What goes around comes around: why the critique of expert-led integral bayesian modelling also undermines judge-led bayesian modelling

Resumen

This article offers an internal refutation of Anne Ruth Mackor’s defence of court-led integral Bayesian modelling in criminal adjudication. Mackor usefully distinguishes between the now well-established Bayesian treatment of individual items of forensic evidence and the more controversial Bayesian modelling of criminal cases as a whole. She also convincingly argues that expert-led whole-case modelling is objectionable because the expert must select hypotheses, evidence, dependencies, and probabilistic inputs in a manner that effectively places the expert in the judge’s chair. Yet she maintains that this objection applies with less force when whole-case Bayesian modelling is performed by judges themselves, especially as an internal “means of inspection” or “sharpening” tool and with the support of a forensic adviser. This article argues that the distinction fails. The decisive objection to integral Bayesian modelling does not arise merely from the formal identity of the modeller. Rather, it lies in the fact that the modelling choices themselves are adjudicatively decisive. Those choices remain equally decisive when made by judges, and the involvement of a forensic adviser reintroduces, in a less visible form, the very institutional concern that Mackor’s critique seeks to avoid.