Data Cop Out

Weather, moon phases, locations of schools and churches -- these are all factors considered in predictive policing, a practice of analyzing crime data and uncovering patterns that may give patrol officers a leg up. Nowadays, predictive policing is powered by expensive and privately developed software, which is used by more than 150 police department across the country. While media outlets and police celebrate lower crime rates, critics argue that programs like HunchLab may not be able to claim responsibility -- especially considering very little research exists on their efficacy. Deb speaks with The Marshall Project’s Maurice Chammah, who isn't convinced HunchLab works. He visited Jennings, Missouri, a city neighboring Ferguson, to find out whether this supposedly unbiased software is really helping law enforcement do their jobs. 

Published in

On The Media

BROOKE GLADSTONE:  This is On the Media. I'm Brooke Gladstone.

DEBORAH AMOS:  And I'm Deb Amos. As we just heard from Matt Levitt, there is no sure way to predict a crime but that's precisely what 150 police departments across the country are trying to do with the help of a new algorithm, and it's getting rave reviews.

  [CLIPS]:

MALE CORRESPONDENT:  When it comes to protecting you and your family from violence or crime, law enforcement is gaining an edge.

FEMALE CORRESPONDENT:  It’s called predictive policing.

FEMALE CORRESPONDENT:  Predictive policing. Their officers can use their experience and their intuition with a scientific program to catch criminals in the act.

MALE CORRESPONDENT:  Norcross police busted two burglars inside this house because a computer predicted the crime. That's right. Is the future of crime fighting predicting the crimes of the future?

DEBORAH AMOS:  But when Maurice Chammah, a reporter at The Marshall Project, went to Jennings, Missouri to see one predictive policing tool in action - it's called HunchLab - he found no evidence that it actually works.

MAURICE CHAMMAH:  What this new technology does is take in more data than any individual police officer could ever on their own, not just the crime, the weather. The phases of the moon has even factored into what HunchLab does, the wind speed, the proximity to schools and to sporting events, and it can factor in the schedule for sporting events so that you know that you want to cluster more police around a stadium when there's a sporting event. And the idea is that the algorithm can kind of be smarter than a person can be.

DEBORAH AMOS:  Although it sounds pretty logical that if you have a sports event, you'd send more police around the stadium.

MAURICE CHAMMAH:  Right, and as these predictive policing technologies have come into the fore, a lot of police officers told me that they're essentially telling them what they already knew.

DEBORAH AMOS:  You went out on patrol in Jennings, and so how did you see it work?

MAURICE CHAMMAH:  For decades now, police officers on the ground have patrolled closer to areas that are called hotspots, which are areas where there's been a lot of crime before. And with HunchLab, if you imagine - you know, there's a computer screen in the front of most police cars right now - we’re sitting in the car, the computer screen is in front of the officer and he has the HunchLab software up on it. And it’s basically a map that has these little boxes, and those boxes are saying, during your shift it's more likely that this kind of crime is going to happen here. So the idea with the technology is that over the course of months and years you’re decreasing the number of crimes because police cars are around.

DEBORAH AMOS:  So you can't really know –

MAURICE CHAMMAH:  You can’t.

DEBORAH AMOS:  - if this works for years and years. So police departments are putting money in this technology but there's no data.

MAURICE CHAMMAH:  There is not great data, no. Studies are starting to be done. Some of those studies are funded by the companies that make the technology, which obviously raises suspicion.

DEBORAH AMOS:  And these are private companies.

MAURICE CHAMMAH:  And these are private companies. And police departments, when I confronted them with this, at least in St. Louis, the line that I got was, it’s not that expensive when you consider what policing costs, so they may be paying $80,000, which is the salary for some officers.

DEBORAH AMOS:  What’s composition of the Jennings Police Department?
MAURICE CHAMMAH:  Mostly white and, and Jennings itself is mostly black.

DEBORAH AMOS:  Embracing this technology, is there anywhere,  even among the police that you were with, that it could lead to more racial tension, rather than less in these communities?

MAURICE CHAMMAH:  Yeah. I mean, I definitely got the sense that racial tension in Jennings in these communities is high, and this adds to the idea that they're going to very specific spots and targeting people because they're in those spots. And there's every reason for people who are then being targeted to suspect, well hey, you know, why were you following me around? Just because I’m in this box doesn't mean I'm a criminal. Police will come back and say, well, these are the areas where there's the most crime. We just want to prevent the crime. And there are plenty of people who say, well, that is sort of blind to history and inequality. And then the police say, well, it’s really not our job to fox those things. Our job is, in the next 12 hours, to make it that there's less crime in this neighborhood than more.

But lots and lots of civil liberties advocates I spoke to in the St. Louis area were very worried that predictive policing technology would aggravate the extent to which a predominantly white police department would stop and frisk and search the cars of African-American residents. I think police officers would say, well, the data is better than my individual sort of decision making. And that conversation, I think, is another way in which predictive policing is now kind of the centerpiece for this very old debate that will continue.

DEBORAH AMOS:  And there's no way to tell that if they just opened the door and got out of the car and sort of did a walk and talk in the community, that that wouldn’t also reduce crime?

MAURICE CHAMMAH:  Right. No, we don't know. And it might reduce crime more. Crime rates fluctuate a lot, and so saying that a specific technology or a specific program reduced crime or exacerbated it is very, very hard to do.

DEBORAH AMOS:  As you say, this is the technology that's in the middle of the table. But HunchLab is only one of the entrants in this field. The big one is something called PredPol.

MAURICE CHAMMAH:  Very Silicon Valley, isn’t it?

DEBORAH AMOS:  It’s the more expensive model.

MAURICE CHAMMAH:  Yes.

DEBORAH AMOS:  And it comes with some responsibilities, if you take it on. Police departments are asked to essentially advertise for the company, to give press conferences, to tout the successes. That is a whole new relationship between technology and policing.

MAURICE CHAMMAH:  It is, and it is a disconcerting one. A lot of policing [LAUGHS] after Ferguson, and I hope this doesn’t sound too controversial, but a lot of it is about PR, a lot of it is about projecting that, you know, we’re fixing things, we’re doing what we do better. On a very real level, police departments are kind of grappling with that. And PredPol comes in and seems to have some answers for how to police in a way that appears to be better and more objective, and I think police departments are clinging to that.

DEBORAH AMOS:  Why do you think that police departments – and I’ll ask specifically about Jennings - is so willing to embrace a technology when there's no data to tell them that it works and there may not be reliable data for a long time?

MAURICE CHAMMAH:  I think there's a sense that this might work a little bit, and that's enough, that it doesn't have to completely reduce the amount of crime in a drastic way for the amount that we’re shelling out to be a little but worth it. We think that it will, just on a day-to-day level, kind of bump up our crime reductions. These police departments are kind of casting about for solutions, especially in this era that they sort of facing such scrutiny and pressure.

DEBORAH AMOS:  Thanks very much.

MAURICE CHAMMAH:  Thank you.

DEBORAH AMOS:  Maurice Chammah is a writer at The Marshall Project.