Computer Models as Play

There is, indeed, an "art" to worldplay in the social sciences that fuses narrative with analytical technique. There is also a kinship with the arts in the relationship between imagined world and reality, a point brought home by political scientist and ellow Robert Axelrod. In the early 1960s the teenage Axelrod won the Westinghouse kience Talent Search for a very simple computer simulation of hypothetical lifeforms behaving in an artificial environment. Ever since, he has worked on the application of computer simulation to biological, economic, political, and social systems, investigating such diverse phenomena as the arms race, the development of ethics in children, and the evolution of decision making.

For Axelrod, the imagined worlds generated by simulation are, in at least some respects, equivalent to the pictures he used to watch his father paint in watercolors:

Dne way to think about this is, he is putting on the canvas his representation of what he sees. Now is that real or not? Well, it's not a photograph...but it certainly looks more or less like what he was looking at—a tree, for instance. So it bears relation to reality; it's his interpretation of which aspects of reality he wants to emphasize and how he wants to emphasize them. And to a substantial extent I see my model building, which I've done throughout my career starting with that high school project, as analogous to that. I'm making my picture of the world as I look at it, and inviting other people to look at it.

Axelrod's picture of the world is scientifically rigorous. "Computer simulation has very clear rules," he said in interview. "It's very explicit. And you have to tell the computer everything, so in fact everything is done as rigorously as a nBtlthenemj proof" But the simulation is also personal in the way that art is personal. Y You can n change the rules or constraints as you wish. And much like the child at play in imagi¬ nary lands, the researcher derives satisfaction ft-om this creative control. "If you m makiake a mistake," he observed, "you can undo it" by recoding the assumptions that govern the behaviors of hypothetical animals or other agents in play.

In that sense, at least, computer simulation proves little different than a self-if-constructed game. As Axelrod recalled it, the Westinghouse project often felt like gan playing. In his adult career the line between work and play has, similarly, be)een "very blurred ... I might say all my work is play—but certainly my research is largebly play, especially when successful."