==More People, More Inertia. ==

As a student I went along to inaugural meeting of the Cambridge Anarchist group; I'm not sure exactly what I expected, maybe people working together without a centralized control, maybe a system and mind set freer from power politics and corruption.

One person was leading the group. The focus was on 'what do we need to do'. I suggested that the circle of around 50 people split into small groups of four to six so that more discussion would take place. There was a general agreement that that wasn't a bad idea, but then nothing happened. The group went on in its large circle, one person preaching to the (mostly) converted. Nothing happened. I do believe that CollectiveIntelligence can be a great thing, but without creative and intelligent structure more people just makes more inertia. More talking, no doing.

==Blind Spot ==

The key problem with CollectiveIntelligence is coming up with a structure that gives greater intelligence than any one individual. It is a tough problem, a BlindSpotProblem. As a hobby I write science fiction. In some of my earliest fiction I tried to describe an AlienIntelligence, an alien brighter by a couple of orders of magnitude than myself. I didn't succeed. Sir Arthur Conan Doyle succeeds in making Sherlock Homes 'brighter' than himself, by trickery. He works back from solutions to puzzles, and then to boot gives Sherlock a bumbling foil in Watson to show up his dazzling intellect. Without trickery you have to work very hard indeed to create a mindset for a more intelligent alien. All I could do was show the alien very adept at learning new things, very intuitive. His intelligence certainly wasn't out of this world.

If we have great difficulty picturing what greater intelligence is, how can we ever work towards it as a group? I think all we can do is extrapolate from lessons learnt elsewhere.

==Communication ==

At the root of intelligence is emotion. Emotion is what focuses your attention. Without emotion intelligence is static (cf DeepBlue).

The main barrier to communication between people (and hence to collective intelligence) is emotional.

Some systems for group work, such as RobertsRules for committees are designed to sidestep typical EmotionalPitfalls to communication. In RobertsRules you never accuse anyone of lying to make their point. The system is based on an assumption of good faith negotiation.

==Structure ==

Large groups can come up with designs faster and more varied than smaller groups or individuals. One project concept I have toyed with is the collaborative design of a micro satellite, a small homebrew answer to Hubble put together by amateurs. Some of my friends launch amateur rockets as a hobby, they can reach low orbit. Some cool ideas for the payload that reduce weight significantly (trading time and cool algorithms for costly fuel and over-bulky sensors) would reduce the build cost to a few thousand dollars. Some technical documents from Phillips dating from the 1960's put me on the track of how to do this using modern components. Back then they were launching satellites with ferrite core memories====

==

For that project to work the review process would need careful design. Different groups could be tasked with BrainStorming on different components of the design - crucially the component designs and integration of components would need feedback from people with key knowledge and experience - e.g. that you can't use oil as a lubricant for the joints - it evaporates in vacuum, that you need shielding for your circuits, or your bits will start hopping with the radiation levels when in sunlight.

But this kind of collective intelligence could function. The reason why is that reviewing takes less time and energy than design. Over time the experience of experts on the review panel would be codified and propagate out to the designers.

The satellite design project is a classic for a HierarchicalOrganisation. Expertise at the top ripples down, energy, originality and creativity flows up from the bottom. You have to start thinking of InformationAsaSubstance that flows to design information structures that will work well.

(See also InformationHeatExchanger for an alternative to a strict hierarchy).