Carlos Bam Torres estimates that he’s been stopped by Philadelphia police more than 10 times in the last two years, both on foot and in his car.

Sometimes a busted headlight turns into a car search. Sometimes he looks like a guy the police are looking for.
“They always say they got a call with my description,” says Torres, a 25-year-old from North Philadelphia’s so-called Badlands. “Hispanic male, tattoos and maybe the same color shirt.”
Torres’ story isn’t unique. Many young men in Philadelphia have run-ins with the police that they feel are unwarranted. In parts of Torres’ neighborhood, where there are four times more police stops than there are neighborhood residents, it’s less of the exception than the rule.
The Philadelphia Police Department, of course, has its own side of that story to tell. And now, nearly six years after the civil rights lawsuit against the city that galvanized public attention over stop-and-frisk policing in 2010, the department did just that. In April, the PPD released a massive log of more than 1.3 million pedestrian and vehicle stops since 2014, along with maps and words. It was the PPD’s way of telling its own story about stop-and-frisk.
But stories are complicated. They are as much about what’s being said as what’s not being said, and some are watching to see what the PPD leaves out. More importantly, in an age where supposedly routine police stops could lead to death by the hands of police, especially in communities of color, there’s great need to reexamine the everyday policing policies, and the narratives, that define our cities.

An unreliable narrator?

A police data dump usually means raw numbers, and later on, some maps, if civic hackers are in the mood.
According to Kevin Thomas, director of the PPD’s data division, a police data dump most often means: “Here’s the tools, here’s the dots, here’s the data. Have fun.” But stop-and-frisk was a different beast.
It was a complex, sensitive, source-of-a-lawsuit kind of beast. It needed context. So the PPD used a series of interactive maps created with mapping firm Esri’s Map Journal tool, which allowed the PPD to provide narration around the data.
“Discussion about Stop & Frisk tends to refer to randomly stopping people and arbitrarily searching them without cause or reason,” the PPD narration reads. “That is not an acceptable practice at the Philadelphia Police Department. The department has never had a ‘Stop and Frisk’ crime-fighting policy.”
The PPD is referring to the fact that officers are only meant to stop people when they have “reasonable suspicion,” but those can feel like empty words to someone like Torres. All he has is an officer’s word that he seemingly always matches the description of a perp.
“Sometimes they use those exact words,” Torres says.

The PPD’s data doesn’t tackle the issue, either. It’s one glaring hole in its narrative. The question of “why” the officers stopped Torres and others like him has been scrubbed from the dataset, due to technical and confidentiality issues, which means that, looking through the data, you can’t tell what the officers recorded to justify a stop.
Even without this “why” part of the data, the PPD narrative tries to answer heavier questions about racial patterns in its policing.
The PPD’s stop-and-frisk data story argues that these stops aren’t occurring in a vacuum.
“The question is often raised about whether a specific racial or ethnic group is over represented in pedestrian and vehicle stops,” the narration reads. “On the surface, a cursory examination of the racial distribution of the pedestrian and vehicle stops, may suggest this is the case. However, it is important to consider all of the circumstances when interpreting raw statistical data.”
The PPD narrator suggests that socioeconomic factors like poverty and neighborhood demographics are driving what watchdogs see as a disproportionate use of stop-and-frisks against people of color.
To make sense of police stops and how they are used, the PPD story also says to consider violent crime. It offers a simple formula: Police presence corresponds with areas where people are most likely to become victims. From the bird’s eye view, it’s a palatable narrative. Swipe back and forth between the maps of where violent crime and police stops occur — they are near mirror images.
But in the case of Torres’s home neighborhood, it’s less about the optics than the sheer numbers themselves. If we want to get the heart of stop-and-frisk policing, we have to look where stops are happening the most.

The context is hyperlocal

The West Kensington and Fairhill areas are ground zero for this practice in Philadelphia.
In two years, there were 24,212 police stops in one West Kensington census tract with a population just over 5,000 people, according to the police data. That equated to roughly 33 stops a day, in an area that measures about one-fifth of a square mile. That’s more police stops in one sliver of a low-income neighborhood than occurred in nearly all of Center City, from I-676 to Walnut Street, from Old City to Rittenhouse Square, from City Hall to the Art Museum.
This is near where Torres grew up, where he hangs out. It’s a polygon that spans roughly from Lehigh Avenue to Westmoreland Street, from B to E Streets. A neighboring slice of Fairhill saw 22,380 stops.
These pedestrian and vehicle stop numbers are off the charts. Tracts with similar violent crime rates saw between 4,000 and 11,000 stops during the same two-year period. Even a tract in Frankford with slightly more violent crime than Kensington and Fairhill had half the number of stops.
So what is it about these neighborhoods?
In the PPD’s headquarters near Chinatown, flanked by his team of data analysts, Thomas is quick to contextualize the situation. Kensington and Fairhill’s drug market is known across the Eastern seaboard. It’s a hotbed for all other manner of criminal enterprises, with its swollen population of drug addicts and dealers.

There's no such thing as a 'definitive' stop-and-frisk narrative.

