An Evaluation of Predictive Policing: Vendors, Municipal Governance, and the "Smart City"

Abstract

Predictive policing, an umbrella term for tools that use data to predict where and when crime will take place, or who is involved. predictive policing reflects a trend, or a paradigm shift towards using data for predictive rather than reactive or descriptive purposes. Advocates of predictive policing argue that the software can make policing more effective, efficient, and objective. At the same time, critics of predictive policing contend that predictive policing exacerbates policing bias and targets individuals. In addition, critics have raised concerns about privacy, legal and constitutional implications, and the lack of public input and transparency that surrounds the technology. Based on this policy debate, my thesis asks two questions. First, how do vendors use crime theories and smart cities goals to build, market, and defend predictive policing technologies? Secondly, how do cities procure predictive policing systems? Why, and in what cases, does predictive policing remain obscure from the public? To answer the first question, I profile and evaluate predictive policing vendors. This thesis establishes that there are key differences between these two categories of vendors. The first group of vendors, including PredPol, HunchLab, and CivicScape, have employed accountability, transparency, and safeguards against bias as their marketing strategy. The second group of vendors, including IBM, Hitachi, Motorola and Microsoft developed predictive analytics products for other markets and then applied them to policing. These large technology vendors demonstrate a superficial or erroneous understanding of the crime and policing issues they attempt to solve. To answer the second question, I overview different ways that cities have procured predictive policing technology. Then, I analyze the process by which a city decides on predictive policing technology in the municipalities Bellingham, Washington and Fresno, California. My analysis reveals a need for close consideration of vendors and the differences in their systems. In addition, municipalities considering predictive policing should follow a procurement process that clearly articulates community goals and involves transparency and community control. Finally, municipal governments should adopt ordinances that require public input on the procurement of predictive policing and attempt to regulate its use.