Julia Kamin / May 17th, 2011 / Tweet
To anyone familiar with human stampedes, financial panics or professional soccer games, the “Wisdom of Crowds” may seem more like an oxymoron than a legitimate pursuit of study.
But as James Surowiecki so adeptly describes in his book of that name, crowds can be wise.
Under certain conditions. First, they must be diverse; homogeneous groups will be limited – or aggravated – by their shared narrow perspective. Next, groups must be set up so that individuals can think independently, thus avoiding the twin traps of groupthink and mis-information cascades. Finally, they have to have a way to aggregate their ideas, inputs and decisions.
Open source communities and (some) markets are good examples that meet all three conditions. Filter bubbles, which encourage homogeneity and cascades while eschewing communal aggregation, are not.
But if filter bubbles may make online communities stupid, could we make algorithms that make us collectively smarter?
We’re probably a long way off, but the Center for Collective Intelligence at MIT is at least moving in the direction. Led by Thomas Malone, the center is looking more deeply at the “DNA” of smart groups; how the “what, who, why and how” of a group correlates to group intelligence.
In a study published last year, Malone and his colleagues discovered that average intelligence, for one, does not predict group intelligence. Other factors, such as group cohesion, satisfaction and motivation, are only moderately correlated. What does make a group smarter? Having a few people who are “socially sensitive;” that is, members who tend to be more open and receptive.
Malone and his crew are taking results like that and mapping them onto a “genome” of group intelligence. Workplaces and organizations are taking note, but so are news sites and online government initiatives interested in harnessing the intelligence of readers and constituents. Have an under-producing team or a comment thread full of flamers? Time for some group dynamic gene-splicing.
It might be too far a reach to translate MIT’s work to the the group dynamics of the internet as a whole – at least in the near future. But perhaps one day we’ll be building algorithms to maximize collective intelligence rather than just personal relevance.