sábado, 3 de noviembre de 2012

Did Nate Silver Let Twitter Get Under His Skin?

Mashable OP-ED: This post reflects the opinions of the author and not necessarily those of Mashable as a publication.

It was shaping up to be a classic 'Revenge of the Nerds' scenario. In one corner, we had Nate Silver, the stats whiz whose algorithmic models have correctly called everything from baseball rankings to the 2008 and 2010 elections. In the other, a pundit class with little on their side other than gut feelings and loud voices.

Then tempers flared on Twitter — and Silver suddenly found himself, rather than his numbers, at the center of the story.

Silver, whose poll-based analysis site FiveThirtyEight.com was bought by the New York Times two years ago, takes a nonpartisan, statistical approach to the election. His model doesn't cherry-pick polls, but weighs them based on how accurately each polling organization in each state has performed in the past.

Allowances are made for economic data, such as the jobs report numbers and stock market rises and falls, which have affected elections in the past. Until recently, that was about the most controversial thing about Silver's model — in that it drew complaints from the Obama campaign.

FiveThirtyEight's numbers, which currently give President Obama an 80% chance of winning reelection, are hardly outliers in the world of election statistics. Indeed, the nonpartisan Princeton Election Consortium, with even more high-octane math genius at its disposal, rates Obama's chances at upwards of 96%.

How does that square with the fact that the average national poll puts the race pretty much even? Simple. Nationwide polls, historically, have never been as accurate as statewide polls. That's not to say they're bad, they just have a higher ratio of noise to signal. Obama has retained a slim but solid lead in most battleground state polls all year — his so-called "firewall."

It's Statistics, Stupid

"National polls have systematic problems," writes Professor Sam Wang, neuroscientist and founder of the PEC. Historically, he says, "they do about 2.5 times worse at predicting the popular vote outcome than expected."

When Wang says that, it's a relatively uncontroversial statement. But Silver appears to be a different case, which is perhaps down to his media prominence. He has been widely promoting his book, "The Signal and the Noise: Why So Many Predictions Fail — But Some Don't," which includes a chapter pointing out that TV pundits are right no more than 50% of the time.

This week, it seems, the pundits decided to strike back. It began with a Politico article on Monday calling Silver a "one-term celebrity" (evidently ignoring his success in the midterms and in baseball).

Silver's response on Twitter was cutting, but still dispassionate:

Then MSNBC host and former GOP congressman Joe Scarborough picked up the baton with an outlandish attack on Silver. "Anybody that thinks that this race is anything but a tossup right now is such an ideologue," he said on Tuesday, "they should be kept away from typewriters, computers, laptops and microphones for the next 10 days, because they're jokes."

Silver's tweets started to show evidence of frustration at the "tossup" notion:

The next day, Silver started addressing Scarborough directly:

Then came Silver's most controversial tweet:

There are a number of odd things about this tweet. Firstly, it appears to come out of nowhere: for all his on-screen invective, Scarborough had not been tweeting at Silver. Secondly, it left Silver open to the charge that he was a partisan. Suddenly, he looked like he had skin in the game.

Thirdly, it was a little tin-eared, as critics pointed out: if people are suffering in the wake of Sandy, why not simply donate to the Red Cross now? Indeed, that's how Scarborough ended the affair Thursday, by taking the high road:

Silver, meanwhile, earned himself a mild rebuke from the New York Times' public editor Margaret Sullivan, who said that making such bets was beneath the standards of a Times reporter. In a call with Sullivan, Silver described the wager as "half playful, half serious" and reminded her the week before the election was "high stress time."

High stress and tweeting, as we've noticed multiple times at Mashable, do not mix well. Take the London Olympics, where multiple athletes got into hot water over Twitter remarks made during training. One swimmer threatened to quit the service because of trolls (though in the end, she decided to stay.)

A few political careers have been damaged by tweets too. Not that Silver is at risk; he has plenty of defenders in the journalism world (such as this article in the American Journalism Review, which takes issue with Sullivan's rebuke).

But there are plenty of us rooting for data, for the wisdom of crowds, for what Wang calls "The Math." As entertaining as the horse race narrative is, it doesn't really get us any closer to the true mood of the electorate. If you think there's a single pundit out there without a personal agenda to their pronouncements, I've got some timeshares in bridges to sell you.

The Silvers and Wangs of the world, folks with more passion for statistics than for individual politicians, have the capacity to change that. But only if they stay above the fray, as cool and dispassionate as their figures. It may behoove them to spend more time crunching numbers, and less time tweeting at critics.

Image via Flickr, Randy Stewart

Publicado por en

No hay comentarios:

Publicar un comentario

Suscribirse a: Enviar comentarios (Atom)