Predictive policing: not yet, but soon preemptive?

Abstract

For several years now, crime prediction software operating on the basis of data analysis and algorithmic pattern detection has been employed by police departments throughout the world. As these technologies aim at forestalling criminal events, they may aptly be understood as elements of preventive strategies. Do they also initiate a logic of preemptive policing, as several authors have suggested? Using the example of crime prediction software that is used in German-speaking countries, the article shows how current forms of predictive policing echo classical modes of risk calculation: usually employed in connection with domestic burglary, they help police to identify potential high-risk areas by extrapolating past crime patterns into the future. However, preemptive elements also emerge, to the extent that the software fosters 'possibilistic' thinking in police operations. Moreover, current advances in crime prediction technologies give us a quite different picture of a probable future of preemptive policing. Following a general trend of data-driven government that draws on self-learning algorithms and heterogeneous data sources, crime prediction software will likely be integrated into assemblages of predictive technologies where criminal events are indeed foreclosed before they can unfold and emerge, implying preemptive police action.