As an InTrade bettor, I think a lot of the mispricing is just driven by wrong beliefs. In the primary season, I could usually get good prices to short Ron Paul and I thought it was likely to be because American InTrade bettors (internet savvy, ok with questionably legal things involving money, nerdy) overrepresent Paulites. No manipulation, just sampling problems.
I assumed that demographics of InTrade made it pretty likely they'd overprice marijuana legalization, too, and I had to pull my money out of the Colorado market with a quickness once I'd heard the referendum was actually getting editorial page endorsements. :)
Overcoming Bias: Was Intrade being manipulated in the last month?
The link is good apropos, but the question that interests me is a bit more general. It seems to me that the apparent failure of Intrade to function as a predictor/knowledge aggregator before the US presidential elections was an important natural experiment. What do you think about the explanations and implications? I don't think it's too hard to discuss the knowledge aggregation issues without being bogged down in the specificities of US politics.