Cass Sunstein offered us a preview of his new book, tentatively titled “Mobs, Markets and Blogs”. As with Republic.com, it sounds like he’s asking tough (and eminently worthwhile) questions about whether the cybertools I know and love make us better citizens and better people.
Sunstein opens by quoting the first post from Judge Richard Posner’s blog, which reads, in part:
Blogging is a major new social, political, and economic phenomenon. It is a fresh and striking exemplification of Friedrich Hayek’s thesis that knowledge is widely distributed among people and that the challenge to society is to create mechanisms for pooling that knowledge. The powerful mechanism that was the focus of Hayek’s work, as as of economists generally, is the price system (the market). The newest mechanism is the “blogosphere.”
Sunstein is curious how groups find ways to pool widely distributed knowledge. He suspects that certain models of deliberation are better at destroying knowledge than creating it. He references a study on deliberation, where groups of Americans were invited to deliberate divisive political issues in small groups with likeminded individuals. In almost every case, individuals’ opinions on the issues being debated were more polarized after deliberation than beforehand – liberals ended up more liberal, conservatives more conservative. And differences of opinion within groups of likeminded people were squashed during the deliberative process, leading to more homogenous opinions.
Sunstein observes that shared information has more weight in group settings than unshared information. We like a speaker more when she says something we already know (and agree with), so a group tends to settle on shared information and filter out unshared information. Groups can amplify error, rather than correcting it. And cascading effects mean that, if the first few speakers state an opinion, it can be very difficult for subsequent speakers to disagree with the stated opinion.
These effects make Sunstein suspicious of the effects documented in Surowiecki’s “The Wisdom of Crowds”, where large groups of people are asked for answers to quantitative questions (“How many jellybeans are in this jar?”). While individual answers are usually off-base, the mean of answers is often remarkably close to the reality.
Sunstein analyzes the effects Surowiecki sees in terms of Condorcet’s jury theorem: if each individual in a group has a better than 1 in 2 chance of making an accurate determination, as the size of the group increases, the chance of a consensus decision being correct approaches 100%. Conversely, if each person in a group has less than a 1 in 2 chance of being correct, Sunstein proposes that the chance of a group making the correct decision approaches 0%.
With this in mind, Sunstein believes he has an explanation for why prediction markets – like the Iowa Electronic Market – are so successful. Most people don’t participate in prediction markets unless they have a reasonably good chance of making a good prediction. And bad predictions are punished economically, and quickly squeezed out of the market.
Sunstein wonders whether other group production methods are able to take advantage of the “Posner-Hayek” effect, using market methods to gather disparate data to make intelligent decisions. He sees Wikipedia and the Open Source movement as two examples where large groups can aggregate information, pointing to Eric Raymond’s assertion that “with enough eyeballs all bugs are shallow”.
Blogs, however, aren’t a Hayekian market, Sunstein speculates. They’re more like “an expansive deliberative forum that contains diverse information”. This can lead to amplification of errors, cascades of false information… in other words, all the bad effects one can see in group decisionmaking behavior.
There was some good pushback from the audience (myself included) to the notion that participants in open software projects, Wikipedia and the blogopshere were undifferentiated, equal participants in a deliberative process. I pointed out that most open software projects tend to be highly hierarchical, relying on large groups to report bugs, but a very small number of developers who fix them and make code changes. Daniel Drezner pointed out that there are hierarchies in the blogosphere based on expertise about different topics – in Rathergate, it became clear very quickly that some bloggers had a great deal of esoteric knowledge about typewriters and typesetting – those bloggers had more authority in the discourse than bloggers who didn’t have expertise.
I’ll be very interested to see how this pushback – which Sunstein accepted graciously and gratefully – affects his thinking moving forwards. Sunstein continues to raise some of the most interesting questions about the implications of our tools and I’m very much looking forward to learning more about his take on this subject as it develops.