Criminal investigations are structured as sequential processes of information production and transmission. Police authorities generate the first formal description of events, which is then transmitted to prosecutors and, ultimately, relied upon by judges when taking precautionary and adjudicative decisions. While this informational chain is institutionally necessary, it introduces a structural asymmetry: downstream actors rarely assess primary evidence independently, but instead form probabilistic beliefs on the basis of progressively filtered information. This article develops a descriptive probabilistic model of informational dependence in criminal investigations. Rather than treating judicial risk as deviation from an objective probability of guilt or as the result of correlated statistical errors, the model conceptualizes belief formation as a function of nested information sets and institutional belief mappings. Informational distortion arises endogenously from filtering, summarization, and reliance across institutional boundaries, even in the absence of bias or strategic manipulation. The analysis shows how early investigative framing can exert disproportionate influence on precautionary decisions taken under conditions of high uncertainty. As institutional reliance intensifies, judicial beliefs become increasingly sensitive to upstream information, reducing the corrective role of independent assessment. The article concludes by outlining a framework of epistemic decoupling aimed at restoring informational independence and mitigating judicial risk.
Abstract This commentary summarizes the novel aspects of the paper by Luby and Kadane, discusses specific points raised by the authors, and presents broader considerations, critical aspects, and remaining problems in the field.
