Introduction: Artificial Anasazi is a custom computer simulation developed at the Santa Fe Institute and the Brookings Institution under the direction of Dr. George Gumerman, who is now the director of the Arizona State Museum. It is designed to simulate the residence patterns of the prehistoric Anasazi who lived in the Long House Valley, Northern Arizona. Unlike earlier computer simulations of archaeological phenomenon, Artificial Anasazi is an agent-based simulation. Agent-based simulations are different from traditional simulations based on Systems Theory in that each actor from the real target population is represented by an autonomous agent in the simulation. All agent-based simulations, including Artificial Anasazi, consist of three basic parts: the simulated environment, the agents, and the rules that govern the behavior of the agents. The following is a breakdown of the Artificial Anasazi program structure:

The Environment: The simulation environment consists of a 80 x 120 square grid, with each square corresponding to approximately one hectare. Each square in the grid is given a maize potential based on its environmental attributes, sand dune status, and zonal soil quality. This potential changes each year based on the paleoenvironmental reconstruction of historical conditions from A.D. 400-1450.

The Agents: The scale of analysis is the household, with each household having both a residence site and a farmland site. Residence sites may be shared by households, but farmland sites may not. At the beginning of a simulation run households may be started at know historical locations and densities or generated randomly by the computer. Two basic events trigger the movement or establishment of a new household: (1) if the household cannot support itself on its current farmland site it will establish a new site in a better location, and (2) when a household fissions after it has been in existence at least 17 years (when a child is expected to have reached marriageable age). It will then fission with a certain probability each year, this probability being an adjustable parameter of the simulation. A household expires when it reaches the "death age," another adjustable parameter of the simulation.

Movement Rules: There are two steps to movement. In step one a farmland site is chosen and in step two a residence site is chosen. Candidate sites are identified by applying sufficiency criteria and then one is selected by applying maximization criteria. Four different household types with different movement rules have been created.

An Experimental Test of the Simulation: An experimental test of Artificial Anasazi was conducted by myself in the fall of 1998 as the basis of my Master's Thesis at the University of Arizona. In brief, the experiment involved conducting a series of runs using the Artificial Anasazi simulation program and then comparing these results with the data from the real Long House Valley in order to test the validity of the simulation. Settlement patterns and population growth were the only two variables that could be tested using the real world data. Significant differences between the simulated and the real patterns were observed, leading to a number of important conclusions about this simulation and the possible contribution of agent-based simulations to the field of archaeology in general. My Thesis is available from University Manuscripts Incorporated for a small fee, if you are interested in learning more.