So imagine it’s 1826 and you want to know what the weather will be doing tomorrow. You really have one choice: pull out your trusty Farmer’s Almanac and get down to business.
The Almanac is still around today. As Sandy Duncan, managing editor, says, “The formula we use dates back to 1818. It is a mathematical and astronomical formula that takes sunspot activity, tidal action of the moon and position of the planets into consideration. The complete formula is known only by our weather prognosticator: Caleb Weatherbee."
Sounds pretty cool. There’s only one problem: analysis shows that its accuracy falls in the 50/50 range. That is to say, garden variety coin toss territory.
So, would I be better off checking in with the National Weather Service?
In short: yes. Meteorology has come a long way since 1818. No longer must we rely on the whims of a man named Caleb. Now we can pull up software that feeds past weather results, current temperature, wind and precipitation patterns into a statistical probability model. For predictions a few days out, there’s something like a 72% success rate. (Science at work.)
Wouldn’t it be great if we had a scientific model that could predict elections with more accuracy than a coin toss? Lest we forget, most major news networks said the race was too close to call the day before the election. With few other options, many of us turn to the pundits. These brave men and women are not afraid to dispense their political expertise and give us the inside scoop on winners and losers. (They get paid for this.)
One of my favorite such figures is George Will, frequently found hanging out with George Stephanopoulos on ABC’s This Week. With his trademark acerbic wit and kicky bowtie, Will never disappoints, whether waxing prophetically about baseball, global warming or the winner of the presidential race. (This year he picked Mitt Romney by a wide electoral margin, carried by unlikely states like Pennsylvania, and Minnesota, home of the late George McGovern.)
As of today, the outcome of this bet is, as Thomas Jefferson might conclude, self-evident. In Will’s defense, one might argue that presidential picking is difficult business, at the mercy of countless variables--natural disasters, voter apathy, candidate momentum, shameless wall-to-wall ads, and the last-minute decisions of those strange and elusive creatures known as swing voters.
All of this might have you reaching for your Almanac, or even a trusty coin to flip, except for one thing. More specifically, one guy.
Enter Nate Silver, a.k.a. “Poblano.” Journalist, baseball scholar, TED Talker, writer, and one of Time’s 100 Most Influential People in 2009. Back when everyone else was saying it was neck and neck, Silver put President Obama’s odds at 90.9%. He wasn’t saying Obama would win--simply that if the race were held 100 times, Obama would likely win 90.9 times to Mitt Romney's 9.1. His numbers lined up perfectly with Tuesday’s results; he wound up picking the correct winner for each of the 50 states and the District of Columbia.
How did he do it? Is Nate Silver a witch?
Daniel Terdiman reported that Silver “relied on a complex recipe that starts with endless polls, weights them by historical accuracy, throws in a dash of economic indicators, sprinkles on some demographic data, and stirs it all together in order to run thousands of simulated elections." Which is to say he runs a statistical probability model. (Science at work.)
There’s something appealing about cracking open the Farmer’s Almanac to find out whether it might rain tomorrow. In the same way that it’s impossible to resist an eloquent speaker willing to sport 1950's neckwear in front of millions of viewers as he explains that Mitt Romney is going to pull off an upset. And I, for one, will continue to read the Farmer’s Almanac and watch George Will. As entertainment, it can’t be beat.
Still, if its accuracy you’re after, the National Weather Service and Nate Silver appear to be a much better bet.