A Comparative Simulation Study of the Fairness and Accuracy of Predictive Policing Systems in Baltimore City

Samin Semsar, Kiran Laxmikant Prabhu, Gabriella Waters, James Foulds · 2026

Read A Comparative Simulation Study of the Fairness and Accuracy of Predictive Policing Systems in Baltimore City by Samin Semsar, Kiran Laxmikant Prabhu, Gabriella Waters, James Foulds (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-02566.txt.

Description

Samin Semsar and 3 co-authors run a head-to-head simulation of predictive policing and traditional hot spots policing in Baltimore, in a 2026 arXiv paper. Predictive systems forecast where crime will occur, and critics charge them with racial bias fed by feedback loops and past biased records. Head-to-head studies, however, remain rare and narrow.

The results complicate the standard story. Predictive policing did show bias from feedback loops, as earlier work reported. Hot spots policing, the older approach, had similar issues. In the short term predictive policing was both fairer and more accurate, yet it grew bias faster, hinting at worse long-run behavior. In some Baltimore cases bias tilted toward over-policing white areas, contrary to earlier studies. The authors offer a city-by-city testing method rather than one verdict for all.

Catalog record

CREATORSamin Semsar, Kiran Laxmikant Prabhu, Gabriella Waters, James Foulds
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE70 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602-02566

Browse the full library · About the Tower of Babel