Wolfers opines on the accuracy of Prediction Markets and Pundits:
As a proxy, let’s look at Nate Silver — one of the most sophisticated poll readers in the country — and see if he did a better job. His afternoon forecastwas that Romney was 64% likely to win Ohio. By 9:40 p.m.: “Pretty sure Santorum is a slight favorite in Ohio. He’s maybe ~60%-70% likely to win.” But at 11 p.m.: “Yes, now Romney starting to look pretty good in Ohio” You’ll notice that, like Intrade, Mr Silver’s forecasts also fluctuated wildly. But unlike Intrade, he spent much of the night maintaining that Santorum would win — which turned out to be a poor forecast.
The question here is: If not prediction markets, then what? The best alternative — indeed, basically the only alternative — is to listen to the pundits. And they fluctuated at least as much (probably more!) than Intrade. For just one example among many, look at the tweets of Slate’s Dave Weigel, who went from calling the entire nomination over, to calling the Ohio race close, to calling it for Romney again.
My point is not that Intrade is perfect. But it may be the less imperfect than the alternatives. And that’s good enough.