Properties are Potatoes?: An essay on ontological parsimony

The cost/benefit analysis is standard fare in contemporary ontology: we measure how a theory performs along a variety of dimensions (simplicity, coherence with intuitions etc.) and then – having completed this for all theories currently being evaluated – we opt for the theory that has, overall, the best cost-benefit profile. §1 is a more detailed exposition of this methodology. One of those theoretical dimensions is ontological parsimony, and one common way of achieving it is 'reduction by identification' whereby an initial theory that commits to the Xs and the Ys has a more parsimonious rival according to which each X is one of the Ys (e.g. properties are identical to classes, possible worlds identical to disconnected spacetimes, material objects identical to spacetime regions etc.). This chapter argues that a theory achieving such ontological parsimony by this method is never a point in its favour – with the exception that it might, in general, push us towards a one-category ontology. This may be because you give up on ontological parsimony as a virtue or the tenability of making reductions by identification. Alternatively it may be because you agree that it's a virtue and the reductions are tenable, but that the only thing you can draw from this is that we should endorse a one-category ontology. This chapter argues for the latter, although the other options – ditching either reduction by identification or parsimony as a virtue – are both lessons you may draw instead. As the former conclusion permits ontological parsimony achieved via reduction by identification to play some role in ontology (since it forces us towards a one-category ontology) and these options rule out even that role, they are stronger than the paper's intended conclusion – but in that all options rule out ontological parsimony playing anything other than a very diminished role in theory choice, I'd be happy for you to accept any of them. I begin by laying out an absurd theory involving arbitrary identifications, 'Spudism': that every property is a (different) randomly selected potato (§2). § §3-6 defend Spudism, showing how the tools already deployed to make other, less absurd, identifications plausible can likewise be used to make Spudism plausible. The lesson generalises: if we accept any reductions by identification, nothing should stop us making absurd and arbitrary identifications between whatever things we like. I conclude that if parsimony is so easily achieved it effectively becomes a …