Computer simulation has arrived big time, in everything from movie effects to election forecasts; here's an expert's overview. Beginning with an attempt to predict the outcome of a Super Bowl, Casti (Searching for Certainty, 1991) chose a popular but fairly sophisticated computer game program to play a series of simulated games between the 1995 opponents, the 49ers and Chargers. He uses his results (which suggest either that the actual game was a fluke or that the program is flawed) to make several basic points about computer simulations and models. They can be predictive or explanatory; perfect fidelity to the real world is not the sole virtue; and they are often most useful in analysis of complex phenomena that in the real world are either dangerous to meddle with (such as the flight patterns over a busy airport) or very rare (the collision of a meteor with Earth). He then gets down to specifics, describing programs to analyze language, to generate artificial ``life,'' or to forecast the weather. A long and fascinating chapter is devoted to limitations and paradoxes that limit our ability to turn every problem into an easily computable simulation. We get close-up looks at a system modeling the traffic patterns at rush hour in Albuquerque; at neural nets, which attempt to simulate the structure of the human brain; and at ``Sugarscape,'' which extracts basic economic principles from a pure supply-and-demand environment. Casti's subject sometimes leads him into esoteric territory, but he tries to keep it down-to-earth with examples from real life. At the same time, he is not afraid to plunge into such deep waters as Gîdel's Incompleteness Theorem. There is the occasional mathematical formula, but readers without advanced math should be able to follow the argument--especially because of the excellent use of illustrations and diagrams. A very solid and useful discussion of the theory and practice of computer modeling in the physical and social sciences.