Sunday, May 21, 2006

Some General Off-the-Cuff Reflections on Stigmergy

Below are some general thoughts inspired by Samuel Rose's response to my article, Stigmergic Collaboration.

* Agents of a stigmergic system don't communicate directly, but rather via the stigmergic medium. The nature of this communication would thus not be one of turn-taking conversation, but rather, “I encounter something in the system that provokes x response from me”. However turn-taking conversation would accompany such efforts: "hey Joe, check out my wiki... can you contribute x" etc, and of course discussion relating to contributions. (Have you ever watched ants when they meet each other on the trail? They almost always stop and touch antennas.)

* The local is the domain of the individual agent, while the system wide level likely to be beyond the individual's comprehension (or ability to conceptualize at any given moment). This resonates when thinking about the origins of stigmergy - how can a single termite know where to put a wee bit of mud so as to contribute to a fantastically complex structure - it doesn't, it just responds to its local conditions – a particular pheromone in an existing bit of mud elicits the response of ‘roll mud ball with x pheromone and deposit here’. Although a whole might be distributed beyond the reach of the individual (I know is already too much for me to know the ins and outs of all its corners), perhaps knowledge of the whole might be naturally broken up into chunks relating to user groups. - If you want to know about x, find the group that stigmergically develops and maintains x and ask them collectively. (An idea which might be worth better supporting in wikis etc – this could also feed Reed’s Law.)

* Filtering... hmm, seems like that is what stigmergy kind of does naturally. That is, when the local gets stigmergically encoded, a filtering process takes place before the encoding does - how do I know what I will encode? I will encode that which is the most meaningful for me in that context. Thus aggregated meaning must emit gradients of reputation which attracts more contributors. - Seems this might also be an area worth exploring: how to build the increase of positive feedback and attract more contributors into today's stigmergic collaboration systems.

* Intuitively I feel there is something big, complex and hairy lurking behind the combination of network theory (mathematical & social) and stigmergy. Perhaps this is humanity’s real contribution to the medium - not that swarm insects aren't social and networky, but let's face it, we have the internet ;-)

* Finally, as I paste all this jabber into my private wiki, I realized that stigmergy doesn’t have to be social, that is, the communication can be with oneself…


Post a Comment

<< Home