Technology watchdogs and technoculture critics have discussed predictive policing software at an abstract level or have tried to reverse engineer its blackboxed code. In this paper, we use the methods of Critical Code Studies, media archaeology, and software studies more broadly to analyze CivicScape predictive policing software, published online, albeit partially. Working from an anti-racist approach, we examine how the CivicScape code calculates which neighborhoods to recommend for heavy policing. Our reading demonstrates what code analysis can add to the analysis of such software and makes a case for the public release of all legislative operational source code for scrutiny under the principles of the Freedom of Information Act.
Ciston et al. (Fri,) studied this question.