In recent years, computer simulations have made enormous progress, even to the point of advancing beyond the limits suggested by their name. The original concept of a computer experiment that "simulates" a real experiment has been replaced by approaches that exploit the greater flexibility and greater information available in a computer simulation. Past examples of this trend include the development of non-physical dynamics (cluster methods), renormalization analysis of the system being simulated, or a histogram analysis to get information on experimental conditions that were not even simulated.

In this talk, I will present a simple new approach to simulation and analysis that extracts a surprising amount of information about the thermal properties of a system from a consideration of the matrix of possible transitions that the system might undergo.