Most Political Predictions Aren't Worth 2 Dead Flies

By Andrew Cohen

Every four years we vote for president -- and fight like idiots over pollsters and prognosticators.

cohenpredictionban.jpg
Darren Hauck/Reuters

I don't know Nate Silver from Adam, but I have been in the business of public predictions for 15 years. In that time, I have been asked to make maybe a thousand predictions about legal events and issues, from the outcome of pending Supreme Court cases to the impeachment of a president to verdicts rendered by jurors in capital cases. I will never forget this moment: On March 31, 1997, on my first "official" day as a legal analyst for CBS Radio News, on the very first day of jury selection during the Oklahoma City bombing trial of Timothy McVeigh, I was asked on the air by a local reporter, "So, Andrew, is he going to be found guilty?"

It's been like that ever since. Sometimes I have accurately predicted the outcomes of a trial. Sometimes I have been dead wrong. Many times, more so recently, I have refused to make a prediction at all, saying only that that there are too many random factors at play to make any sort of prediction worthy of much value to the rest of the world. The more I have learned about life and the law, in other words, the more I have come to understand how little I know about why certain things happen the way they do. I'd like to think that's the beginning of wisdom rather than just an acceptance of life's randomness.

Predicting legal events, by rights, should be easier than predicting something as unwieldy as a national election. There are far fewer variables in play. There are only 12 jurors who deliberate in a capital case. There are only nine justices on the U.S. Supreme Court. There are only 100 senators who get to vote on a judicial nominee. There are only a handful of federal circuit courts. There is only one First Amendment. But just think, for example, about how many of us -- on all sides of the ideological spectrum -- were dead wrong this June about the justices and the Affordable Care Act? It's still more art than science.

Political analysts -- the folks who pore over the polling data to tell us who is likely to win an election -- would likely agree. What distinguishes the law from politics or economics or math or science, they would likely argue, is that with the latter disciplines you can soak yourself in numbers, statistics, and charts. There is no mathematical equation to define the scope of the Commerce Clause, they would contend, but if you have enough raw data about early voting in a certain district you should be able to come up with a formula that tells you which candidate is more or less likely to win.

Yes and no. I have no problem with political predictions. I can take or leave them. They give comfort to people looking for a sign. But I have little use for political predictions which are designed to have some sort of cynical impact upon the races they purport to predict. And I have even less use for people who cannot tell the difference between the two. So while I have absolutely no standing to make any political predictions about the 2012 election, and while I have no standing to evaluate the merits of this cycle's most controversial predictors, I thought the following few points were worth sharing as we near the 75-hour mark before Election Day.

Dead Flies. The pre-election predictions of partisans like Karl Rove and David Axelrod, or anyone else with skin in the game, aren't all put together worth two dead flies. I suppose journalists repeat these predictions so that the rest of us can look back after the election and see how wrong this people were. And I suppose that these voices are "experts" in the sense that they have personal insight into the inner workings of campaigns. But the idea that they are going to subvert their party allegiances in the name of accuracy, that they are going to be candid against self-interest, is absurd. Shut it down.

Random acts. No prediction, by a computer or a person, can factor into the equation all the random acts of nature and destiny that will impact voting (and an individual's vote) between now and the time the polls close on Tuesday. We all should have been harshly reminded of this immutable truth this past week with the presence (and now the aftermath) of Hurricane Sandy. It sounds silly to say, because it's so self-evident, but the election won't be determined by polling data. It will be determined by votes. People still have to vote. And vote accurately. And have their votes counted fairly.

Same old story. Memories are short. But the truth is, every presidential election cycle we go through some form of this nonsense. As if we don't have enough to fight about, when we are tired of fighting over the candidates, we start fighting over pollsters and pundits. Remember the great exit-polling scandal of 2004? Remember the kerfluffle about polls in Florida before all those poor old ladies in Palm Beach County mistakenly voted for Patrick Buchanan? We used to fight about a single poll. Now we fight over the calculus of polling. Soon we'll be fighting about the microchips inside a voter's head.

Glaciers. It's worth sharing this graph from Reid Cherlin's fascinating take in GQ on the election of 2012. It's true about all elections, of course, and it's also true about the pollsters and prognosticators:

The problem with writing about elections is that it's difficult to prove anything until they're over. Presidential campaigns are not sprints, or even marathons. They are glaciers: huge, slow-moving, and nearly eternal, capable of moving earth but constantly fracturing and groaning under pressure. You cannot just hike out onto the ice field and know its depth or its force. You can survey the top. You can snap photos and say you went to see it, as I've done here. But you don't really know what it's doing until it recedes, and look! it created the Finger Lakes or Yosemite Valley or what have you.

Oh, we'll say, clearly Obama was never going to be able to overcome the sluggish economy and that awful debate performance. Or we'll say, Romney was a stiff, terrible candidate with no clear agenda; the better candidate always pulls it out. The one thing we know is that we won't look back and admit to much uncertainty when it was actually happening -- even though, for most of us, the uncertainty is the story of this campaign.

Likewise, I see things in terms of Onion headlines. If Mitt Romney wins Tuesday, the headline will be: "Poor People Elect Millionaire Who Pledged To Help Rich People." And if President Barack Obama wins Tuesday, the headline will be: "Angry Whites Reelect Unpopular Black Man to Hardest Job in the World." The point is that the 2012 election narratives already are in place -- the only thing missing are the votes. You could write a lede-all for a Romney or Obama victory today, if you wanted to, and have a decent chance of being correct if you were to submit it when the votes are counted.

Secrets and lies. Some of the criticism of Nate Silver has been based on the fact that he does not share his secret formula by which he conjures up his predictions. I have no problem with this, except to the extent it raises many important questions: Does the formula factor in all the ugly things we now know about the inaccuracies of our vote counts? Does it discount for voter-suppression efforts which redirect registered voters to the wrong precinct? Is there a value placed on the possibility of electronic voting fraud? It's all well and good to predict the outcome of a race. It's another thing for all the valid votes in that race to be counted.

America loves her predictions. Loves to disagree with them in advance and then dissect them afterward. Loves to praise those who got it right and ridicule those who got it wrong. That's why there are so many gainfully employed political and legal analysts and why so many people watch pregame football shows. But the truth is that the recent history of this country proves that what can go wrong in our elections does go wrong, and that there are a million -- there are 10 million -- variables which can, and which will, determine who wins on Tuesday. I don't know who is going to win. And neither does anyone else. And I'm okay with that.

This article available online at:

http://www.theatlantic.com/politics/archive/2012/11/most-political-predictions-arent-worth-2-dead-flies/264476/