Darryl Reeves

Symbolic Systems 205

Commentary on Rauch’s “Seeing Around Corners”

During the first few weeks of the quarter we read Wolfram’s A New Kind of Science. The underlying theme of that book was that simple rules could result in complex outcomes. This is an underlying theme of the field of simulation in the social sciences as we have learned this week. From simple rules complex behavior such as that seen in society can result. My issue with starting with simple rules is the potential danger that we might make these initial rules too simple and/or make more of the results than we actually should. This idea will be developed using the “Seeing Around Corners” article written by Jonathan Rauch.

As the article mentioned (and as was discussed in class) the use of simulation in social science using agent-based modeling began when Thomas Schelling created a model of segregated neighborhoods. The simulations contained agents of two types: red agents and blue agents. Rauch first discussed a simulation where 2 of the 4 agents surrounding each agent must be of the same color. If an agent does not have neighbors such that this is the resulting scenario, they move until 2 of their 4 neighbors are of the same color. After running this simulation (enough times so that it is not an anomaly), the resulting scenario is a segregated situation such that the two colors are clumped together with “tipping” at the boundary such that the boundary agents flip back and forth on the border. The same result occurs when agents only seek a situation where 1 of their 4 neighbors is of the same color although this end result takes longer to develop.

It is obvious that this result is interesting. The question is, “what are we to make of these results?” We discussed in class that the rules are made simple to extract out what is important about the phenomena that we are trying to simulate. This is done by removing less important factors to concentrate on what is important in the creation of segregated neighborhoods. But a heterogeneous rule such as a desire to be neighbors with 2 or even 1 other agent of the same type seems to be too simple of a rule. In reality, (as Todd pointed out) it is very unlikely that every person in a community would have the preference of 1 or 2 neighbors that are the same color as them. It is more likely that there will be some with these preferences but there will also be others with the preference of having all neighbors of the same color and those with the no preference regarding the color of their neighbors. What would a simulation with less uniform preferences look like? And how are we to determine the breakdown of how many agents have one preference versus another preference? What if we added more colors into the mix? A more comprehensive picture of possible preferences would make the simulation more complicated but it would make it more substantial as well.

Another missing ingredient to this simulation is the impact of economics, which is a point that I raised in class. Granted, factoring economics into this simulation would make the simulation much more complex. But this is an example of how an oversimplification of the rules can make the resulting situation questionable regarding its significance. It definitely seems to me that majority groups might have a preference to live around people of their same race. But it also seems that economically disadvantaged minority groups do not even get the chance to act on any type of preference because their economic situation forces them to live only in neighborhoods, which they can afford to live. This is often in low-income areas with neighbors who share the same race as them. I do not think this situation is what Schelling’s simulation shows. Disregarding this factor in creating a segregated neighborhood greatly impairs the conclusions that can be drawn from such a simulation.