The tech tool for police accountability that's like "Yelp for cops"

Jul 24, 2018

In the age of the Black Lives Matter movement, the question of how to hold police accountable gets brought up a lot. One Brooklyn-based startup, Elucd, has partnered with the New York Police Department to develop a tool they say aims to answer this question. Marketplace host Amy Scott talked to Simone Weichselbaum, who covered it for The Marshall Project in a piece called "Yelp for Cops." The following is an edited transcript of their conversation.

Amy Scott: So how does this tool actually work?

Simone Weichselbaum: It's a very complicated data algorithm tool, which is two-pronged. So the first part is akin to a Yelp for cops ... so what does that mean? They use this technology to ping cellphones the same way pop-up ads appear in gaming apps like Candy Crush. And up pops a survey! And you fill out the survey — it's usually 10 questions long — and the goal of the survey is to ascertain your opinion on local policing. The surveys never name an actual police department because the whole point is what is your view of your local cops. The second part of this thing is taking these answers, and this is sort of the black box and the challenge of reporting on private startups, [they] magically [translate] into a score, which is similar to a credit score.

Scott: And why do this ... why did the NYPD invest in this tool?

Weichselbaum: So it's an interesting history. For those policing nerds of the world like myself who are familiar with Bill Bratton, he goes back to New York from the '90s with CompStat. And that's developing a tool, which was pretty novel of the time, of mapping crime and having meetings holding police commanders accountable to crime in their precincts. So that was his first supposed or so-called revolution, if you will. Secondly, Bill Bratton in L.A., when he was police commissioner there, got behind the predictive policing movement. And predictive policing is really huge right now. So that's like hot spots, hot people ... Chicago has a heat list. And that, I would say, sort of cropped up in the aughts. 

Scott: And that's all based on data to predict crime trends, basically?

Weichselbaum: Right. So it's looking at data like time of day.... In Chicago, they're also looking at your history as a human being. It's a whole separate rabbit hole we can go down ... of predicting whether or not you are likely to be shot or do a shooting. But again, that's predicting behavior on past variables, which Bratton was behind, and that is alive and well in policing today. So next up on the list ... he wanted police polling [for public trust] the same way we do CompStat, which is like monthly, weekly ... [he wanted] these scores all the time.

Scott: So real-time data about how people are feeling about the cops.

Weichselbaum: Right. This is basically measuring emotion. So he came back to New York, and I was at the New York Daily News at the time, so I wrote a lot about this. Bill Bratton is praised by some in academia or some who follow policing, but in black and Latino neighborhoods in New York City, he was not praised. And there was a lot of pushback [when] our mayor, Bill de Blasio, chose him. So he's coming back to a city where people were so upset with him and he wanted to understand why — why despite crime going down in New York, there's still tension in minority communities. So when he came back, he ordered a survey, a traditional survey, which the founders of Elucd did. And to speed this whole thing up, Black Lives Matter happened, Ferguson happened, and then we had two officers here in New York shot and killed in Brooklyn by a gunman, I believe from Baltimore, who was angry and upset about the headlines about police brutality. So that really put this thing on like fast forward. So NYPD [worked with] the co-founder of Elucd, a man named Michael Simon, who's this wunderkind in data analytics who in his 20s worked for Obama and ... was looking at voter data. So his claim to fame is [identifying] undecided voters, which some say helped Obama into office in 2008. So sort of taking those tactics and applying it to policing.

Scott: So he founded this company, Elucd, that is offering these services not just to New York but now other cities as well? 

Weichselbaum: Right. So they're building it out in New York, and this is an interesting sort of branding/marketing conversation.... Once you have the NYPD behind you in policing in this country, you're good to go. So the NYPD basically co-developed this thing. They helped it from the ground up. LA is next. LA has now a letter of intent, meaning they plan on doing this. They're fundraising money to pay for this thing. Chicago's going to roll it out August first in some of their police districts. So it's growing in popularity. Once you have the three biggest police departments in the U.S. curious about a product, you're going to start rolling it out to small markets.

Scott: Whether or not it's actually successful.... I mean, you found that there are a lot of questions still about how these data are actually being used, how they can be effective. So how does it work if you see, say, a decline in the trust score on the Lower East Side. What's the response?

Weichselbaum: So that's the interesting part of this.... So when I push back with the commissioner of the NYPD and one of the heads of the LAPD, based on my reporting that this is really confusing — A) even how you're getting these scores, and then B) you know, as commissioner or deputy chief of big police departments, what do you want your commanders to do? So what was interesting is the top brass really didn't have a clear answer. Their answer was, "Oh, let the commanders figure it out." So it was interesting reporting out the story. The biggest critics were actual police commanders, who were very frustrated. And [were asking], "Do we really need this tool?"

Scott: And how representative are these surveys, particularly if they're happening mostly online, on social media versus a beat cop walking the street and talking to people?

Weichselbaum: Right, so, whoever these people are — people clicking, whether it's popping up on a gaming app or ads popping up on your Facebook ... you're a certain type of person. Whether you're a millennial, whether you have time on your hands ... whoever you are, do you represent an average resident of the city? You know, academics may feel differently, but the other side of this is that it's low cost and cheap to do. Traditional surveying is really expensive and takes a long time. So those who actually are into this idea feel that, well, it's not the greatest, most statistical gold standard of surveying, but it's cheap and it's easy. So let's keep building this out.

Scott: And you did find at least one police force that has a response plan for using this data. Talk about Grand Rapids and what they're doing.

Weichselbaum: So it's a smaller police department, it's not, of course, big by any means compared to New York or LA. And their chief said, "I'm going to use this as a tool to determine where I can send my community policing resources." So if scores are down in a specific area of his city, he'll figure out ... "Well, maybe I should have a backpack giveaway, maybe I have a car seat giveaway, maybe I should have a pop-up community meeting," is what he calls them. And my favorite solution was a police karaoke car. I was like, what's that? And that's when you send out a patrol car to a neighborhood and encourage the kids to come over and sing songs with the officer over the loudspeaker. So it's a little gimmicky, but he at least had a response that he mapped out and thought about. And when I interviewed the [New York] police commissioner last and explained Grand Rapids, he was like "No way... like, I don't want to just drill this down to gimmicky things like giving away backpacks. I want my commanders to have responsibility. I want them to think about this. I want them to really come up with creative solutions, and that's why we're not going to give out a playbook." So that was the interesting tension: You had these police commissioners and top city officials really excited about this, saying "We're not going to write a playbook, we're going to let the commanders figure it out" and the commanders are like, "We don't know how to figure this out, and we don't really think we need this tool."

Scott: So if we could pull back for a second.... Just from a law enforcement and public safety standpoint, I mean, is this a sign at least that police forces are taking issues of trust and public perception more seriously?

Weichselbaum: Yes, there is a positive story thread here.... So the first thing is you do have big city police departments now wanting to quantify and analyze sentiment, and that's a good thing. And their ultimate goal in New York and LA is to roll this out into CompStat. So as you're presenting your information that week or that month about murders in your precinct or robberies in your precinct ... one day, perhaps, you will also be asked about your trust numbers or your satisfaction numbers ... and perhaps one day, you will be asked in front of a big room of a top police officials: What are you going to do about it? And perhaps one day you will be held accountable. So that is what they say, at this point, is the ultimate goal. And accountability raising is always a good thing.