What about central Northeast Philadelphia, where three census tracts stand out with a relatively high number of stops? The neighborhoods are largely white and middle-class. Violent crime incidents are few and far between. But in these areas that hug Roosevelt Boulevard, police made the same number of stops that they made in areas with ten times the violent crime and poverty.
“Motor vehicle violations,” Thomas says, is the likely explanation.
It’s another story. One that has much less to do with race, crime and poverty. That story would need its own color-coded maps on the PPD’s data journal, pointing viewers to the rates of vehicular carnage along the notorious Boulevard.
Are these plot holes in the police department’s narrative? No, Thomas says, they’re more like untold stories within the story. That’s the problem with any sort of “definitive” stop-and-frisk narrative. It couldn’t capture all the complexities.
Thomas says his data team couldn’t possibly tell the stop-and-frisk story from every angle.
“Just this car stop and ped stop [data], you could slice this thing up in so many millions of different ways,” he says. “We would never stop.”
A recent audit of stop-and-frisks revealed that 433 frisks yielded only two firearms. Moreover, in cases where “a bulge” justified the frisk, police found only one gun in 78 frisks. But in the Kensington-Fairhill area, the record-high number of police stops are clearly yielding something else.
This is an area with a record-high number of annual narcotics arrests, an area where ticketing nuisance offenses saw an uptick in recent years, an area in which every policing practice from broken windows to beat cops has been tried and tested. As the focus gets smaller, this is less of a story about stop-and-frisk than it is about how the city fights the “war on drugs.” Has any of it worked?
That’s one story the PPD map doesn’t tell.

Reclaiming the stop-and-frisk narrative

When it comes to bad practices, the PPD’s research team says it is monitoring its own officers, too.
While they were storyboarding the Esri map journal throughout April, Lisa Howdyshell, a data scientist for the department, discovered an alarming pattern. It fit civil rights attorneys’ description of the practice as disproportionately targeting Black and brown people. In the predominantly white Roxborough neighborhood, 5th District cops were stopping a far higher percentage of young Black males than actually lived in the neighborhood.
So the research team dug deeper into the data. They passed questions around the department. It turned out that the overwhelming number of these stops were happening around a high school with a large African-American student body. Officers reportedly made these stops to address truancy during schools hours. Data on those stops — geo-mapped, timestamped — backed up that account, Howdyshell said.
But there are some watchdogs who remain unconvinced of the police’s justification of racial patterns for stop-and-frisk, like civil rights attorney David Rudovsky, whose firm co-filed the civil rights lawsuit against the city in 2010.
Looking at the PPD’s stop-and-frisk Map Journal, he points out that the police’s story doesn’t mention, or explain, the high number of stops conducted by police without reasonable suspicion, regardless of race.
According to an audit of the PPD’s stops, 33 percent of stops made in the first two quarters of 2015 were conducted without reasonable suspicion, meaning the stops were illegal, and African-Americans and Latinos were more likely to be targeted.

The public can directly fact-check what the PPD is saying about stop-and-frisk.

Another watchdog, Mary Catherine Roper, the deputy legal director for the American Civil Liberties Union (ACLU) Pennsylvania, said that she disagrees with “the Department’s suggestion that stops only occur where the crime statistics show the need for greater policing.”
“And,” she adds, quoting from the PPD’s narration, “it is historically untrue to say the Department ‘never had a crimefighting policy of stop and frisk.'”
Despite their criticisms of how the department is attempting to reframe the story, both Roper and Rudovsky give credit to the new transparency tool. The fully mapped narrative story could be the next big chapter in a push for a more transparent department, an effort spearheaded by former Commissioner Charles Ramsey and carried forward under Commissioner Richard Ross. The Map Journal data release was a “courageous” move, Thomas adds, one that was met with heavy internal criticism.
The police’s marriage of narrative and data will hopefully encourage more honest public dialogue, so long as people keep asking questions.
“I would hope that when the public gets data, they look at the data and don’t just read the sidebar,” Roper says.
She continues:

Having the narrative and the data side-by-side means that, finally, the public and the press can directly fact-check what the PPD is saying about stop-and-frisk. That’s great. If the data doesn’t match the narrative, that will say something important about the PPD’s willingness to confront the evidence about its practices. Up until now, the public has only had narrative from the PPD.

The open data allows the public to interrogate the sheer volume of police stops — where they occur and who they target. That’s especially powerful information, but if people know how and where to access that information is another question.

New technology, limited resources

The department’s first attempt at data-driven storytelling didn’t have the widest reach. Since the launch of the Esri map journal in early April, it has accrued around 1,800 page views, according to Thomas.
Right now, data-driven narratives fall low on the research team’s priority list. Thomas says that 95 percent of the research department’s time is spent on data-driven crime prevention strategies, which are for internal eyes only. What’s more, the Esri geomapping narrative tool hasn’t gotten federal approval as a method to publish sensitive information. (The stop-and-frisk story was a trial run and the data was scrubbed to protect individuals’ identities.)
If resources were infinite, the data team could tell a geo-mapped crime story for every neighborhood of the city. Or they could use the narrative platform to address quality-of-life issues that concern city residents every day. They could theoretically reconstruct officer-involved shootings, rather than issue the standard press release 72 hours after the shooting occurs.
Thomas says there has been talk of using Esri’s map journal tool to tell often-overlooked stories of officers who receive commendation. But all of those require extra personnel in the research lab, and the PPD is currently lacking across the board.
It’s a brave new world, but one that won’t be fully explored unless the order comes from higher up.
“If this is something that the mayor or commissioner wants us to run with, of course we would,” Thomas says. “That would just mean something else would take a backseat.”