The Wisdom of Crowds
I'm finding that The Wisdom of Crowds is upsetting a lot of presuppositions I've long maintained about, well, the stupidity of crowds. For example, a political scientist at the University of Michigan has done some experiment using computer-simulated problem-solving agents that demonstrated that a group that consists of all smart agents does worse than a group that consists of some smart and some not so smart agents.
There was a conversation that a bunch of us were having at the Rapid Prototyping session at Digifoo a couple weeks ago about the benefits of having some less intelligent or less informed people in a group. I think this works because you have to slow down to explain to these people what you are doing and why -- which is never a bad exercise to engage in. As the group becomes better informed over time, it is important to throw in some new, uninformed people into the mix, so you can get a fresh perspective.
Don MacAskill argued that having someone who generally and actively disagrees with what the group is doing is the most beneficial group member of all, since they provide a completely different perspective, insure against groupthink, and also because (this is something I hadn't thought of) the other members of the group anticipate that this devil's advocate will counter them whenever they present an idea, and so will prepare better, question themselves in advance, and make sure they can defend their idea in the next meeting.




