Abstract

Framed as a critique of David Hume’s analysis of miracles, Peirce offers a sustained argument against an approach to historical inference he calls the “Method of Balancing Likelihoods‘ (MBL). In MBL the posterior probability that a disputed historical event has occurred is computed on the basis of the prior probability of that event occurring and the probability that each purported witness of the event has given accurate testimony. Peirce’s critique of this method is hierarchical: he denies that an objective probability obtains for the truthfulness of witness testimony. Conceding this point, he asserts that, even if such objective probabilities exist, it is implausible to believe that witnesses are independent of one another. Conceding the first two points, Peirce argues that the very sampling process inherent to history necessarily introduces a strong probabilistic dependence that makes MBL unreliable. Finally, irrespective of the success of his first three criticisms, Peirce argues that MBL can be shown by empirical means to fail as a reliable method of inference. I reconstruct this hierarchical critique from a handful of Peirce’s manuscripts, and emphasize its continuing relevance for modern accounts of judgment aggregation