The wisdom of crowds is a much misunderstood term. It does refer to the result of a large group of people thinking - but only in special circumstances. Not as a least common denominator/aggregate as it is commonly and incorrectly reduced to.
I would strongly recommend James Surowiecki's 'The Wisdom of Crowds' to discover in what circumstances it works. In short, a prediction system with WoC requires diversity of representation, independence of decision and a particular kind of decentralisation.
Prediction markets are a special circumstance where the necessary conditions for the wisdom of crowds come together.
And the source for my confidence about prediction markets - James Surowiecki's book, the number of prediction market implementations currently ongoing and their results along with my own participation in many of these prediction markets.
The Iowa Electronics Market, HSX.com (which began as a game but due to the accuray of its results is currently being used by Hollywood studios to 'peek' into the future), Yahoo Tech Buzz (a predition market for technology run by O'Reilly Media and Yahoo) and NewsFutures (another prediction market about news events.)
In fact, several companies like Microsoft have already implemented them internally as they are a wonderfully cost-effective way to get information about the future.
Post 9/11, the Pentagon wanted to implement a prediction market to 'aggregate' the wisdom of crowds and find out the next likely target of terrorists. The system was squashed because of public protest. I know it sounds harsh to bet on which place is likely to be hit next - but it was a system that would have yielded the best possible results.
Your example about an individual betting against the wisdom of crowds is not proof of the system not working. For our discussion, we are talking of stockmarkets as information systems - giving us outsider observers a picture of the likely scenarios in the future. It works because the players are willing to gamble and take more or les risk than others, which evens out to give us a fairly accurate picture. Their personal profit and loss keeps the system running - it doesn't mean the system has failed.
I apologise for stressing on the "reliable" though. What prediction markets will do is give us an accurate probablity picture of what is likely to happen - of course, the only way to be sure about the future is to wait for it happen.
But the point I am making is that though we'll never know with relaibility what will happen in the future, in all likelihood prediction markets will present us with all the options and their likely probablities (however miniscule.)
In which case, we'll never have reason to be surprised that something happened. We'll only be surprised that something happened sooner or later than expected.
And of course like all prediction systems, a prediction market's accuracy will increase or decrease depending on how near or far the system is looking into the future.
