Dave Pollard offers some insights based on James Surowiecki’s book:
Just to restate the basic principle: Many cognitive, coordination and cooperation problems are best solved by canvassing groups (the larger the better) of reasonably informed, unbiased, engaged people. The group’s answer is almost invariably much better than any individual expert’s answer, even better than the best answer of the experts in the group.
The reason for this superiority is that each individual brings to the problem some valuable unique knowledge or perspective, and any errors in that knowledge or perspective are balanced off against those of others in the group, so the collective wisdom of the group is likely to be extremely accurate, reliable, knowledgeable, and predictive. If you’re skeptical, please read the book — Surowiecki presents dozens of examples to support this thesis. The average prediction of one such group, the Iowa Electronic Market, over the several months before the election, was that Bush would win by a comfortable 3% margin and that Republicans would make gains in both houses of Congress. They were exactly right.
My ‘Collective Intelligence’ model realizes (a) that there are some things that crowds can’t do (they need to be given a problem with a discrete or quantifiable set of possible answers from which to choose), (b) that care must be taken in the ‘qualification’ of the crowd to meet Surowiecki’s conditions of nonbias (they must understand the problem, be diverse in their perspectives, independent of groupthink tendencies and each able to bring a bit of unique knowledge to the problem, and (c) that there needs to be some incentive for people to participate in the crowd (those guessing correctly the number of jelly beans in the large jar at least win the jelly beans).