “The Signal and the Noise,” and some of the most annoying. Silver is clearly a smart guy, and is refreshingly free of the glibness and smugness that can creep into the work of “Freakonomics” authors Steven Levitt and Stephen Dubner.
Yet it can be exhausting to follow the twists and turns of his mental gymnastics. Perhaps we could have made do with a shorter account of how International Business Machines Corp.’s chess-playing Deep Blue computer spooked world champion Garry Kasparov, or a little less on the various strains of flu that never triggered a pandemic. Both the book and the reader would have benefited from fewer somersaults and more editorial focus.
And for a guy who really made his reputation in politics -- his 2008 model successfully forecast the presidential outcomes in 49 of the 50 states -- there’s surprisingly little on the subject in “The Signal and the Noise.”
For that, we have to turn to Silver’s blog -- a form that, with its short-attention-span conventions, may actually prove better at enforcing literary discipline than the bigger canvas of a 454-page-plus-footnotes book.
Still, Silver displays a knack not just for mining data but for explaining his thinking in an accessible manner. It’s nice to come across someone as insightful as he is who doesn’t seem completely dazzled by his own brilliance.
“Our bias is to think we are better at prediction than we really are,” he concludes, hoping for a future in which we’re “a little more modest about our forecasting abilities, and a little less likely to repeat our mistakes.”
“The Signal and the Noise: Why So Many Predictions Fail -- But Some Don’t” is published by Penguin Press (534 pages, $27.95). To buy this book in North America, click here.
(Rich Jaroslovsky is a Bloomberg News columnist. The opinions expressed are his own.)
Muse highlights include Jeremy Gerard on theater and Lewis Lapham’s podcast.
To contact the reporter on this story: Rich Jaroslovsky in San Francisco at firstname.lastname@example.org.
To contact the editor responsible for this story: Manuela Hoelterhoff at email@example.com.