The triumph of mathematics (or how Nate Silver got drunk)
November 19, 2012
Posted by on
“Drunk Nate Silver stumbles through traffic on the Jersey Turnpike, screaming out what time each driver will get home.” @davelevitan
I know… I am late to the game… let’s chalk it up to a very busy schedule in the lab. But I want to write about the elections (queue eyes rolling).
I arrived in the US in 2006, so I was fortunate enough to witness the Obamania that swept this nation in 2008. I was quite fascinated with the dynamism of the elections and I was watching it VERY closely. That was the first time I came across fivethirtyeight.com, a blog started by a sports statistician named Nate Silver. His simple yet elegant model correctly predicted the election outcome in 49 out of 50 states. Despite his rise, the one-sided 2008 election was not a very good indicator for the supremacy of his model. In 2012, however, everyone believed the race to be a significantly close one. While pundits called the race a virtual toss-up, Nate Silver (and other statistician/bloggers like him, including Sam Wang of Princeton Election Consortium) were assigning very high chances of winning to president Obama throughout the campaign season. This made Nate Silver a punching bag for the TV hosts, and the punditry in general, in the run up to the elections… however, the accuracy of his statistical model proved to be quite impressive (it called all 50 states and all but one of the senate races). This made him the true winner of the elections… with his book becoming a best-seller and #drunknatesilver becoming a popular hashtag on twitter (where I got the quote in the beginning of this post from).
Nate Silver’s model is fairly simple and is very similar to the models used by other poll aggregators (who predicted pretty much the same outcomes). I think, anyone with an adequate knowledge of statistics would have come up with a comparable model. I really don’t want to talk about the model or why it worked so well (which I don’t think is very surprising to any scientists). What caught my attention however, was the extent to which people were shocked by the efficacy of these statistical models. This, I think, clearly indicates that people underestimate science and its ability to deliver. I think, as scientists, we should be worried about this. Why this is the case, I really don’t know… is it the successful war on science? Is it the botched PR dramas by fraudulent scientists? Is it going head-to-head with religion and losing? I don’t know… what I do know, is that Nate Silver is not an extraordinary researcher/mathematician. He has a job and he does it well, but what he’s doing is not groundbreaking. Nevertheless, in this election, science squared off against ideology and won a decisive victory. And we should take this as an opportunity and build upon this. How? I am again not sure… I just know that this opportunity should not be wasted.
Dan Levitan started a game on Twitter: #DrunkNateSilver: things Nate Silver might do/say when he’s drunk.