Monday, October 02, 2006

Diagrams and Networks

Responding to Paul Ellerman:

I was actually pretty careful with the diagrams, though on reflection I considered that I should have, for the network diagram, used the standard connectionist (neural network) diagram. See eg. this.

Now in fact there are even in networks people like myself, Will Richardson and Dave Warlick, and they are sometimes called “leaders”. But from the perspective of a network, what makes an entity emphasized in this way is the number and nature of the connections it has, and not any directive import. People like Will, Dave and I stand out because we are well-connected, and not (necessarily) because we are well informed, and certainly not (necessarily) because other people do what we say they should do.

There are in fact two major ways that such people can emerge in a network:

First, as a consequence of the power law phenomenon. This is discussed at length in the discussion of scale-free networks. It is essentially the first-mover advantage. The person who was in the network first is more likely to attract more links. This is also impacted by advertising and self-promotion, phenomena I would not disassociate with the list of names you provide.

Second, as a consequence of the bridging phenomenon. Most networks occur in clusters (prototypes of Wenger’s communities of practice) of like-minded individuals. Philosophers of science, say, or naturalistic poets, or the F1 anti Michael Schumacher hate club. Some people, though, have their feet in two such clusters. They like both beat poetry and the Karate Kid. And so they act as a conduit of information between those two groups, and hence, obtain greater recognition.

In neither case is the person in question a ‘leader’ in anything like the traditional sense. The person does not have ‘followers’ of the usual sort (though they may have fans, but they most certainly don’t have ’staff’ - at least, not as a conseuqnce of their network behaviour). They do not ‘lead’ - they do not tell people what to do. At best and at most, they exemplify the behaviour they would like to see, and at best and at most, they act as a locus of information and conversation .

In the diagram, this difference is represented by depicting the ‘traditional’ leader and group as a ‘tree’, with one person connected to a number of people, while at the same time depicting the network as a ‘cluster’, with many people connected to each other.

Pushing the model back to the ‘leader’ mode suggested by this item would push the diagram back to a ‘cluster’ or ‘hub and spoke’ model. See this or this — which I was anxious to avoid. Not because it doesn’t depict a network (it does, technically, a scale-free network) but because it depicts what I would call a ‘group’ dominated by ‘leaders’, where the leaders have a directive function.

No comments:

Post a Comment

I welcome your comments - I'm really sorry about the moderation, but Google's filters are basically ineffective.