What does it mean to trust the results of a computer simulation? This paper argues that trust in simulations should be grounded in empirical evidence, good engineering practice, and established theoretical principles. Without these constraints, computer simulation risks becoming little more than speculation. We argue against two prominent positions in the epistemology of computer simulation and defend a conservative view that emphasizes the difference between the norms governing scientific investigation and those governing ordinary epistemic practices.

In the late 1990s, computational technology had advanced sufficiently that astrophysicists were able to construct reasonably high resolution computer simulations of the Local Group of galaxies. These simulations indicated there should be around 250 small satellite galaxies orbiting the Milky Way and Andromeda. In the real Local Group, however, only around 40 satellites had been observed, and only twenty or so more have been discovered since then. Despite this discrepancy in numbers, claims have been made in recent years that the (...) ‘missing satellites problem’ has been solved. Using the examples of the constructed luminosity curve, and hydrodynamic simulations, this paper explores how simulations are used in conjunction with observation to ‘solve’ the missing satellites problem. It is suggested that the simulated universes have sufficient complexity to be understood as worlds in their own right, ones that can be measured and observed. By demonstrating that these virtual worlds are sufficiently ‘realistic’ with respect to observations, astrophysicists are able to make a robust argument for the existence of ‘dark’, non-observable astrophysical objects. Observational and simulated data are combined to demonstrate the plausibility—a term that develops more ontologically meaningful connotations—of both the existence and the maintenance of dark satellites. It is thus through the conflation of the real and virtual worlds, the blending of simulation data with empirical data, that the missing satellites problem is ‘solved’. (shrink)

There are many tangled normative and technical questions involved in evaluating the quality of software used in epidemiological simulations. In this paper we answer some of these questions and offer practical guidance to practitioners, funders, scientific journals, and consumers of epidemiological research. The heart of our paper is a case study of the Imperial College London covid-19 simulator, set in the context of recent work in epistemology of simulation and philosophy of epidemiology.