Decision markets

I've made some posts on James Surowiecki's financial column in the New Yorker before. Now, I found a short piece of his (related to his new book 'The Wisdom of Crowds') in Wired.

He describes the uncanny knack of large crowds getting things right more often than experts. Markets, in particular, seem to be good at forecasting where experts fail. According to the Publishers Weekly review, the book outlines 4 necessary criteria for large crowds to be 'smart.'

(1) diversity of opinion
(2) independence of members from one another
(3) decentralization
(4) a good method for aggregating opinions

Strangely, (financial) incentive isn't listed, even though Surowiecki has argued it's relevance before, during the Political Analysis Market scandal.

In Wired, Surowiecki describes some companies' experiments with internal decision markets and their encouraging results.

Firms need to aggregate the collective wisdom [...]

One intriguing method of doing this is to set up internal decision markets, which firms can use to produce forecasts of the future and evaluations of potential corporate strategies. Few companies have tried such markets. But the few examples we have suggest that they could be very useful. In the late 1990s, for instance, Hewlett-Packard experimented with artificial markets to forecast sales. Only 20 to 30 percent of employees participated, and each market ran for just a week, with people trading at lunch and in the evening. The market's results outperformed the company 75 percent of the time. Even more impressive was a recent experiment at e.Lilly, a division of Eli Lilly, which set up a market to test whether it was possible to distinguish between drug candidates likely to be approved by the FDA and those likely to be rejected. Realistic profiles and experimental data for six hypothetical drugs were devised by e.Lilly, three of which it knew would be approved and three rejected. When trading opened, the market - made up of a diverse mix of employees - quickly identified the winners, sending their prices soaring, while the losers' prices sank.

The evidence is clear: groups - whether top executives evaluating a potential acquisition or sales reps and engineers analyzing a new product - will consistently make better decisions than an individual. Companies have spent too long coddling the special few. It's time for them to start figuring out how they're going to tap the wisdom of the many.

I'm not so sure that groups are always better. For one thing, the market mechanism probably wouldn't work for fuzzy questions or where there are not yet clearly defined alternative solutions. Nevertheless, it's an interesting tool. And, given the difficulty of reaching decisions in large groups or decision-making online, it certainly deserves more attention.