Space, time and parsimony

Abstract

This paper argues that all of the standard theories about the divisions of space and time can benefit from, and may need to rely on, parsimony considerations. More specifically, whether spacetime is discrete, gunky or pointy, there are wildly unparsimonious rivals to standard accounts that need to be resisted by proponents of those accounts, and only parsimony considerations offer a natural way of doing that resisting. Furthermore, quantitative parsimony considerations appear to be needed in many of these cases. When scientists evaluate scientific theories, they often have to rely on many criteria besides whether the relevant theories are consistent with the evidence so far. For one thing, it is harder than it looks to have evidence flatly contradict hypotheses, especially when we remember how easy it is to chalk recalcitrant results up to noise, or experimental error, or bias, or as-yet-undiscovered factors, and the fact that our theories often only have probabilistic connections to data. For another, it is relatively easy to come up with different theories that agree on any given body of data, though many will look like mere philosophers' tricks rather than serious scientific rivals. Finally, we need theories to help tell us about tricky cases that we have not observed so far: and when we are setting up experiments or observations we are not already sure about, we want to put extra effort into testing plausible theories of the new phenomena, rather than doing undirected data collection. But this suggests that we want criteria of plausibility that help choose between theories that have not already been ruled out by the evidence we already have. Scientists do seem to use criteria such as simplicity, coherence with theories in other areas, inductive considerations, valuing explanatory hypotheses, and so on. Call criteria of theory choice like these "theoretical virtues". 1