Months after end of ‘predictive policing’ contract, Cantrell administration works on new tool to ID ‘high-risk’ residents

29th October 2018   ·   0 Comments

By Michael Isaac Stein
The Lens

Months after the city abandoned a controversial data initiative used to predict and identify which New Orleans residents were most likely to be entangled in violent crime, Mayor LaToya Cantrell’s administration is quietly developing a similar tool, documents obtained by The Lens show.

The new identification tool is one component of a larger plan to target high-risk citizens with “social interventions.” Additional details about “the method for actually implementing and tracking our target population,” will be made public by May 2019, according to an internal memo.

In February, an article in The Verge shed light on a largely unnoticed arrangement between the city of New Orleans and Palantir, a company created by Silicon Valley billionaire Peter Thiel with funding from the CIA. It gave the city access to Palantir’s “Gotham” software, which, the company claims, can identify people at high risk of committing or falling victim to gun violence by analyzing police reports, social media, and other databases, including jailhouse phone logs.

The ACLU bemoaned the program as “predictive policing,” and argued that algorithms relying on police data exacerbate existing biases in the criminal justice system. Other critics said it unjustly punished people for hypothetical crimes they hadn’t yet committed.

The city had been using the software since 2012. But in March, former mayor Mitch Landrieu declined to extend the Palantir agreement for a fourth time. And current Mayor LaToya Cantrell said she would not revive it.

Cantrell has kept that promise, but documents obtained by The Lens show that her administration is in the beginning stages of developing a new tool for identifying likely victims and perpetrators of gun violence, partnering with the same criminologist whose research formed the foundation of the Palantir software.

The tool is being developed by the New Orleans Gun Violence Reduction Council, which Cantrell created in May.

In an August op-ed published by NOLA.com, Cantrell said that the council’s central task is “to come up with a plan to execute the violence reduction recommendations produced by my transition team.”

The council is made up of Cantrell’s director of strategic initiatives, Joshua Cox, as well as unpaid volunteers including Flozell Daniels, the executive director of Foundation for Louisiana, and Melissa Sawyer, the executive director of the Youth Empowerment Project.

Taking a lead role on the council is Jeff Asher, one of the two primary analysts who used the Palantir software to create a list of 3,900 people — about one percent of New Orleans’ population — who were at high risk of being involved in gun violence, according to Palantir documents.

Cantrell’s office did not respond to repeated requests for comment on this story. Asher said he was not authorized to comment on the record.

The program is still in its formative stages, but the broad idea, according to documents, is to use the new identification program to target individuals for “impactful social interventions” rather than strict criminal enforcement.

It is a modified version of Landrieu’s Group Violence Reduction Strategy notification sessions, more commonly known as “call-ins.” The call-ins brought together hundreds of suspected gang members for “interventions.” These residents were selected, in part, using the Palantir software, city officials told NOLA.com.

They were given a choice: engage with social programs or prepare for swift police action and harsh prosecution if you commit another crime.

“We know who you are. We know who your friends are. We know where you go. We know everything about you,” Landrieu said in 2012 after the first call-in.

The results were mixed, according to a 2015 study published in the American Society of Criminology. Reviews of the program showed a decrease in violent crime in the wake of the call-ins.

But the drop was short-lived.

“Firearms violence experienced a substantive and statistically significant reduction early in the post-intervention period, with diminished returns over time,” said a 2017 review of NOLA for Life cited in Cantrell’s mayoral transition report.

Critics said the interventions were too focused on law enforcement, and that the program made participants targets for heavy-handed policing. A 2016 NOLA for Life progress report found that about 75 percent of call-in attendees signed up for social services. But the 2015 American Society of Criminology report found that only 25 out of 158 call-in attendees reviewed actually received services.

The participants, as well as the people the Palantir software determined they were associated with, still remained exposed to “enhanced prosecution,” however.

Meanwhile, the Palantir software became an “integral part of the investigative process” for the New Orleans Police Department, according to a powerpoint co-authored by Asher. The city’s multi-agency gang unit used the software to aid in the arrest of 83 suspected gang members between 2012 and 2014.

