Pollsters are making elections increasingly possible to predict, but their clairvoyance doesn't serve the public any good
Nate Silver at the New York Times provides a thoughtful, informed answer to the question "Is Obama Toast," but the better answer is "let's wait and see." No answer to the question "who will win" is actually the best answer, not because elections are unpredictable -- statistical wunderkinds like Silver make elections increasingly possible to predict -- but because the predictions don't serve the public good.
Who benefits from handicapping elections? Private interests. The handicappers -- pollsters and pundits who are paid to tell us how and why we're likely to vote -- are obvious beneficiaries. The media outlets employing them benefit or expect to benefit as well by maintaining or increasing their audience shares. Individual candidates benefit by increasing their chances of success with the aid of polling data, focus groups, and analyses of the electorate's mood. And major individual and corporate donors seeking access, influence, or ambassadorships benefit from data that reliably inform their bets on the candidates most likely to succeed.
You could argue, I suppose, that electoral handicapping helps keep the public engaged, but it seems more likely to depress and disengage people whose favored candidates seem destined to lose while encouraging complacency in those who are predicted to win. You could argue that sophisticated polling and analysis can help the best candidates win, but it can just as easily help the worst candidates. Or, you could speculate that grassroots movements that represent significant portions of the public can similarly use polling data to fashion their appeals and advance their ideals; but as Occupy Wall Street shows, genuine grassroots movements tend to be more spontaneous than strategic.
In questioning the public interests served by electoral predictions, I'm not questioning the value of all electoral analyses. The best analyses are interesting, insightful, and have considerable historical value. In other words, their virtues are not predictive but retrospective: they help make sense of our political history. Perhaps someday we'll learn from it.