The Finnicky Wisdom of Crowds

I was tickled to see Crooked Timber, one of my most-​​frequented blogs, report on Cosma Shalizi reporting on Scott Page’s analytic work on diver­sity. Back home in Ann Arbor, the com­plexity reading group is meeting on patios to discuss Page’s new book “The Difference”, and I gather Scott (if I may call him Scott) has been showing up to some of the events. Or perhaps he was there for the sangria and decided to stick around. Scott (and I suspect I may call him Scott, I’ve only briefly met him but he seems like a super nice guy) is one of the most engaging pre­sen­ters I’ve seen and a crack­er­jack thinker and I’d love to be in on that dis­cus­sion, whether with or without him. And hey, I like sangria. I’ve not read the book but Cosma Shalizi does his usual great job of boiling down to the sauce of essence, and the basic pitch seems to be: multiple diver­gent weak hueris­tics applied serially can solve problems with multiple inter­acting factors better than a single strong hueristic. Thus groups composed of diverse people can be more suc­cessful than homoge­nous ones. With the caveat that there must be some degree of agree­ment on what the goals and success criteria are. Cool.

As a perhaps-​​interesting test case of the idea, Wired has a big fat report up on the results of their “Assignement Zero” project. The idea was to let anybody con­tribute to an effort to generate a large body of high-​​quality reporting on a subject, that subject being the ability of dis­trib­uted crowd­sourcing to produce high-​​quality work. Apparently it played out a little rocky. Apparently there was a great many lessons learned, a certain fraction (esti­mated at between 1 and 3 quarters of the total) of decent material produced, and appar­ently a lot of the problems related to people not knowing or agreeing on… what the goals and success criteria for the project were. Interesting.

leave a comment