In discussing the US presidential elections with American friends over the years, I’ve heard frequent complaints, and I quote “the entire thing is treated like a horse race surrounded by a media circus“. I have to confess that despite my sensibilities, when it comes to this particular horse race I find myself an entertained spectator through ‘Five Thirty Eight‘.
Reading Asimov’s Foundation series as a kid made me long-time predictive statistics enthusiast. To my delight, I found and have followed Nate Silver’s blog and his election forecasting since his pseudonym (Poblano) days back in 2008. But, what really caught my attention in this case were not just Silver’s forecasts. It was the logical way in which he laid out his arguments and the remarkable clarity in which he delineated his supporting evidence for the layperson. This was in direct contrast to the pundits I was reading from who appeared to be, very literally, making up things. Of course, there was also the thrill of watching the numbers go up and down, and change left to right and right to left.
It was not a tremendous surprise, at least for me, to see that Silver’s predictions were right for the 2008 election. However, this election was presented by the majority of pollsters and pundits as one that would be incredibly close, a sentiment which Nate Silver contested at least on the basis of his electoral college number-crunching. Consequently, there was a lot of speculation not only on which candidate would win, but also which pollsters, pundits, and analysts would be redeemed and which would burn away in fiery shame after the election results rolled in. Gambling on the gamblers. One interesting article was one that came out this week, a column by Jonah Goldberg that questioned Silver’s predictions, and simultaneously appealed for a move away from the “cult of numbers”.
“I’m not saying Silver’s just lucky or shoveling garbage. He’s a serious numbers guy. But so are the folks at the University of Colorado’s political science department whose own model is based on economic indicators. Its Oct. 4 findings predict Romney will win, as do others… I like to think that people are different, more open to reason, and that the soul — particularly when multiplied into the complexity of a society — is not so easily number-crunched.”
Well, Nate Silver was right. Which has led some to speculate that he maybe be Nostradamus reincarnate or a witch. To be fair, so were a a number of other witches and wizards. The implications of the election outcomes aside, I was pretty satisfied that FiveThirtyEight turned out to be completely on target (though I think the writing is worth it even if the predictions were completely wrong).
Given the coverage that these ‘grand wizards’ of the ‘cults of numbers’ are getting (stretching the metaphor here), it is clear that there will more and more demand to reach forecasting into quarters where it hasn’t really been applied before, outside perhaps advertising research. In fact, the future is already here. There was an article in nature back in 2009 that attempted to model insurgencies, laying the groundwork for predictive models (see Sean Gourley’s great TED Talk), millions of dollars are being spent to develop computational social science models that can forecast “Cloudy, with an 80% chance of war”. Earlier this year Slashdot had an article on these models with the headline “Asimov’s Psychohistory Becoming a Reality?” I’m personally very interested in health planning and policy, and curious how this emerging predicative statistics arms race will affect policy making in this arena.
Asimov’s psychohistory (and the driving force behind the novel’s narrative) was based on the premise that “while one cannot foresee the actions of a particular individual, the laws of statistics as applied to large groups of people could predict the general flow of future events.” But, this vision might not even be the most ambitious, as neuroscience and psychology are fields moving towards creating a deterministic model of the individual as well (though realistically, such developments are many years away). I’m personally a numbers gal but in the courses I am taking and the groups I collaborate in, there is always the reminder that there are people behind the numbers and that there are facets of individuals and societies that absolutely cannot be quantified.
Don’t get me wrong, I am absolutely looking forward to a future where policies are based on empirical evidence rather than somebody’s gut feelings or popular opinion. However, I do wonder whether there are things that we miss as we seek to quantify and fit every facet of social patterns and behaviour into statistical models. Though I am not per se concerned about number crunching ‘the human soul’ I think Goldberg presented at least one important idea in his article: are there in fact facets of humanity that are simply not quantifiable, and not predictable? Or can we encompass everything we are and we will be as a society into neat statistical models?Permanent linkMarzieh Ghiasi