Paper short abstract:

The paper looks at how predictive policing software shapes the practices of crime prevention. By using ANT concepts of “script” and “de-scription” to grasp the relations between algorithms and police officers, the paper aims to stress ANT contribution to the “power in algorithms” discussions.

Paper long abstract:

As shown by Amoore, post-9/11 discourses on domestic security have been shaped by the idea that terrorist attacks can be prevented by "connecting the dots", i.e. by using algorithms to detect potential threats through disparate sets of data. This paper looks at the government of risks in the urban realm by examining how the contemporary practices of urban crime prevention are reframed by this algorithmic logic of "connecting the dots".

In particular, I intend to focus on current "predictive policing" software, which connect and analyze data from disparate sources (past criminal activities, 911 calls, smart CCTV images etc.) to target areas or individuals that could present or be exposed to a risk. As evidenced by a growing number of companies developing their software solutions, as well as by an increasing number of US and European cities using such software, "predictive policing" seems to be on the rise.

This paper looks at these data-driven security networks, by examining the enrolment of a particular actor in the control room: the officer, who interprets, negotiates and enacts algorithms claims about potential crimes. By using ANT concepts of "script" and of "de-scription", I wish to comprehend the relations between the algorithms claims (to what extent are they prescriptive?) and the officer actions (can she negotiate those claims?). Drawing on Murdoch´s (1998, p. 364) idea that "prescription and negotiation are two sides of the same coin" I wish to underline how ANT can contribute to the scientific discussions about "power in algorithms".