What Makes a Prediction Arbitrary? A Proposal

What makes a prediction arbitrary? This article explores the possibility that one source of arbitrariness is asserting “P iff Q” when the justifications of P and Q are, in the relevant sense, independent. It uses this idea to draw a formal distinction between non-arbitrary and arbitrary predictions, even if they are also correct. It initially illustrates with Goodman’s New Riddle. There are, by now, so many different literature-prominent variants of grue, emerose or other odd-looking predicates that, surprisingly, the criterion handles them relatively straightforwardly. As the proposal is markedly distinct from existing resolution strategies centred on predicate projectability or counterfactual analysis, it sidesteps their well-known problems. The article makes the case that testing entailments for arbitrariness enjoys advantages over the currently dominant approach of directly putting conditions on when an F that is G should support “All F are G.” The article then surveys possible broader objections and offers two formal tools for identifying when and especially why a hypothesis makes good – that is, correct and non-arbitrary – predictions or not.