Computer simulation and the features of novel empirical data

In an attempt to determine the epistemic status of computer simulation results, philosophers of science have recently explored the similarities and differences between computer simulations and experiments. One question that arises is whether and, if so, when, simulation results constitute novel empirical data. It is often supposed that computer simulation results could never be empirical or novel because simulations never interact with their targets, and cannot go beyond their programming. This paper argues against this position by examining whether, and under what conditions, the features of empiricality and novelty could be displayed by computer simulation data. I show that, to the extent that certain familiar measurement results have these features, so can some computer simulation results.