Overcoming sentencing inconsistency - a proposal for algorithmic guidelines and juridical misalignment index

This study examines algorithmic support for punishment adjustment in judicial discretionary systems, based on the example of drunk-driving cases in Poland. Using an extensive case study, the research explores how algorithm-based sentencing models can address problems in judicial decision-making processes and mitigate inconsistencies arising from misapplied human discretion. The study’s objectives are the following: (1) identify factors influencing punishment severity in drunk-driving cases in Poland; (2) compare these findings to statutory rules governing punishment in Polish criminal proceedings to uncover inconsistencies; (3) evaluate the practical implications of discretion in this category of cases and propose remedial measures; (4) integrate algorithmic sentencing guidelines into the proposed solution; and (5) introduce a novel metric to quantify undesirable judicial decisions, enhancing automated judge-support systems in criminal proceedings.