Bulgarian study urges courts to treat evidence as behavioral algorithms

A new academic paper from Bulgarian researchers Dobromira Bankova and Vladimir Dimitrov examines the growing role of data from information systems in court proceedings. Drawing on current judicial practice in Bulgaria and recent substantive legal changes, the study highlights legal and practical challenges when courts collect, analyze, and evaluate datasets as evidence. The authors walk through specific cases to show how judges struggle with raw data that was not originally designed for evidentiary purposes, especially under the country's new regulatory framework.

The core argument is that legal analysis must undergo a conceptual shift: instead of treating "evidence as an information unit" — a single data point or record — courts should approach it as "evidence as a behavioral algorithm." That means looking at aggregated digital footprints as patterns of behavior that can be interpreted algorithmically, not just isolated facts. The researchers stress that this shift demands more than new software; it requires a methodological transformation in how evidence is gathered, validated, and presented. The paper, published on arXiv (2608.16901) under the Computers and Society category, contributes to a growing debate on data-driven justice, offering a framework for judges and lawyers working with complex digital datasets in legal environments increasingly shaped by automation and information systems.