- Over the long run and many predictions, markets outperform most individuals;
- The more participants there are in a market, the more accurate it is (although there are diminishing returns);
- The more participants there are in a market, the fewer individuals are able to outperform its accuracy.

Read the rest, and subscribe to his new blog.

More.

-

“A financial adviser or stock broker should be licensed to provide any consultation on investment in securities.”

-

The intelligence and security community also has mixed views. John F McCreary, the Chief Analysis Officer for KGS Security, who had a senior role at the United States Defense Intelligence Agency, notes that the National Intelligence Council successfully trialed prediction markets, until political pressure ended the experiment. However, he believes that, since they “violate the scientific principles of auditability, replicability and irrefutability”, they lack credibility. Retired Naval Intelligence Commander Jennifer Dyer sceptically points out that, “anyone who has bet on sports would be wary of the effects of a large, very specifically motivated group of bettors”. According to Professor Anthony Glees of the University of Buckingham, “there must be a few analysts in the FCO and at GCHQ looking at these sorts of websites but I doubt if they spend much time on them”.

See also the excellent analysis of Mike Smithson.

Every company faces the challenge of [finding out] what our company really knows — how do I access it, when I need it, how I need it — to help drive decision-making or results

Note that Emile estimates the total value of the collective intelligence market, globally, at “roughly $150 million to $250 million”. Ah.

Appendix: Lumenogic (ex-NewsFutures).

Roxana Strohmenger (e-mail me at chrisfmasse at gmail dot com to get a copy of the report):

At the end of the day, a prediction market must have sufficient “information completeness” even if the individuals interacting in the market do not, to accurately predict outcomes with a reasonable degree of certainty.

Paul Hewitt:

Prediction markets must have sufficient information completeness to accurately predict outcomes with a reasonable degree of certainty.

American Civics Exchange:

Having just executed an initial round of closings for the funding we’ve been raising, we’re entering the final run-up to our real-money trading launch. With that in mind, we’ve launched a community feature where prospective traders (and other interested members of the public) can propose political outcomes they’d like to see traded on the exchange.

You can submit your proposals here: http://amciv.com/ideas .

I don’t think there is any question that the usefulness of prediction markets has been exaggerated. We don’t hear much from James Surowiecki anymore. So a few of those polled probably indicated that he doesn’t exaggerate the usefulness (at least not anymore). Justin Wolfers isn’t as well known as Robin Hanson in the prediction market arena. This probably accounts for the large size of the “Don’t Know” vote. This leaves the “Yes” category, which would be made up of Robin Hansonites, who believe the slightest possible improvement in predictive accuracy over other methods justifies a significant investment in prediction markets.

If the question had been asked in a prediction market instead of in a poll, such answers would have led us to believe that the market knows little about the subject, if anything at all. Of course it would have been difficult, if not impossible, to create a prediction market in which the “outcome” is an opinion!