According to Reason, the wisdom-of-crowds-based prediction markets failed for the Senate race this last election. Most prediction markets were putting the likelihood that the Republicans would keep a majority in the Senate at 75-80%.
Does it really mean a failure of the system? If the probability that most people in the “crowd” would predict that the democrats would win 6 out of 7 tight races in order to win a majority in the senate is less than 50%, then the wisdom of crowds would only reinforce that – at least that is what the Condorcet jury theorem says. Besides, predicting that there is a 20% chance that the dems would win those races and thus take control of the Senate is not a negligible chance. In fact, it probably is somewhat higher than the straight probability that pollsters would have come up with.
What do you think?