Posted by
Michael Agar on
Jun 06, 2006; 4:27pm
URL: http://friam.383.s1.nabble.com/FW-SFI-Seminar-Complexity-Parallel-Computation-and-Statistical-Physics-tp521927p521946.html
Interesting idea. In various and sundry experiments with drug ABM's
it seemed like assumptions were being made that were candidate
assertions for a "canonical society." For instance, we set up
networks on a power law distribution assumption following Barabassi.
We assumed that openness to change among agents followed the normal
distribution shown by Rogers' work on Diffusion of Innovation. Do
these tend to be how "normal" small societies organize themselves? If
so, why is that? If so, in what kinds of social ecologies do they
depart from it?
Then there's an interesting connection between ABMs and the robust
trend across many social and psychological theories that a theory has
to be "trifocal." Agents are the centerpiece, then a level down to
model their knowledge and rules, then a level up to observe the
system that they create on the one hand and that effects them in
turn. Is that a minimal requirement for a unified social theory?
Then there's the natural selection principle. Some sort of co-
evolutionary mechanisms would seem to be required, but they'll have
to be different from the classic Darwinian. For instance, human
agents are telic, they organize around imagined future states. If we
consider memes--a problematic concept, I know, but one that brings
ideas into the picture--reproduction rates can vary from extremely
slow to extremely quick. With memes mutations occur frequently and
sometimes dramatically. Memetic crossover occurs in all kinds of
interesting ways. A unified social theory will have to take all this
into account in addition to natural selection on biological variation
if it wants to explain human social conditions.
Been out of the FRIAM loop for a bit so hope all that isn't a re-run.
A good challenge, Jochen, that phrase. Vielversprechend.
Mike
On Jun 6, 2006, at 4:18 AM, Jochen Fromm wrote:
>
> Where is the difference between steps, "depth" and time,
> if "the depth of a system" is simply defined in terms of the
> number of parallel computational steps needed to simulate it ?
> Depth seems to be just another word for (virtual) time.
>
> Much more interesting is the question if there is a unified
> theory for complex systems in terms of agents and multi-agent
> systems. In psychology and sociology we have a patchwork of
> theories, which arises from the complexity of the research object.
> A complex system is often described by several theories and
> multiple models, depending on the particular perspective. We
> have the psychology of Sigmund Freud, of C.G. Jung, of Skinner,
> of William James, etc. In sociology we have the sociology of
> Durkheim, of Weber, of Luhmann, a few smaller theories like role
> theory and "rational choice theory" and a lot of vague theories
> like Giddens "theory of structuration".
>
> These theories can be correlated to one another if we
> place them in a grid or coordinate system with two axes:
> * historical vs. regular behavior (exceptional vs. expected events)
> * micro vs. macro behavior (low-level vs high-level patterns)
>
> The behavior of a complex system depends neither solely on
> individual events and accidents nor on universal laws.
> Both sites play an important role, historical accidents (see
> for example the principles "sensitivity to initial conditions",
> butterfly effect, frozen accidents, path dependence) and
> regular laws. Likewise, the behavior of complex systems
> depends neither solely on individual microscopic actions nor
> on macroscopic structures, institutions and organizations.
> Both layers are important (see for example the principles emergence,
> swarm intelligence, self-organization).
>
> The most interesting behavior occurs in the center or at the
> middle, if microscopic actions have a strong effect on macroscopic
> behavior and vice versa, or if historical accidents become global
> patterns. An ideal theory would combine both aspects, historical and
> regular behavior, micro and macro behavior by defining universal
> "laws of history" or "theories of emergence". Do you think it is
> possible to discover or formulate such a unified theory? Or at
> least a unifying principle, such as evolution in Biology ?
> Probably evolution is again the unifying principle here..
>
> -J.
>
> -----Original Message-----
> From: Stephen Guerin
> Sent: Monday, June 05, 2006 9:10 PM
> To: friam at redfish.com
> Subject: [FRIAM] FW: SFI Seminar: Complexity, Parallel Computation,and
> Statistical Physics
>
> Has anyone seen any papers on logical depth in the context of agent-
> based
> modeling? I know we could talk about n agents * t steps * a rough
> description of agent and environment complexity, but I was
> wondering if
> anyone's done some more formal work...
>
> -Steve
>
>
>
> ============================================================
> FRIAM Applied Complexity Group listserv
> Meets Fridays 9a-11:30 at cafe at St. John's College
> lectures, archives, unsubscribe, maps at
http://www.friam.org