Paper short abstract:

Predictive policing systems are increasingly employed to anticipate criminalized behavior through statistical analysis. Several case studies highlight the various political and epistemological entanglements encountered by these programs and their 'evidence-based' evaluation.

Paper long abstract:

Policing practices are changing. Security assemblages in different parts of the world are increasingly incorporating a diverse set of data-driven technologies. Next to a new generation of surveillance equipment and data infrastructures, emerging devices include algorithms that aim to anticipate a variety of criminalized behavior. These predictive policing systems aggregate information from crime statistics, ongoing investigations and complementary datasets (such as weather forecasts) in order to recognize patterns, extrapolate into the future and identify optimal locations for crime mitigation measures.

Beyond concerns about data security and privacy, or the effectiveness of specific programs, predictive policing applications raise a novel set of questions for social science research. What does it mean for assumptions about criminal behavior and mitigating policies to be hard-coded into operational infrastructures? In which ways are emerging platforms challenging separations between the state and the private sector, or domains of crime and national security? Finally, how can the technical specifics of these programs be reconnected to the political contexts in which they emerged?

The proposed paper will present preliminary findings of an ongoing research project that closely examines variegated predictive policing systems in the United States, the United Kingdom and Germany. Of special interest are efforts undertaken to statistically prove the effectiveness of these programs in order to meet the requirements of the 'evidence-based policing' paradigm. This contribution's STS-inspired approach is interested in a detailed technical understanding of the concepts, systems and interfaces deployed, as well as the operational concerns by professionals tasked with their implementation and evaluation.