The Dissemination of Culture Axelrod 1997

Axlerod is one of the fathers of agent based modeling. Him and Schelling.  Yet after thirty years, he still has to make a careful case to sociologists, who remain skeptical.  

This paper, presented to the journal of conflict resolution, walks carefully through the idea of agent based modeling, how it works, and how it can contribute to an understanding of society.

He spends the first part of the paper essentially covering his ass; addressing common misconceptions and being careful that no one thinks he is making unreasonable claims.  This care is a sign of how inclined (naturally) sociologists are to misunderstand theoretical work.  
*"It should be understood  that there is no connotation that a uniform culture exists with a single society."
*"because the proposed mechanism can exist alongside other mechanims, it can be regarded as complementary with older explanations rather than necessarilty competing with them."

"Anthropolgists...The diffusionists treated a biven culture as a set of distinct traits, each of which could be passed along to another culture.  More recently, most anthropologists have amphasized the interconnections between the many traits that made up a culture, viewing culture as a system of symbols by which people confer significance on their own expperience"  -very concise review

The particular work he goes into shows some nice features of the method.  I particulalry liked one part where his model had a 'wrong' prediction, and, fromthe perspective of understanding, that was jsut as instructive as if it had been right, because it gave him a question to ask about society, namely, what are the relevant factors to take into account when modeling a given social phenomenon, (and why was a particular factor the wrong choice)?
And whe he is confronted with something that is counter intuitive it is an oportunity to learn something deeper about the dynamics of the model. He actually spends most of the paper exploring the counter intuitive result.  It was itneresting, and the explanation was good.  But again, the parallels to society are not obvious.  I think the work is done responsibly, but sometimes it is more apparent that he is exploring the models than society, forest for the trees kind of thing.

This model has 'culture' on a lattice, each location on it has n kinds of beliefs, m possible values for each belief.  his conclusion is interesting (and hard to apply meaningfully to society: "the complexity of the culture needs to be differentiated to account for the number of stable regions.  Having more features (i.e., dimnesions) in the cuture actually makes for gewer stable regions, but having more alternatives on each feature makes for more stable regions"

Changing the range of interaction doesn't change the fundamental dynamics.  This is important to check to accomodate people who are concerned about the realism of the square lattice.

he ends up spending most of the time trying to figure out//explain why a bigger map leads to fewer stable regions.  you would expect that the bigger a world, the more diversity.  but somehow, a bigger map leads to major decrease in diversity.  The explanation: stable regions are formed by borders along which different groups have nothing in common.  bigger maps take much longer to converge, during this time many 'ideas' are propagating.  This means that, along a given border, the probability that there will be a 'breach', or one single digit that it shares, increases.  Again, whether this applies to society is not something that can be tested easily, the metaphor is a little tough to map.

"The mere obsevation that a practive followed by few people was lost does not necessarily mean either that the practive had less intrinsic merit or that there was some advantage in following a more common practive. ... even unbiased changes in adherence lead to a less common practive disappearing simply because it is more vulnerable to rendom fluctuations"

He also just breaches dynamic measures, getting into situations where there is no stable convergence.

Definition:
"It treats culture as the attributes that social influence can influence"

Circular, but not in the traditional sense.  I like it.  Also not immeidately controversial, thought I haven't thought about it really.