Easy Chair — From the June 2016 issue

The Improbability Party

Download Pdf
Read Online

I like to think I’m unique. Don’t you? Complicated. Surprising. Unpredictable. I like to think that people who’ve only just met me or who know only the basic facts about me still have a lot to learn. I also like to think that people who’ve known me for a long time — family members, say — still don’t know the essential, the true me. What I really like to think, however, is that corporate statisticians, who track my consumer choices and feed them into algorithms to forecast my behavior for Google or Amazon, are capable, on occasion, of getting me wrong. No, just because I bought Tender Is the Night doesn’t mean I’ll like A Moveable Feast; the Fitzgerald book is fiction, see, and the Hemingway book that mentions Fitzgerald and happens also to be set in Paris is memoir. And I’m not a fan of memoir.

The slightly demeaning guesses are inescapable. I recently spoke to a young woman who accidentally left a streaming-video service running on her television while she went out. Show after show played with no one in the room. When the woman returned, the service’s suggestion engine had cued up a slate of grim reality programs about unwed teenage mothers. Such shows aren’t to her taste, she told me, and so the recommendations confused her. Then she solved the puzzle. If you averaged the subjects of all the programs she had supposedly watched, the result would be unwed teenage mothers.

This is similar to what Nate Silver does with polls, if I understand the super-pollster’s methods. On his vaunted website, FiveThirtyEight, Silver lumps together lots of polls and takes a weighted average to yield a single result. He then makes subtle adjustments to this result using a proprietary formula that I’ve seen referred to as his “secret sauce.” (Being more a language guy than a numbers guy — on the team that’s rarely favored to win — I can’t let it pass unremarked that “secret sauce” feels like a high-end rebranding of “bullshit.”) The word on the street is that if you bet with Silver, especially if you’re betting on elections, which were once thought to be incredibly complex events but turn out to be more like ball games without the balls, you can’t go wrong.

Except you can! For us insecure humanists, the high point of the presidential campaign thus far has been FiveThirtyEight’s blown prediction of the Michigan Democratic primary. With 99 percent certainty — a degree of confidence unknown even among hallucinating psychics — Silver’s dream team predicted that Bernie Sanders would lose to Hillary Clinton. This forecast depressed me when I read it, chiefly because of poetic considerations to which Silver’s breed seems deaf. To me, Sanders wasn’t merely the underdog in Michigan, he was Charlie Chaplin splashed by mud from the tires of a financier’s big car. (Clinton was the car.) Clinton’s claims of superior electability had a harsh, metallic ring to my ears. Vote probability, I heard her say, and let desirability ride. This allied her, I felt, with the glinting Silver himself.

Sanders’s one-in-one-hundred win in Michigan encouraged me personally, not just politically. It seemed to mean that my daughter, a high-school senior, had a shot at an Ivy League college. It gave me hope that my lastest book, in development as a miniseries, might actually air someday, and even win an Emmy. It allowed me to think that I might live to ninety-nine, still witty and brisk, still full of secret sauce. By flubbing his forecast, Silver had regressed to the mean, finally joining the pack. And I felt I’d sprinted ahead of it somehow.

To be fair to the once-exceptional Silver — or to insult him further — it is a huge pack. For more than a year now, the twenty-four-hour news cycle, lately subdivided by Twitter into a second-by-second news vibrato, has buzzed and hummed with poor predictions. Spectacular errors are the norm on the televised panels that shuffle together, three-card-monte style, top journalists with top campaign consultants — the press and the propagandists. Trump is finished. Trump is back. Trump is unstoppable. Wait, he’s stopping. Bill Kristol, editor of The Weekly Standard — a publication that celebrates the market’s punishment of poor performance — has publicly declared the arrival of “Peak Trump” nearly a dozen times in the past year.

Silver and his modeling were supposed to correct this lack of accountability by adding some statistical precision to the mix, but after Michigan he seems to have become just another member of the band. If the expert political observers — and political participants on tactical sabbaticals — had started out as Las Vegas bookies, most would be working as fry cooks in Reno by now. Why aren’t they?

Previous PageNext Page
1 of 4

You are currently viewing this article as a guest. If you are a subscriber, please sign in. If you aren't, please subscribe below and get access to the entire Harper's archive for only $45.99/year.

= Subscribers only.
Sign in here.
Subscribe here.

Download Pdf

More from Walter Kirn:

Get access to 168 years of
Harper’s for only $45.99

United States Canada


December 2018


You’ve read your free article from Harper’s Magazine this month.

*Click “Unsubscribe” in the Weekly Review to stop receiving emails from Harper’s Magazine.