Two weeks ago I bet that John Edwards will come in fourth place in Iowa, or, if he comes in third, he will beat the fourth place candidate by 2% of less. It seemed like a great bet: Iowa was to be a two-party faceoff between Dean and Gephardt and the latest polls put Edwards soundly in fourth place with 11% support.

Needless to say, I lost. I've been wondering ever since how conventional wisdom proved so misguided in this case. Before Kerry's and Edwards' last-week resurgence in the polls, every major newspaper concentrated its coverage on Dean and Gephardt and repeatedly characterized the chances of other candidates as marginal.

It seems to me that the all the mainstream reporters and analysts are terrible at predicting counterintuitive movements. Their role is fundamentally reactive - they comment on trends which have manifested themselves but never seem to accurately predict the trends that are about to happen.

Early coverage of Dean -- for example, this brilliant and quirky piece by Johnathan Cohn published before anyone had heard of the man -- tended to question whether Dean would be able to have any impact at all on the Democratic race. No one managed to foresee the legion of support Dean would be able to drum up.

Why is this? Are political movements really so difficult to predict? Or is this a culture problem in the media?