Becker-Posner on Prediction Markets

Becker and Posner, over at the aptly named Becker-Posner blog opine on prediction markets being used to forecast political races. I scanned Posner's, but read Becker's, and it sounds like Becker sees the markets as having some biases, despite their winning track records (the Iowa Electronic Market has successfully predicted the Presidential race every time since 1988, when it began). He thinks that because the bets are relatively small, the usual theories about risk and uncertainty don't apply. Instead, he thinks people bet because they get utility out of the bet, and so there's a tendency for a home team bias as people bet on their favorites, and not necessarily on the outcome itself. Entirely possible. But it doesn't explain the accuracy, if so.