Social Services, Not Arrests

Part of the ethos of Cantrell’s new council is that regardless of past mistakes, focusing violence reduction efforts on a small, at-risk portion of the population remains a promising strategy.

“Research has shown that gun violence is incredibly concentrated in American cities with only a small fraction of a city’s population involved in that violence,” says a memo sent from Cox to the rest of the council. “Previous efforts in New Orleans and elsewhere have shown that focusing on that small concentration can be effective though these efforts have largely relied on law enforcement levers to effect change.”

This time, city officials say, it will be different. The new program will “engage with at risk population with social services rather than law enforcement mechanisms,” according to notes from the council’s Aug. 22 meeting.

But it’s still unknown what those social programs will be, how they will be funded, or why they will be more effective than former initiatives. The transition report suggests that the strategy would employ a mix of new programs and additional funding for existing programs.

“There is not sufficient data to support the effectiveness or return on investment (ROI) in programs that are currently funded by the city and aimed at violence reduction,” the transition report says.

The role of the police is also still largely undefined.

While the violence reduction strategies portion of Cantrell’s transition report stresses the use of social services over law enforcement, one of its recommendations is to improve the NOPD’s homicide clearance rate by examining “the use of technology to solve crimes.” The transition report also recommends “enhanced data collection” and increasing law enforcement’s “predictive analytic capacity.”

According to meeting notes, there seems to be some confusion within the council about the function of the NOPD as well.

During their August meeting, Daniels mentioned that the council had to “reconcile idea of how we can rely so heavily on social services without including law enforcement,” the meeting notes say. He talked about the low capacity of the NOPD homicide unit and the “incompetency of some responders.”

“Approach should be comprehensive,” the notes say.

Asher responded by saying this was “not necessarily under purview of [the Gun Violence Reduction Council]. Though, we must address this in some way.”

Another council member, Nathalie Simon, suggested increasing law enforcement resources and capacity for deterrence. “If we solve more crimes, less likely to feel like you can get away with it,” the meeting notes say. Simon is a special council at Laitram, a Harahan-based manufacturing firm.

Even in the absence of an official role in the program, it’s unclear if there’s anything that would prevent the NOPD from accessing the data and identities produced by the new program for its investigations.

Papachristos’ involvement

The blueprint for the new identification tool and its method for calculating people’s individual risk is also ambiguous, mostly because it hasn’t been created yet. What we do know is the person who will build it: Dr. Andrew Papachristos, a professor at Northwestern University.

A council memo calls him the “country’s leading nation expert on the concentration of gun violence in American cities.”

Asher contacted Papachristos in July to ask if he wanted to help “create a new violence reduction program in the mold of many of the ideas we have talked about for years — namely identifying individuals at high risk and delivering impactful social interventions.”

“Your timing is insanely perfect,” Papachristos responded. “In a nutshell, I’ve been in conversations with a few funders (whom will remain nameless for now) about trying this out.”

Papachristos’ research has played a large role in an increasingly prevalent premise in criminology: a large portion of urban gun violence is concentrated among a small portion of a city’s residents, and the population at greatest risk of being shot is very similar to the population that is most likely to shoot someone.

In a 2013 Washington Post op-ed, Papachristos said that gun violence “in the way it is transmitted, bears striking similarities to public health epidemics such as cholera in Haiti or HIV/AIDS in the United States.”

“You’re more likely to ‘catch’ the disease if you engage in risky behaviors with someone who might be infected,” he wrote. “The solution is not broad, sweeping policies, such as New York’s ‘stop and frisk’ or mass arrests, but the opposite: highly targeted efforts to reach specific people in specific places, akin to providing clean needles to drug users to prevent the spread of HIV.”

But government-run gun violence reduction programs have displayed a tendency to target at-risk individuals with strict enforcement rather than aid, according to the council.

“No American city has yet created a gun violence reduction program delivered primarily through social program interventions rather than by law enforcement professionals,” says a council memo.

The Chicago Police Department’s controversial “heat list,” which was inspired by Papachristos’ research, is one example. The list uses an algorithm to numerically rank Chicago residents based on their likelihood of being involved in a shooting. It has guided the police to make large-scale arrests, which Jeff Asher noted in a 2017 New York Times article.

