Predpol is a location-based, algorithmic Predictive Policing system that was developed by a team of UCLA academics in collaboration with LAPD.
After developing the algorithm, Predpol was also established by its creators as a for-profit business, funded early on by UCLA as the company’s “lead investor.” Predpol claimed to use reports of crime to make predictions about where crime was going to occur.
The Predpol logo, before the company rebranded as “Geolitica,” which has since been subsumed by SoundThinking.
Pages from a Predpol zine created by the Free Radicals and the Stop LAPD Spying Coalition:
The algorithm was developed by academics, led by UCLA anthropologist Jeffrey Brantingham. Together with Andrea Bertozzi (UCLA), George Mohler (Santa Clara University), Martin Short (Georgia Tech), and George Tita (UC Irvine), Brantingham worked with LAPD to create and launch the policing algorithm Predpol. LAPD provided PredPol “an automated data dump that happened every 2 hours” and was compiled and sent directly to the PredPol website for use. Using this LAPD data, Predpol claimed to be able to make predictions about when and where crime was going to occur, down to a 500 ft x 500 ft area, which was also called a “hot spot” (see Hot Spot Policing). Predpol reports with these “hot spot” predictions were distributed to officers daily to guide their Patrols.
Image description: Predpol creator Jeffrey Brantingham (of UCLA) pointing to a Predpol map on a computer screen. Image source: lapd.com
PredPol’s guesses about future crimes were driven by three elements of historical crime data: crime type, crime location, and a crime timestamp. This reported crime data originated from crime reports filed by police along with calls for service, both of which can be entirely unsubstantiated allegations.
A report issued by Predpol to LAPD’s Central Division for April 26th, 2016, showing Predpol “hot spots” indicated by red boxes on a map. Source: Document production from a Public Records Act request about Predpol.