This is really a posting about automating the creation of light weight groups.

I wish I had tools to filter some of the feeds that I read so that I could filter them better. For example if I subscribe to Mr. Diverse Interests I want to setup filters so that I get those of his postings that overlap my diverse interests. Doc. Searls for example thinks about lots of stuff; but I’m only interested in a portion of that stuff.

One idea, among many, is draw on the work of the tagging hordes. Intersecting my the tag cloud of my interests with the tag cloud of others. An approach like this helps with the UI problems. The filtering software could present me with a source’s tag cloud and I can toggle those for white/black listing. The neat thing that’s hidden in here is that we can probably build two models – one of the reader and one of the writer – and then intersect them. The model built can, of course, draw on various sources of the meta data: tagging, text analysis, link graph analysis, semantic web, etc. etc.

I’m already doing a certain amount of this using pubsub. When I notice a turn of phrase I add it to a query I have at pubsub and see what it turns up. That is, of course, similar to what Amazon is doing with their statistically improbably phrases.

But back to group forming.

This would be particularly useful for some of the planet sites that are trying to aggregate postings from all the loud mouths who are thinking in public about a particular problem domain. The trigger for this realization was planet identity. Which is an attempt to aggregate all the very very diverse voices that are talking about the internet identity problem.

I call these pools of loud mouths the mob: i.e. the set of disaffected intelectuals and artists who actually provide the most substanative analysis of an emerging topic. The identity bloggers is only one example. These are very tenous groups. They have common cause in so far as they are working on the same problem; but in many cases they have no common cause around how to solve that problem. Such groups – ones with a common problem but very little consensus about the right solution – are of course extremely intereting.

All the other planet’s I’ve encountered have, go in, agreed on a common body of work they are engaged in. Planet Apache for example is a tiny subset of the Apache committers; it’s a minor complement to the real thing that brings that group together.

Mob sites are a different kettle of fish. The mob members aren’t organized into a coherent group. Recall that my definition of a mob says they are “disaffected.” The benefit of going to the mob for information about a topic arises out of their collective lack of loyality to any given solution.

Many years ago my wife was in charge for a while of an “Artists Collaborative.” A phrase I consider the penultimate oxymoron.

So the really fun idea here is that mob sites might be brought into existance without resolving that disaffection. It looks entirely straight forward that one can create very very light weight collaboration forums from out of the work of entirely disaffected artists that happen to be working on that problem. Distilling the liqour of the mob’s conversation on a given topic without forcing the mob’s members to join a drinking club.