Papachristos didn’t have a direct hand in creating the Chicago heat list. Nor did he design the Group Violence Reduction Strategy in New Orleans. But according to a Palantir white paper, his research served as the program’s foundation. Papachristos said he was not available to comment in time for the publication of this story.

Papachristos has disavowed the use of his work for predictive policing and heavy-handed enforcement. In a 2017 Yale Daily News article, he called for cities to “treat these individuals as victims and afford them the same care we would of people who are involved in any other epidemic.”

In emails, Papachristos proposed that New Orleans model should incorporate more community input.

“What do you think about holding some sort of series of meetings with residents to talk about ‘predictive prevention’ (we should coin that phrase, btw)?” he wrote in an August email to Asher. “Rather than repeat Palantir, we should start to tell people what we’re going to do and have an avenue for input?”

Asher liked the idea. But not everyone on the council — which is mostly made up of unpaid, part-time volunteers with full-time jobs — was enthusiastic.

“A ‘series of meetings with residents,’ however, sounds like a potentially VERY time intensive and painful process,” Sawyer wrote. Sawyer declined to comment for this story.

“I def agree with [Sawyer] on the meetings and ‘avenues for input’ piece,” Cox responded. “Don’t know that we have the capacity to execute on that really well at scale.”

Phases and funding

According to an internal memo, the plan will be developed in five stages. The public should see a formal, written description of it by May 2019.

The first phase is identifying the at-risk population.

“After reviewing historical gun violence data in New Orleans, Mr. Papachristos will develop a scientifically backed program to identify individuals to whom we should deliver interventions,” the memo says.

This population of high-risk individuals is estimated at 2000 people, according to meeting notes.

Then, the council will review and evaluate local and national programing to decide what is most likely to succeed in New Orleans.

Third, the council will build the actual organization that will oversee and manage the programming. The yet-to-be-named leadership of the new organization will be housed within City Hall.

Only then — three phases after the target population is identified — will the council focus on sustainable funding for the program, according to the memo.

“This initiative will be sustainably funded for decades to come,” it says.

The only apparent expenditure that the council has made so far is for the services of the Caulfield Consulting Group. In their proposal, the consulting firm says it will charge $24,600 to help facilitate council meetings, conduct research, and produce a 30-page written “Gun Violence Reduction Plan document.”

Sustainable funding was identified by the council and Cantrell’s transition report as a major flaw of Landrieu’s NOLA For Life Program.

“Total funding for the program has never exceeded $7.5M per year and total general fund expenditure has never exceeded $2.5M per year,” the transition report says. “This is an insufficient level of funding to invest in violence reduction in New Orleans given the scope of the problem.”

Cantrell’s mayoral transition report recommends “establishing a public-private partnership as a primary funding source” for any replacement for NOLA for Life.

“We are going to have to raise a lot of money,” says one starred meeting note.

“This project cannot be solely dependent upon philanthropy because that funding is fickle and does not scale,” a council memo says. “It cannot be solely dependent upon government because politics and changing administrations alter priorities and funding decisions.”

The council is considering a novel solution. Upfront costs would be covered by charities, and annual costs would be paid by the city. In the long-term, the council hopes the city can recoup its costs through social impact bonds.

Social impact bonds have been used to fund social services that deliver quantifiable financial benefits. Investors gain, or lose, based on the program’s ability to deliver those benefits. In this case, the gun violence reduction program hopes to decrease the hospitalization costs for gun victims in New Orleans, which the council estimates to be $35 million a year.

“Many victims of gun violence are unable to pay the costs of these medical expenses, and if the hospital cannot bill Medicaid, those expenses become a financial loss,” the memo says. “We could set up an agreement for [the University Medical Center] to pay the City a percentage of those savings, which would be re-invested into the program.”

The above article originally appeared in The Lens on its website (www.thelensnola.org). The Louisiana Weekly enjoys a partnership with The Lens.

This article originally published in the October 22, 2018 print edition of The Louisiana Weekly newspaper.