Nate Silver On How Not To Predict The Future

Photo by Randy Stewart

Think you can predict the future? Good luck. In his new book The Signal and the Noise, statistician Nate Silver—he of the New York Times's FiveThirtyEight blog, who called the 2008 presidential election with near perfection—says to question your faith in many trusted prediction methods. Here are four:

1. ECONOMIC INDICATORS
"They're one hot mess," Silver says. The government produces data on 45,000 indicators annually, but few are anything more than correlation. And folklore indicators—like which team wins the Super Bowl—receive routine press attention but are almost always bogus.

2. EXPERT OPINIONS
After conducting a survey of 1,000 predictions made by panelists on the TV show The McLaughlin Group, Silver concluded that so-called experts display as much acumen as "a barbershop quartet." The pundits, he says, were no better than a coin flip.

3. EXTRAPOLATION
Back in 2009, as cases of H1N1 went from 20 to 2,618 in two weeks, U.S. authorities described a plausible scenario in which half the population became infected, leading to 90,000 American deaths. The strain (thankfully) turned out to be exceptionally mild, with just a 0.02% fatality rate.

4. FORECASTS
Commercial meteorologists deliberately overestimate the likelihood of rain, forecasting a 20% chance when there's only a 5% to 10% chance. Why? It's a way of "covering their butts," Silver says.

Illustration by Vidhya Nagarajan

Add New Comment

0 Comments