Predictive policing is an umbrella term for various technologies meant to "combine crime mapping, statistical analysis and law enforcement expertise to provide forecasts of where and when crimes will most likely occur," according to a Department of Homeland Security note. While the software has received critical attention in larger cities, its rollout has been slower in New Mexico.
When SFR asked Santa Fe Police spokesperson Greg Gurulé whether the department is using predictive policing in its approach to crime, he wrote that the department uses "Geographic Profiling Analysis." This actually isn't a form of predictive policing, but rather a way to use geographic indicators to trace a criminal suspect's residency or "place of origin" in order to locate and arrest them.
The police department uses the Rigel Analyst program developed by the company Environmental Criminology Research Inc. SFPD has had some form of it since 2005, but Gurulé says its use has been interrupted by "changes in personnel."
Even if Santa Fe police aren't quite yet using predictive policing programs, a report from The Santa Fe New Mexican this week indicates the department is using past data to decide where to allocate patrols for DWI checkpoints. The paper found that 22 of 27 DWI checkpoints in the city were set up in the less-affluent south and southwest regions since 2016.
These patrols created a similar situation as what's happened in cities where the software is in use: People living in less affluent and less white areas are targeted for greater policing because that's where algorithms tell police to go. As a result, existing biases in policing are reinforced.
"We do [conduct DWI checkpoints on the south and southwest areas of Santa Fe] quite a bit, but that's where the data tells us to go," police Capt. Marvin Paulk told reporter Daniel Chacón.
This defense—that data-based patrolling isn't unfair because data is supposed to be objective— echoes the one used by police departments with predictive policing tech.
In Los Angeles, where grassroots groups are protesting the LAPD's use of predictive algorithms, officers are daily given refreshed maps of where property crime is forecast to occur based on past crime data. The LAPD's algorithm, designed by a company called PredPol, is sophisticated enough to track how much time officers spent in an area, and correlate it to the subsequent occurrence of crime.
"It's focusing on a geographic area. PredPol has no information on suspects," LAPD Police Captain Jeff Nolte told CBC, adding that it was "no different than the police radios telling us where to go."
Is use of the technology likely to become more common in New Mexico?
At a discussion convened by the state's Criminal Justice Reform Subcommittee on Thursday, the chief of policy and planning for the Second Judicial District, Adolfo Mendez, told legislators that his office's use of data analytic technologies developed by New Mexico Tech had resulted in a sustained drop in crime in the district since last August.
One of these tools is "micro targeting geographically," which Mendez described as the attorney's office receiving granular-level crime data from specific regions and using it to decide how, or whether, to prosecute particular crimes.
"We can see the specific language of the police narrative," Mendez said, while showing the room a visualization of property crime in Albuquerque's Nob Hill. "[What] did the criminal complaint say about that incident? We can drill down all the way to that level of detail with this particular tool."
The Second Judicial District attorney's office initiated the use of associated criminal network analysis tools in July, Mendez said. Albuquerque police have had predictive policing technology since at least 2012, according to KOAT.
Santa Fe Police also share geographic profiling data with the First Judicial District Attorney's office, but according to Mendez, his district has the most advanced tools. State Sen. Sander Rue (R-Albuquerque) expressed hope that the analysis capabilities could spread throughout the state.
"Maybe the Legislature will appreciate what [the second district] is doing and provide funding to expand it statewide," Rue said at the subcommittee meeting.
Whether data on past crime is used to inform policing or prosecutorial decisions, all of it comes from arrest reports made by police. These reports are based on the patrolling biases of departments, which isn't typically subject to independent oversight that could address how police enter data into their systems or where they patrol.
Alex Vitale, a professor of sociology at Brooklyn College, says a lack of transparency is common in cities where police are increasingly relying on predictive technologies. In a city like Santa Fe, whose police haven't yet acquired such software, there's still time to implement public oversight before it's ever deployed.
"Why should we agree to letting police departments determine the rules on what technology they're going to use and how they're going to use it? That's a public issue," Vitale tells SFR.