Two approaches are described with their philosophic implications: building a model as preliminary to simulation game design, and discovering a model after the game has been developed. A third alternative presented attempts to span the chasm created by the first two and integrate both design processes. (VT)