Bill Bratton (left), Bill de Blasio (second from left) and others after an active shooter training exercise in 2015.

One day in 1995, two officers from the New York Police Department walked up to a podium at Harvard University’s Ash Center. Louis Anemone and John Yohe were representing the department as a finalist in the Kennedy School’s Innovations in American Government competition, and their excitement about the force’s new, computerized crime-fighting system was palpable. “It’s revolutionizing the way the NYPD polices the city of New York,” Anemone told the judges. Giving officers rapidly-updating maps of crime all over the city, the system was “a shot of adrenaline to the organization of the NYPD,” the officer stressed, “right to the heart.” Previous decades had seen a tremendous rise in crime, but with the advent of CompStat, as it was called, the police said they were finally able to flatten the curve.

CompStat not only won the competition; it went on to become the new gold standard for measuring and combating crime. Most major cities in America have adopted forms of the system, and so have many world capitals. By bringing crime-fighting into the age of big data, CompStat led a global revolution in policing, homegrown in New York City.

But 25 years after Anemone’s laudatory speech—June 24, 2020, to be exact—the NYPD captain’s union was singing a different tune. “I believe Compstat to be the primary driving force in undermining police/community relations,” wrote union president Chris Monahan in a letter to Mayor Bill de Blasio and Police Commissioner Dermot Shea. “This inherently creates tensions between black and brown communities and the police. Compstat has always been used as a means of embarrassing and coercing commanders into more proactive policing. Simply put, the NYPD MUST find another way.” 

In other words: end CompStat now.

In its short life, the NYPD’s technological golden child has garnered as much criticism as it has praise, with opponents in and outside the force. Some, like the captain’s union, claim it encourages racial profiling, blaming it for the outgrowth of stop-and-frisk under Mayor Michael Bloomberg. Others say the system has led to widespread manipulation of crime statistics. Still others question whether CompStat can really take credit for the massive decrease in crime it supposedly precipitated.

In a time when police tactics are coming under intense scrutiny, when people across the country are calling for comprehensive reform or even total abolition, CompStat—one of the most influential innovations in modern policing—deserves a hard look. And the only way to do that is to go back to the beginning, to when CompStat was just a couple pins on a map.

————

Jack Maple, a subway cop who’d joined the force at age 18, was not a computer geek. Officers on other beats looked down on the so-called “cave cops,” but Maple made up for his low status by living beyond his means, frequently donning a Homburg hat and wing-tip shoes to get a drink at the Plaza Hotel’s Oak Bar. “I always felt I belonged there,” he told New York magazine for a 1983 profile.

As he chased after pickpockets and purse thieves in the 1970s and early ’80s, he began to develop what he called “Charts of the Future”: 55 feet of maps of the city plastered over his office walls, with color-coded pins marking the date and time of each crime reported. Instead of blindly making patrols through the subway, police could station themselves at hotspots—”putting cops on the dots,” as Maple would say. They moved from simply reacting to crime to anticipating it.

The results Maple reported were remarkable, but they went unheralded until Bill Bratton, formerly transportation police head in Boston, was hired as NYPD transit commissioner in 1990. Like Maple, he was dead set on success. “My whole career,” he would later write, was about “making it to the top.” Like Maple, he saw the crime charts as the way to get there, and he gave the then-transit lieutenant 100 cops to implement them throughout the “caves.” According to the magazine Government Technology, they quickly reduced robberies and felonies in the subways by about one-third.

When Bratton was promoted to NYPD commissioner by Mayor Rudy Giuliani in 1994, he took Maple with him as his deputy commissioner. Both promotions received sneers throughout the force: Bratton’s for his perceived lack of dues paid on the street, and Maple’s for his rapid jump through the ranks. Maple would write in his memoir that it was like “an ensign in the Coast Guard waking up as a three-star admiral in the Navy.” But they were intent on shaking up the department, by computerizing and automating Maple’s Charts of the Future into the digital screens of CompStat. Where crime statistics were once cobbled together every few months, every precinct was now responsible for collecting and reviewing them each week.

The real revolution, though, was not so much the data collection as what Bratton and Maple did with it. To ensure that their cops were actually taking advantage of the statistics, Bratton and Maple held weekly meetings at 7 a.m. where they would grill precinct commanders on the nitty-gritty of what crimes were happening under their watch, why, and what they were doing about it. NYPD brass used CompStat to both track and measure productivity.

The intense interrogations could make or break careers, and not without pushback. Police columnist Leonard Levitt wrote that the early meetings were “literal free-for-alls,” complete with fistfights, chair-throwing, and accusations of “treason” and “heresy.” Eventually the force accepted the new reality. “Most captains, including myself, would rather have had monthly root canals,” recalled NYPD Captain Ernie Naspretto in 2013. “But we soon realized that the process made us all sharper, more focused and, quite frankly, smarter.”

