When the city of Camden, New Jersey, deployed its first predictive policing algorithm in 2014, officials promised a smarter, fairer approach to public safety. A decade later, the results tell a more complicated story — one that reveals how algorithmic decision-making is quietly redrawing the boundaries of urban life.
This is not just a story about policing. Across American cities, algorithms now influence where affordable housing gets built, which neighborhoods see investment, and how much you pay for a ride home. The cumulative effect is a kind of invisible zoning — one that operates without public hearings or community input.
The Prediction Machine
Predictive policing works on a deceptively simple premise: if you can identify where crime is likely to occur, you can prevent it. Software like PredPol (now Geolitica) analyzes historical crime data to generate "hot spot" maps that guide patrol deployment.
But historical crime data reflects historical policing patterns. Neighborhoods that were over-policed in the past generate more data, which the algorithm interprets as higher risk, which leads to more patrols — a feedback loop that academic researchers call "runaway feedback."
Dynamic Pricing and the New Redlining
The same pattern plays out in less obvious ways. Ride-hailing surge pricing, insurance risk models, and even grocery delivery fees all use location-based algorithms that can effectively create price discrimination by neighborhood.
A 2023 study from the Brookings Institution found that residents in predominantly Black neighborhoods paid an average of 12% more for ride-hailing services compared to white neighborhoods with similar demand patterns.
Automated Zoning
Perhaps most consequentially, cities are beginning to use algorithmic tools to make zoning decisions. These systems promise to optimize land use for efficiency, but efficiency is not a neutral concept — it depends entirely on what you choose to measure.
This is part one of a five-part series. The full investigation continues in parts two through five, examining case studies in Chicago, San Francisco, Atlanta, and Detroit.