Eight years in, LAPD can’t measure PredPol’s effect on crime

Eight years in, LAPD can’t measure PredPol’s effect on crime

“[T]he OIG cautions against drawing strong conclusions from the available statistics.”

Written by
Edited by Miranda Spivack

The Los Angeles Police Department, an early adopter of data-driven policing, needs tougher standards for data collection, recordkeeping, and communicating its policies to the public to guard against targeting minorities and certain neighborhoods, a new report from the department’s inspector general said.

The 48-page report from the office of Inspector General Mark Smith, issued on March 8, said data tracked by the department is inadequate for determining the impact of specific programs on crime. The report, reviewed on March 12 by the Board of Police Commissioners, examined databases, software, and crime statistics used in three specific programs. It found that data on “dosage,” the time an officer spends in an area deemed to be of interest, was too vague.

One critic of the LAPD’s predictive policing systems said that the report itself fell short by not accounting for the harm to targeted individuals and their communities from the department’s targeted policing practices and policies.

Over the past decade, LAPD has used several predictive policing systems, combining strategies that target individuals and locations. In 2009, the department began development of the L.A. Strategic Extraction and Restoration (LASER) Program, using nearly $1 million in grants from the federal Bureau of Justice Assistance, designed to identify for removal particular “high risk” individuals. Two years later, it began to employ the predictive policing software PredPol, which uses historical data to create daily reports on anticipated crime “hotspots.” And the agency recently began testing ELUCD, a platform to survey community sentiment. The IG’s report specifically reviewed LASER, PredPol, and ELUCD, noting use of a platform developed by data analytics firm Palantir to streamline information used in both the LASER and PredPol programs.

The 9000-member department has had a long history of controversy in which local groups have questioned its ability to overcome bias in policing. Stop LAPD Spying Coalition released a report in May 2018 critical of the department’s predictive policing and asked the inspector general’s office to conduct its own review. At a special July 2018 police commissioners meeting on data-driven policing, representatives from the American Civil Liberties Union of Southern California, Stop LAPD Spying Coalition, and the community testified in opposition to the department’s use of predictive technology - including the audited programs, automated license plate readers and video recording systems - to drive decision making and to identify individuals for warrantless surveillance. The five-member commission, following a request from Stop LAPD Spying Coalition, asked in August 2018 that the IG examine the agency’s existing data-informed strategies.

The IG report offered new information about the department’s use of technology, noting that in August the department had suspended use of the LASER system. Jamie Garcia, a Stop LAPD organizer, told MuckRock that the program’s suspension was news to her and had been done without public notice. However, the temporary suspension, she said, will not enable the department to compensate for its shortcomings.

“This isn’t a new program that just got started and it’s going to have some bumps,” Garcia told MuckRock. “This is a program that has been in effect for 10 years. This is 10 years of people’s lives that they have been screwing with.”

Know of an algorithmic development near you? Click here to let us know.

The LASER program relied on the identification of “Chronic Offenders” to be monitored for removal from the community. It also highlighted areas, referred to as “LASER Zones” or “hotspot corridors,” for increased officer attention. According to the report, the department is working to revise the program and plans to continue using it.

Garcia was also disappointed that the report included no mention of the community request for the investigation or the human and civil rights impact these programs have had, and instead ignored potential victims of the program and the activists who have worked to bring to light problems.

Stop LAPD released a response to the audit in which they call on the LAPD to “stop legitimizing harmful predictive policing programs by attempting to ‘reform’ them.”

LAPD has been employing PredPol since 2011. The software intends to help police departments target high-risk crime zones and is used by an estimated 50 police departments. However, activists and academics throughout the country, such as the ACLU, are worried that the reliance on old data generated by previous LAPD policing or arrest patterns will perpetuate existing biases in deployment of officers and investigative resources.

The department has begun to rethink some of its predictive policing methods, the IG report said. “[S]ome of the proposed changes for a revised offender-based program include more narrowly constraining the selection process, incorporating disclosure and appeal processes, and developing a centralized oversight component,” the report said. “The Department also expects to implement additional technology to assist in more accurately tracking data related to officers’ activities in the field, including those related to data-driven policing strategies.”

The LAPD did not provide comment by publication time.

Garcia, however, believes any reform that cannot properly address concerns that LASER, PredPol, or other predictive systems are inherently unjust.

“To even think about bringing something back that violates human rights, to me, is completely, completely unacceptable,” Garcia said. “How do you even talk about bringing back a program that has unjustly targeted and stalked people?”

The full report is embedded below.

Creative Commons License
Algorithmic Control by MuckRock Foundation is licensed under a Creative Commons Attribution 4.0 International License.
Based on a work at https://www.muckrock.com/project/algorithmic-control-automated-decisionmaking-in-americas-cities-84/.

Image by Cliff via Flickr and is licensed under CC BY 2.0