Sim City social science

Computer simulations of artificial societies are leading to social science that seems illuminating, and conceivably might even be useful at some point. At least that’s what an account by Jonathan Rauch in The Atlantic Monthly suggests. Some interesting results:

• “Power laws” order a great many social phenomena: the size of firms, population of cities, sales of best-sellers and so on tend to follow a pattern whereby the figure for the top firm or city or whatever is roughly twice that for the second, three times that for the third, a hundred times that for the hundredth, and so on. (A quick look at my logs suggests that something of the sort also holds for visits to the pages on my website.)
• Such unintended stable patterns are commonplace, and where they exist they rebut the common impression that behind-the-scenes machinations are needed to explain some feature of the social world. For example, they tend to show that moralizing explanations for common patterns of communal relationships (“ingrained racism” and whatnot) are unnecessary. If you do a simulation that divides cyber-agents into different “races,” and gives members of each race even a very moderate preference for being near some of their fellows, you’ll end up with rather strict segregation.
• Indeed, it seems that it might be as well to let that process go forward, especially if communal tensions may become serious. Simulations suggest that police forces can prevent or stop genocide if the conflicting groups live in separate enclaves, so the problem is policing the borders. If the two groups are all mixed together, though, it appears that even flooding a society with UN peacekeepers—one for every ten inhabitants—will not prevent an outburst of genocidal violence sufficient to overwhelm the peacekeeping forces if conditions are such that the violence will likely happen absent their presence. The point can be extended. Genocide (luckily) is a comparative rarity, but lesser degrees of communal ill-will are very common. Could it not be that friction and bad temper will be minimized, and the need for government control reduced, if people are allowed to establish for themselves what degree of interaction with other groups is desirable?

The strong point of the new method is that it treats society not as one big thing to be studied at large in hopes of finding ways to make it do one thing or another, but as a series of situations arising out of the dealings of myriad agents, each with his own habits, goals, and limited perspective. For some reason Rauch thinks the success of the method proves that Mrs. Thatcher was wrong when she said “There is no such thing as society.” To me it seems more reasonable to interpret what she said as pretty much a restatement of the method’s presuppositions.

Be that as it may, Rauch is a liberal and thus a committed social engineer. The new science confirms that large-scale coercive interventions from the top don’t work as intended. His response is to hope that new knowledge will enable him and his fellow members of the Brookings Institute to construct cleverer interventions to manipulate the system in the directions they think good.

Maybe that approach will work and be beneficial, but maybe not. Here’s a different proposal: the new science says that social problems mostly have to do with unintended patterns that arise out of the interactions of myriad agents, each with his own motives and understandings. If that’s right, then the solutions are likely to involve affecting the motives and understandings of the agents by introducing principles of action that aren’t obviously rational from the point of view of either social technologists or individual utility-maximizers, but nonetheless have achieved credibility and attracted loyalty because experience shows things turn out better if people live by them. The usual expression for such principles is “traditional moral standards.” So if Mr. Rauch or anyone else wants to make a contribution, why not give up liberalism altogether and start pushing traditional morality?

(Note: I also touch on traditionalism and this “new science” stuff in my review of Kevin Kelly’s Out of Control.)