I’m biased, so I would give the most weight to what the computer science department says. Predictive policing absolutely does work in some aspects. The opposi

I’m biased, so I would give the most weight to what the computer science department says. Predictive policing absolutely does work in some aspects. The opposite of predictive policing is to have all beat cops equidistant from each other in their jurisdiction. This is clearly wrong. The next step is to have cops near population centers, then around areas where there have been more reports, requests, etc. After that you move to time to day. Where you place cops during the day vs night is clearly different. Shops close, bars open. Places where young adults vs seniors gather will clearly have different coverage. Pretty much every study in the subject shows people age out of crime. This is used as supporting evidence against lifetime sentences and geriatric incarceration. I have to assume only the most naive software will have an algorithm that uses secondary indicators / simplest predictive techniques would get such attention. For example “black neighborhoods” (that happen to be poor, with no other stats other than other departments say they have issues with black neighborhoods) and self fulfilling prophecies (we put most cops at location X and what a surprise, that location had the most arrests, w/o looking at the arrests themselves)