Epistemic Opacity, Confirmation Holism and Technical Debt: Computer Simulation in the Light of Empirical Software Engineering

The output of a simulation model does not, prima facie, appear to have an objective status comparable with data captured by observation or experiment. Counter to this Winsberg, Humphreys and others emphasise parallels between experiment and simulation in practices which are said to “carry with them their own credentials” (Winsberg, 2010). Humphreys holds that that the physical implementation of computer models places constraints on simulation methods not present in traditional mathematics, creating essential epistemic opacity. By this he means that it is impossible for a cognitive agent, given its characteristics, to know all of the epistemically relevant elements of a computational process. Humphreys views essential epistemic opacity as reflecting the limitations not of the simulation method itself but of the human agent, and thus as evidence for a “non-anthropocentric epistemology” recognising computational tools as a superior epistemic authority.