The Internet and collective problem solving

The whole Dan Rather-Bush national guard forgery episode appears to have brought the blogosphere into a new prominence. (Though, check out Matthew Yglesias’ objections to all the self-congratulation.) In some ways, it does appear to be an instance of collective problem solving.

As the recent and peculiar embarce of the masses (all the rage these days thanks to James Surowiecki’s The Wisdom of Crowds, but see this review by Daniel Davies) implies, if for a pool of people who are to come to a collective decision, each individual has, say, a 51% of being correct, as the number of people grows, the chances that the collective decision will be right is significantly greater than 51%. For a group of three people, each of whom has, say, a 2/3rd (67%) chance of being right, the chances that a collective decision is right is significantly greater, nearly 75%.

When the chances that an individual is right is greater than his or her chances of not being right, majorities are very likely, more likely than any individual, to produce the correct decision. This is the upshot of Condorcet’s Jury Theorem.

(Of course, if each individual’s chance of being right is less than half, y < .5, the chance that a majority will right will be less than y, by symmetry.)

Add to this the fact that for reasons of Bayesian rationality, we should weight what majorities believe heavily, unless we have good reasons. (I’m discussing empirical issues, and not disagreements about values.) If the Jury Theorem gives us reasons to believe that the outcome is correct, we should adjust (update) our beliefs to what the majority has thrown up as the correct answer, submit, as it were, to a tyranny of the majority’s ontology, as Robert Goodin suggested as a set up to his argument.

With the advent of, not simply blogs, but large scale collective problem solving enabled by new technologies, is there the chance that we receive better information and better answers? (Again, none of this stops value conflict.) The experience of wikipedia.org, in which, incorrect information is corrected quickly, seems to suggest “yes”.

I don’t live in Korea, nor do I speak Korean, but the case of OhMyNews—a South Korean news service in which “citizen reporters”, ordinary people, call in news–may over time prove the extent to which the involvement of “crowds” improves information, though strictly speaking the number of people who report any single story isn’t clear, and neither is how stories are corrected. The stories are ranked according to credibility. But the reception of OhMyNews in Korea does suggest that the logic holds.

“When some Yonsei University students recently met with a visiting reporter to discuss the future of news, one psychology major put it simply: ‘How can you ever get truth from one source? The Internet allows us to check multiple sources, to explore message-board postings, to debate issues with others—that is the only way to find truth. And besides, what good is information if you can’t react to it?’ ‘We’re not stupid,’ added a business student. ‘We know that there is a difference between a message board, a traditional journal and OhmyNews. But by putting them together, our understanding is better. We can piece together truth.’