A Non-Exhaustive Taxonomy of Tools of Data-Driven Policing

Keeping track of the wide range of emerging technologies increasingly available to and used by police departments can be overwhelming. This taxonomy, while not necessarily comprehensive of every specific tool or product in the policing tech toolbox, provides a high-level overview of data collection resources and types of data used in policing.

Cell site simulators (often reported on as "Stingrays" the product name used by manufacturer Harris Corp.) imitate a cell tower and can be used to locate, identify, and on occasion intercept data from cell phones. The primary vendor for cell site simulators, Harris, requires police departments to sign extremely secretive non-disclosure agreements that can make it difficult to find out whether or not the technology is being used at all.
Police-worn body cameras vary in size, recording capacity, and deployment across police departments, but were generally heralded as a panacea for police brutality when the Department of Justice issued a $20 million grant to help police departments acquire them. Public records rules, access to camera footage, and policies around applying image-recognition technologies to footage vary from city to city and among body camera vendors.
Mugshots are not only a useful resource for police departments attempting to identify individuals, but also a valued source for public-domain imagery that can be used as training data for facial recognition software.
Automated License Plate Readers (ALPR) capture images of cars' license plates on roads, which can be used both for monitoring traffic violations and for cross-checking or tracking a vehicle's movements through a city.
In addition to cameras deployed by police departments, private businesses often will share footage with departments. Sometimes this happens upon request during investigations; for much of New York City's Domain Awareness System, feeds from private sector cameras of initiative corporate partners (primarily Wall Street banks) feed directly into the city surveillance system.
This is the more familiar and common tool of predictive policing, sold by companies like PredPol and HunchLab to police departments across the country.
In 2015, the NYPD began a program rolling out customized Windows phones to its officers as part of a new program to bring the department into the 21st century. Although the department claimed that it had improved NYPD response times, by 2017 some 36,000 outdated phones had to be replaced in part because Microsoft stopped maintaining its operating system.
In addition to standalone ALPRs usually positioned at intersections, some cities' police vehicles (including some of the NYPD's) are outfitted with mobile ALPRs. In 2016, the American Civil Liberties Union obtained a copy of a $442,500 contract between the NYPD and Vigilant Solutions, a company that provides license plate imagery data from a huge range of sources to municipalities.
Backscatter X-ray imaging is a technology sometimes used in airport body scanners and for scanning shipping containers at ports. The Z Backscatter Van (ZBV), developed by American Science and Engineering, is essentially that same technology in a mobile unit. While more typically deployed by military or federal customs and border patrol agencies, in 2012 investigative reporters at ProPublica sued the NYPD for refusing to disclose information about their use of the vans (a suit that ProPublica ultimately lost in a split decision on appeal). To date, there is little public information about the NYPD's deployment of these vans or the potential radiation risks they pose to the general public.
Computer vision (CV) can refer to a number of different techniques or software tools but generally refers to methods of automating image recognition by training a computer to identify specific items in digital images. Facial recognition, optical character recognition (turning images of text into text), and other biometric tools rely on CV. Increasingly, companies selling body cameras and other surveillance equipment to police departments are working to build image recognition directly into imaging products.
Police departments have been deploying surveillance cameras in cities since the 1960s, with varying levels of public knowledge and varying degrees of public access to camera footage via records requests.
Tattoos are useful biometric identifiers and sometimes correlate with individuals' gang affiliations. Tattoo software recognition tools make work previously done by police departments with binders full of photographs into an automated task of cross-referencing and searching.
Social media data analysis involves a fair amount of both image and text-based data, and is a means of tracking relationships and networks. In 2013, the NYPD engaged in multiple mass arrests of suspected gang members in Harlem as part of an initiative called Operation Crew Cut. The gang arrests relied heavily on Facebook data, entangling a number of innocent Harlem teens in criminal conspiracy charges. In California in 2015, the San Diego district attorney attempted to use Facebook posts and, effectively, living in an area with gangs to implicate Aaron Harvey in a criminal conspiracy charge. The case was ultimately thrown out, but law enforcement standards for social media evidence and investigation remain a nonstandardized and poorly documented practice.
These spaces (sometimes called "Domain Awareness" or "Real Time" systems or centers) are in theory useful for maintaining a high-level perspective on all of the surveillance systems operating in a city at a given time. Generally, they're experienced by the public as photo ops or something imitated by televised procedural crime dramas.
A more controversial tool of predictive policing, "heat lists" are attempts to identify individuals likely to be at risk of committing a crime or being involved in a violent incident based on previous crime records and data. The approach was tested by the Chicago Police Department in 2013 and 2014 through a $2 million grant from the National Institute of Justice; media controversy and a damning report from the RAND Corporation undermining the technique's efficacy in fighting crime has led to few private vendors actively pursuing a heat list model.
Common deployment of dashboard cameras in police vehicles came in two waves in the United States: first, in the 1980s, when Mothers Against Drunk Driving fundraised to have them installed in cars to strengthen DUI evidence; then in the 1990s during the War on Drugs as an aid to evidence that individuals had in fact consented to having their vehicle searched.
As GPS receiver devices have become smaller (due in part to the advent of cell phones), location-based data collection has become easier to deploy. GPS is so frequently added into metadata determining camera location and tracking phones and vehicles that it's easy to neglect in a larger inventory of data-driven technologies deployed by police departments.
Shotspotter is an acoustic surveillance product and platform created by California-based SST, Inc. The listening devices are supposed to pick up the sound of gunshots in an area, triangulating its location in order to improve police response times and address the recurring problem of unreported gun violence. The technology came under scrutiny when it was revealed that the microphones in Shotspotter devices can, in fact, pick up and record audio other than gunshots (said audio was used as evidence in shooting investigations in 2007 and 2012).

Ingrid Burrington is a writer and artist based on a small island off the coast of America. She’s the author of Networks of New York: An Illustrated Field Guide to Urban Internet Infrastructure and has previously written for The Atlantic, e-flux journal, and The Verge, among other outlets.

Series

The Location of Justice

An examination of the pervasive and often overlooked infrastructure of criminal justice in New York and the spaces that could serve a more just city.

In This Series