I, along with every other data/political/media geek out there, have been fascinated by the high drama surrounding New York Times election stats guru Nate Silver. He accurately forecasted the presidential election in a race when others said it was impossible — that the race was too close to call.
Slate has a wonderful graphic showing just how far off most pundits were.
As they say, victory is the best revenge, but the pro-Silver adulation is, in some circles, getting carried away. Among the digital set, “innumeracy” has become as cutting a putdown as outing one as a Hotmail user.
And that’s a good thing: Replacing party talking points with data points is way overdue.
But innumeracy is no worse a sin than false equivalence.
But take Chris Taylor’s Op-Ed that proclaimed Tuesday’s winner “Nate Silver and his running mate, big data.”
Setting aside Taylor’s vision that smart modeling takes out all “myopic human bias out of the equation,” the data that Silver was working with was, in today’s terms, relatively small data: The input that Silver took was dozens of polls, not tens of thousands.
Daniel Engber throws a little more cold water on the Silver-as-genius theory, but my favorite analysis was from Felix Salmon, who wrote that Silver’s genius wasn’t in his prognostication, which was good, but in his writing, which made that prognostication accessible and exciting to a wide audience.
But even then, these discussions are all about prognosticating the elections, and not influencing them. Well before Tuesday, both campaigns invested heavily in big data crunching.
TIME has an excellent look inside the Obama data machine, which leaned heavily on email and demographic datapoints. But Romney’s campaign also tapped into big data, or at least tried: It launched an ambitious, crowdsourced polling dashboard, as the Washington Examiner reported, which was supposed to help drive turn-out-the-vote tactics on election day.
“In recent weeks the campaign came up with a super-secret, super-duper vote monitoring system that was dubbed Project Orca,” the Times’ Byron York reported. “The name ‘Orca,’ after the whale, was apparently chosen to suggest that the project was bigger than anything any other campaign, including Barack Obama’s in 2008, had ever imagined.”
By election night, rumors were swirling around Kendall Square that Orca had failed, and York’s reporting backed that up.
Orca’s beaching certainly didn’t help things, but there’s a lot of indicators that more traditional, non-big data decisions and events ultimately drove the election: Jobs data improving, Hurricane Sandy, a luckluster debate showing. And, in a very old-fashioned way, a big bet of when and how to launch attack ads, as the Wall Street Journal reported.
But where big data is getting interesting is here, in Boston, particularly in the life sciences space. We recently highlighted Patients Like Me, which built a social network that connects patients while analyzing their symptoms, treatments, and outcomes. And Xconomy reported today that Cambridge’s GNS Healthcare recently inked a partnership with Dana-Farber Cancer Center in Boston and Mount Sinai School of Medicine in New York to analyze multiple myeloma.
To me, that’s the really exciting impact of big data: Not to determine whether you should end subject lines with more colons, but to determine how to transform medicine, cities, and societies.
With scores of true big data companies and scholars in the area, and big data incubatorhack/reduce’s official launch tonight, Boston is perfectly poised to help make those transformations happen.