HubDub is a great company based in Edinburgh (where I used to live), who apply the principles of the free market to predicting the outcomes of news stories. I’ve had a good chat with them, and they say that their market tends to be spookily accurate. I’ve heard anecdotal evidence that when this type of market is applied to predicting when internal projects will be completed within Microsoft, they were universally more pessimistic, and more accurate, than the project manager’s estimates.
Here is the current prediction for who will win the US Presidential election, right now it has Obama winning (phew!). It will be interesting to look back and see how accurate it was in retrospect: