Back From CI Foo Camp

I'm back from the Collective Intelligence Foo Camp - it was a great experience. The format of the meeting, which is essentially a collection of spontaneous discussions, informal presentations and demos, really made me think about how to get the best value out of other, more traditional meetings (like ICWSM!). I met a bunch of great people (many of whom I had been reading in the blogosphere) and participated in some very engaging and rich discussion.

I plan to write a number of posts about the camp, but perhaps the best place to start in terms of covering the meeting is at the end, when the group discussed definitional issues of collective intelligence. Definitions are always problematic as they can often define away interesting related areas that aren't captured by some 'pure' or elegant description. Thus I think it worthwhile summarizing some of the dimensions which surfaced in the discussion.

• Parallelism - a key aspect of collective intelligence is the notion that there are many agents, all of whom are working in parallel in some way.
• Homogeneity - are all the agents identical?
• Systemic effects - does the system that networks the agents contribute in some way to the quality or form of the result. Another way to think about this is: are all the agents doing work which is simply summed together linearly, implying that a single agent, with enough time, could effectively carry out the same task.
• Efficiency - does the system result in quicker solutions.
• Intelligence granularity - is the resulting behaviour producible by the component agents? In other words, is the emergent intelligence attributed to the system at a higher level than that of the agents? A single ant (which we assume is not intelligent to any great extent) couldn't devise an algorithm for the efficient gathering of food, but the colony can.

Of course, for definitions one could do worse than refer to Wikipedia:

Collective intelligence is a form of intelligence that emerges from the collaboration and competition of many individuals. Collective intelligence appears in a wide variety of forms of consensus decision making in bacteria, animals, humans, and computers. The study of collective intelligence may properly be considered a subfield of sociology, of business, of computer science, and of mass behavior — a field that studies collective behavior from the level of quarks to the level of bacterial, plant, animal, and human societies.

The use of the term 'collaboration' is interesting here as - one might argue - collaboration requires intention.

Comments

[Lukas Biewald]

Matthew, it was fun to meet you in person. Looking forward to your posts about the camp.

[Alex Tolley]

"...single ant...couldn't devise an algorithm for the efficient gathering of food, but the colony can."

The colony does not devise anything. The efficient gathering of food is the result of evolution working on ant genes that create behaviors, that in the colony aggregate, result in a competitive search algorithm. Thus all the behaviors are in the ant and the algorithm is nothing more than the emergent behavior of the rules acting in parallel.

[Matthew Hurst]

Alex - um, yes - I think that is what I was saying.