Wlodek Rabinowicz (Lund) "Aggregating Value Judgments"
Abstract: My focus will be on the contrast between aggregation of individual preference rankings to a collective ranking and aggregation of individual value judgments to a value judgment of the collective. What I want to look at, in particular, is the case in which the two aggregation scenarios exhibit a far-reaching structural similarity: more precisely, the case in which, in the judgment aggregation scenario, the individual value judgments that are to be aggregated are value rankings. This means that, formally, the individual judgments in this case have the same structure as preference rankings over a given set of alternatives. But while in a preference ranking the alternatives are ordered in accordance with one’s preferences, the order in a value ranking expresses one’s comparative evaluation of the alternatives. I will suggest that, despite of their formal similarity as rankings, this difference in the nature of individual inputs in two aggregation scenarios has important implications for the task of aggregation: the kind of procedure that looks fine for judgment aggregation turns out to be inappropriate for aggregation of preferences. The procedure I have in mind is similarity maxization. There has been an extensive research on judgment aggregation in the last decades, and especially on reasonable rationality conditions for the aggregation procedure (Pettit & List, List & Dietrich, Nehring & Puppe, Dokov & Holzman, Mongin, Gärdenfors). Impossibility results have been proved, analogous to Arrovian impossibility results for preference aggregation. A different perspective, which I want to explore in my talk, would be to treat aggregation procedure as an optimization task: the task of finding an output that is maximally similar to individual inputs. Or – to put it in geometric terms – the aim is to find an output that minimizes the distance from individual inputs. (Cf. Pigozzi 2006, Miller & Osherson 2009, Rabinowicz 2011.) On this approach, the output need not be uniquely determined. When applied to judgment aggregation, maximization of similarity can also be approached from the epistemic standpoint: the questions will be posed concerning its advantages as a truth-tracker (on the admittedly controversial assumption that value judgments can be true or false). In this context, what matters is not only the probability of the outcome of the procedure being true, but also the expected verisimilitude of the outcome: its expected distance from truth.