We use models to answer questions and solve problems. Often these problems involve designing solutions in terms of physical form, functional capabilities, and policies intended to incentivize or inhibit particular behaviors. The key point is that models are intended to support problem solving.
Initially, our models are limited to visualizations, perhaps just sketches. These visualizations may evolve to become more elaborate and perhaps interactive. They enable exploration of connections among entities and how relationships among entities work. Such visualizations are models. They express the problem solver's perceptions of what phenomena matter, how they interact, and key trade-offs.
Sometimes, a good visualization is all that is needed. The problem-solving group's discussion and exploration of the visualization lead to a conclusion on how to proceed. In other situations, deeper explorations are needed. These explorations may involve more formal representations of phenomena and relationships. Deep computation may be warranted, but perhaps used sparingly.
The discussions and explorations usually lead to creative suggestions for possible courses of action. All the creativity comes from the group of problem solvers, not computers. In other words, policy flight simulators seldom fly themselves. Instead, computers provide the means to explore the implications of seemingly good ideas. Bad ideas are rejected and good ideas are ...