The most current issue of the Harvard Business Review has an interesting article in which they marry the wisdom of crowds “theory” with the Condorcet jury theorem – which was developed by a Frenchman in 1785 (requires subscription).
To understand how the theorem works, imagine that a number of people are answering the same question and that there are two possible answers – one correct and one incorrect. Assuming that the chances that an individual will answer correctly is more than 50%, the theorem proves that the probability that a majority of the group will answer correctly increases towards 100% as the group size increases.
That explains why the wisdom of crowds works well for consumer product testing, or to predict well documented political races. It also explains why it failed in other areas, like predicting whether Iraq had weapons of mass destruction. With little information available, individuals had a higher likelihood of picking the wrong answer, making the chance that a majority would predict correctly close to 0% as the size of the group increased.