Fairness and randomness in decision-making: the case of decision thresholds.

Abstract

This paper defends the role of lotteries in fair decision-making. It does so by targeting the use of decision thresholds to convert algorithmic predictions and classifications into decisions. Using an account of fairness from John Broome, the paper argues that decision thresholds are sometimes unfair, and that lotteries would be a fairer allocation method. It closes by dealing with two objections. First, it deals with the objection that lotteries should only be used to break ties in cases where individuals' claims are equally strong. Here, the paper gives a new argument for Broome's view, targeting decision criteria that are arbitrary and highly standardized. It then defends the arguments of the paper against the objection that lotteries are not morally superior to other methods of arbitrary choosing.