Re: [Swarm-Modelling] social dynamics

[Darren Schreiber]

The next versions of the model will do exactly what you are describing. 
 I am basically going to take the current model and incorporate my 
advisors' theories about persuasion (John Zaller "The Nature and 
Origins of Mass Opinion" 1992).  Later, I will use a major result from 
my brain imaging work (supporting the contention that two distinct 
brain systems with different computational properties are responsible 
for some of the confusing features of public opinion) to motivate a 
computational model of political cognition.  I will either use Ron 
Sun's Clarion model ("Duality of Mind: A Bottom Up Approach Toward 
Cognition" 2001) or something similar.  This computational model will 
replace the very very simple five rule system that drives my current 
party model.

While I have lofty ambitions for this party model, I should note that 
its simple incarnation has lots to offer.  One of the main benefits of 
agent-based modeling, in my mind, is that we can often show really 
complex and interest behavior emerges from a very simple model.  I 
strongly believe that if we stray from the KISS principle (Keep It 
Simple Stupid) it should only be when we have demonstrated that simple 
methods are inadequate.  And, even our extensions to models should be 
extremely parsimonious and high leverage (explaining a lot with very 
little.)

I also agree that changing the agent's motivation to be in coalitions 
is a laudable direction.  My friends and I love to quote E.E. 
Schattsneider's phase "the organized beat the unorganized."  But, I 
think agent based modeling would be a very nice way of testing the 
conditions under which that is true.

        Darren


On Wednesday, April 23, 2003, at 02:47  PM, Otto Cordero wrote:

> Darren,
>
> I have reviewed your article and I have found some very interesting 
> issues
> on it, it is mostly what I was looking for. If I understood well, the 
> agents
> join in parties according to their political preferences, but those
> preferences are static. I would like to see what happens if the 
> criteria
> (upon which two agents decide to join together or to split away) 
> changes
> over time, probably this criteria would be conditioned to the state of 
> the
> environment or to the inner state of the agent. Another thing that 
> caught my
> attention is that the agents' goal is to form the biggest coalition, 
> so in
> this sense the coalition is an end by itself. I would rather like to 
> see
> agents forming groups as a medium to achieve their goals.
> I would work on this and let you know as soon as I have some results,
>
> Thanks,
>
> Otto.
>
> ----- Original Message -----
> From: "Darren Schreiber" <address@hidden>
> To: <address@hidden>
> Sent: Wednesday, April 23, 2003 1:48 PM
> Subject: Re: [Swarm-Modelling] social dynamics
>
>
> > My model of political party formation does essentially what you are
> > talking about.  Individual agents look around in the "issue space" to
> > see what other agents have similar political views to them.  They form
> > a coalition with their closest neighbor.  Recursively, the coalitions
> > look to see what other agents or coalitions are closest and form
> > super-coalitions.  The end result is political parties composed of
> > coalitions of coalitions.  I haven't been working on this project for 
> > a
> > few years, but an explanation, graphics, and the paper are located on
> > my website: (http://www.bol.ucla.edu/~dschreib).
> >
> > I'm finishing my dissertation right now using brain imaging to study
> > political thinking.  My next step is to merge the insights from my
> > neural microfoundations into the macro-scale party model and show how
> > ideology is an emergent property of political dynamics.
> >
> > On a much less serious note, the cocktail party model
> > (http://zia.hss.cmu.edu/econ/homework00/2/cocktail.html) that Troy
> > Tessier and I wrote at the Santa Fe Institute a number of years ago
> > also has people forming social groups.  Agents search the room looking
> > for someone interesting to talk with.  They talk until they get bored
> > or someone more interesting passes by, at which point they move off in
> > search of more interesting conversation.
> >
> > I am similarly interested in anyone pursuing this line of work, so 
> > keep
> > me informed.
>
>
> _______________________________________________
> Modelling mailing list
> address@hidden
> http://www.swarm.org/mailman/listinfo/modelling