Coming hand-in-hand with CompStat was the advent of “broken windows” policing, which posited that crime at its core was driven not by poverty or racial oppression, but by a “disordered” environment. The theory has been accused since its inception of unfairly targeting people of color and the working class, but Maple and Bratton were pious adherents. In order to curb serious crimes like drug trafficking, sexual assault, or murder, the duo pressed commanders to scrupulously clean the city of small-time offenses like graffiti, public drinking, and the well-known squeegee men.

And, by their account, clean up they did. In the first nine months of 1995, barely a year after CompStat was implemented, murders fell by nearly 30 percent, car thefts by 25 percent, and robberies by 20 percent. Over the rest of the decade, the city’s crime rate dropped to record lows not seen since the mid-1960s.

Bill Bratton even made the cover of Time magazine in 1996, under a caption declaring, “Finally, we’re winning the war against crime.” New York was becoming the poster child of a new era of proactive policing and safe cities. And yet, Giuliani was irked. Rumors spread that he coveted the spotlight shining on Bratton. He launched a probe into some of his commissioner’s questionable conduct, like signing a $350,000 book deal about his successes and vacationing in the Bahamas on the dime of wealthy financiers. Rather than weather an investigation, Bratton resigned in March 1996, just 27 months after becoming commissioner. Maple left soon after.

Nonetheless, CompStat’s renown, as well as its inventors’, was sealed: Bratton and Maple’s crime-fighting computer had made New York safe again. The duo took it to the private sector. Bratton consulted for the LAPD before becoming their commissioner, while Maple and one of his colleagues received a $1 million contract to implement CompStat in New Orleans. Bratton’s book, Turnaround: How America’s Top Cop Reversed the Crime Epidemic, came out a few years later and was a bestseller. Maple’s own memoir, The Crime Fighter: How You Can Make Your Community Crime Free, was not, but before his death in 2001 he helped develop a television series, “The District,” partly based on his life. If CompStat could make or break careers, it certainly made theirs.

————

By the turn of the millenium, police commanders across the country were visiting CompStat meetings hoping to adopt the system, and even the CIA was showing interest. A whole “cottage industry” of NYPD consultants, in the words of sociologist Andrew Karmen, had developed to implement the system throughout the U.S. and abroad. But the cracks in the system were already starting to show. In 2003, the NYPD admitted that 203 crimes reported to police on the West Side the year prior had been improperly downgraded—felonies had been changed to misdemeanors in order to artificially maintain the decrease in crime. This kind of intentional misclassification had been discovered at least five times since CompStat’s inception in 1994, but the police insisted that the bulk of their data was accurate. Besides, a spokesman assured the press, “the citywide misclassification rate has steadily declined in recent years.”

The question of data manipulation nevertheless remained open. In 2010, criminologists Eli Silverman and Joe Eterno released a study that they said demonstrated endemic data manipulation within CompStat. They had surveyed hundreds of retired senior NYPD officers who admitted  being aware their colleagues were cooking the books. When a robbery was reported, an officer might search for the stolen item on eBay to find lower prices that would let them downgrade the crime. Or they might just refuse to report the crime, telling the victim it just was a case of lost property. Either way, the pressure to keep crime numbers low was encouraging police to shirk their responsibilities. Many officers in the survey said that the biggest challenge on the job was not risking violence in the field but rather facing upper management at the 7 a.m. meetings. “My job is to make sure we don’t take a hit on a number unless we have to,” said one.

“We were actually attacked in the press” for releasing the study, Silverman recalled to Bedford + Bowery. Op-eds appeared in local tabloids and the New York Times decrying the professors’ “anti-cop idiocy.” Mayor Michael Bloomberg questioned the legitimacy of the research, claiming incorrectly that it had been funded by a police union, while the NYPD defended themselves with a previous NYU study that purported to find no manipulation, without mentioning that that study was sponsored and overseen by the department itself.

The official denial of wrongdoing was swift and unequivocal, likely because another even more outrageous controversy had already been brewing. In 2010, the Village Voice released tapes made by officer Adrian Schoolcraft showing that cops in his precinct, the 81st, were being instructed to make arrests simply to keep up arrest numbers, typically in Black and Latinx neighborhoods. “Everybody goes,” the precinct commander orders in one recording. “I don’t care. You’re on 120 Chauncey and they’re popping champagne? Yoke ’em. Put them through the system. They got bandannas on, arrest them. Everybody goes tonight. They’re underage? Fuck it.” Before going to the press, Schoolcraft had reported these issues within the department. In response, members of his precinct dragged him out of his apartment while handcuffed and committed him to a psychiatric ward for six days.

