My trip to San Francisco was timed to ensure attending one of the “thematic” Foo Camps – this one on Collective Intelligence, held at the GooglePlex. The concept of an emergent intelligence from a group is not new. It is the basis of democracies and also more recent books such as “The Wisdom of the Crowd”. However, the evolvement of the web has made engaging a huge amount of users not only possible, but incredibly easy. And the ability to monitor nearly every single action they take moves the question from “what if you could harness the collective”, to “what do you now do with all this data”?

From what I gathered at the conference, it is very early stages for the web crowd on discussing these ideas. The fact that there was a very small representation of sociologist or anthropologists that no doubt have a greater understanding of the general concept is indicative of typical technologists having difficulty engaging traditional experts and gathering experience being applied to new techniques. There was an interesting mix of business/market analysis, developers, designers, and technophiles. Because of this broad representational range there was a difficulty in having a common conversation since the taxonomy of collective intelligence is not well understood.

I’ll go into this more in follow-up posts, especially as applied to Geo, but there are many varied aspects of CI: collaboration, collection, explicit and implicit, superlinear vs. mechanical turk. Each has incredible power and capability, but it’s ineffective to apply a broad brush stroke of design and understanding across the entire gamut.

Kim Rachmeler, VP Customer Service at Amazon, Inc., summarized it best in her “award-winning” quote:

The network knows what the nodes do not.

There were a number of notable projects, a few summarized here:

Beth Noveck showed the very excellent Peer-to-Patent review system that is leveraging public analysis of patents in order to help alleviate the mess the USPTO is currently in. Blaise Aguera y Arcas had an excellent demo of SeaDragon, scaleless image zooming – and hoping that we can do something similar with OpenStreetMap. More CI based, Eric Horvitz had interesting concepts on selective sharing of user-gps tracks for route and traffic prediction building. Humans as sensors.

There was also discussion about “reputation”, which really is like saying Collective² Int. – identifying experts and then using their intellingence to solve problems or make suggestions. Which gets to more underlying questions about whether a “crowd” can really be smart, or if it is just extra power behind a few smart kernels.

I did find it particularly humorous of the potential the underlying purpose of the conference on “Collective Intelligence” was in fact to get together a number of intelligence people to garnish their ideas.