Only a small fraction of human history is known through texts. For the rest, archaeology is the main source. By examining ruins, artifacts and remains, archaeologists have painstakingly constructed a series of pictures showing human societies as they existed thousands and even millions of years ago. It is much more difficult, however, to determine the processes that produced and changed these societies. Researchers are still struggling to understand the long chain of cause-and-effect (and chance events) stretching from our hominid ancestors of four million years ago--small bands of upright-walking primates with no stone tools and scarcely any conversation--to the communities and cultures we see around the world today.

With the advent of computers, archaeologists began to experiment with simulation as an aid to exploring human prehistory. The logic is simple: you program the computer to mimic processes such as population growth and resource usage, then see how well the software's predictions coincide with the archaeological record. An early example is the well-known attempt in the late 1970s to examine the collapse of the Classic Maya civilization, which dominated a vast swath of Mexico and Central America from A.D. 300 to 900. Led by researchers at the Massachusetts Institute of Technology, this effort looked at the relations between variables such as total population and the rate of construction of Mayan monuments. Because the study considered the variables in aggregate form, however, it could not provide information on spatial relations--for instance, which areas of the Mayan territory had the highest agricultural production.