Algorithmic Policing: An Empirical Study of the Algorithmically Mediated Construction of Individual Risk in a UK Police Force

• In Event: Technologies of Policing; Policing as Technology – II

Abstract

Predictive policing from risky places to threat scores has captured the imagination of both enthusiasts within police forces hoping to increase public safety and opponents raising, among others, concerns around algorithmic bias and total suspicion. Based on in-depth interviews with police officers in different functions in a UK police force, this paper addresses the lack of empirical research in these discussions. Oftentimes the discourse is informed by techno-determinist visions in which police officer robot-like enact an algorithm’s orders. Contrastingly, this paper examines the process in which police officers and software co-construct the risk of individuals. Drawing on STS, the paper reveals the complex power dynamics between risk scores as a reified reminder of an individual-focussed strategy and the operational conditions and alternative priorities pursued by officers. Particularly influential among these concerns are conditions of austerity shaping both the perceived need for the algorithm and its dismissal.

Author

• Daniel Marciniak, Max Planck Institute for Social Anthropology