Kinds of decentralization?

[EMERGE_sl_users@umit.maine.edu (Larry Latour) (by way of VanessaStevens Colella)]

bh@CS.Berkeley.EDU,.Internet writes:
>I think that this points out the weakness of the whole concept of
>central
>versus decentral as an organizing principle.  There are different
>*kinds*
>of decentralization.  It is one thing to allow people to pursue their
>own
>goals.  It's quite a different thing to allow people to work out small
>details about how to achieve someone else's goals.

Your point is well taken about *kinds* of decentralization.  I wouldn't
call it a weakness of the concept, just not necessarily a concern.
We've been having a running discussion here about the difference (if
indeed there is one conceptually) between an intelligent and purely
reactive agent.  Pulling away from people and morality issues for a
moment,  there is this issue of whether or not an individual agent can
learn from and adapt to its environment, to develop its own goals in a
sense.

As an example, Mitchel's ant foraging and termite algorithms follow a
fixed set of rules that don't change as the environment changes.  There
are also a few iterated prisoner's dilemma models (evolutionary
algorithms) where the agents also don't change BUT where the system
learns by changing the mix of agents.

Thoughts?  Is anyone familiar with the AI Intelligent Agent crowd?  Are
they addressing some of these issues?

Larry Latour
UMaine CS Dept.