This shooting-from-the-hip policing wasn’t limited to the 81st precinct. Under NYPD Commissioner Ray Kelly, who took over from Bill Bratton in 2002, stop-and-frisks—where police detain and search people without warrants—had skyrocketed, from 97,296 stops per year in 2002 to 685,724 in 2011, according to the NYCLU. Eighty-eight percent of those stopped in 2011 were innocent, and the majority were Black or Latinx. Criticism grew that the stops were the very essence of racial profiling. Mayor Bloomberg came to the NYPD’s defense, claiming that what looked like racial profiling was simply more of Bratton and Maple’s “putting cops on the dots”—the police were searching people where crime was known to happen more. “Incidentally,” the mayor said in a 2013 radio interview, “I think we disproportionately stop whites too much and minorities too little.”

————

NYPD officials have been adamant that arrest quotas do not exist, and that officers are not instructed to target specific racial groups. But Schoolcraft’s revelations and other lawsuits that followed bely that claim.

Silverman believes that Compstat was a victim of its early successes. Once the crime numbers start dropping, no one wants them to stop. “Every mayor and every police commissioner since 1994 has said, ‘I was responsible for crime declining,'” said the professor. So they increase the pressure on beat cops to deliver good numbers, to not only maintain the current low but make it even lower. “It’s like squeezing an orange,” he said. “When you first start squeezing, the juice comes out quickly and readily.” But as you tighten your grip, the fruit dries up. 

In our 21st-century world of big data, careers in government rest on good-looking numbers. This holds as true for mayors seeking reelection as it does for beat cops trying to hold on to their posts. Even doubt about data manipulation can become a political tool. In 2015, a year after Mayor Bill de Blasio kicked out Ray Kelly as commissioner, brought back Bill Bratton, and dramatically cut stop-and-frisk rates, Kelly accused the mayor of playing with the numbers to create record lows. He made the comments as he was preparing for a mayoral run, which he later aborted.

Cops have long been frustrated by the pressure cooker of CompStat meetings. A 2016 NBC News report shows police preparing by pouring over their neighborhood stats like teenagers at a debate tournament, before facing the wrath of the top brass the next morning: “This is going on too long,” the current commissioner, Dermot Shea, tells one police inspector. “We need it fixed yesterday!” Memes have even been made satirizing the tense meetings.

If this has been going on for decades, then why the call to end CompStat now? With the NYPD under intense scrutiny over its response to the George Floyd protests, the captains are in a tough spot, said Silverman. According to the professor, they are still getting pressed to bring their crime numbers down, but many feel their patrolmen have been more hesitant to take action.

But change looks unlikely. If you ask Commissioner Shea, who rose through the ranks on a reputation for intense analytic acuity, ending CompStat is out of the question. “There is an assumption, that I disagree with, that CompStat is directing activity,” he told Pix11. “That is the furthest from the truth. It was never about summonses and arrests for the sake of numbers. It’s about improving the quality of life. So there is no plan to cancel CompStat.” 

After all, who would want to get rid of the one thing that finally made New York City safe again? Yet, even CompStat’s role as city savior must be put into question. The reality is there was a massive drop in crime all across the nation beginning the late 1980s and early 1990s; New York’s began in 1991, two years before Bill Bratton became commissioner and three before he and Maple began CompStat. Cops columnist Leonard Levitt noted that in 1995, the first full year the system was in place in the city, the leading city in decreased crime was not New York, but Seattle. The police there did not attribute the success to a fancy policing innovation, nor, in fact, to anything at all. “In all honesty,” the Seattle police spokeswoman told Levitt, “we don’t know why. It’s too soon to tell whether it’s a trend or a fluke.”

The economist Steven Levitt (no relation) found that while innovative policing strategies like CompStat were given the most credit for the crime drop by the media, there is little data to support these claims. He concluded that “the impact of policing strategies on New York City crime are exaggerated, and that the impact on national crime is likely to be minor.” Likewise, Patrick Langan, former senior statistician at the Department of Justice, wrote in a study of New York’s crime decline that “scientific proof of CompStat’s success is hard to find.”

Crime is a nebulous thing to measure and an even harder one to explain, but nonethetheless scholars have put forth myriad hypotheses for the crime rate plummet that began in the 1990s, some straightforward and some rather kooky: economic growth, an aging population, mass incarceration, an increase in executions, a growing police force, wider access to abortions, and, strangely but convincingly, the phasing out of leaded gasoline. Each of these theories has its critics and counter-theories, but it is easier to entertain the idea that multiple factors worked in tandem to decrease crime rather than insist that only one did. 

————

Despite these debates, CompStat has maintained its shining status within the police. If anything, it’s only gained esteem nationally after cities like New Orleans and Santa Cruz made controversial forays into predictive policing. Even though CompStat provided the data-collection infrastructure that predictive policing is based on—Bratton himself pioneered it in Los Angeles and cited sci-fi film The Minority Report as inspiration—early-adopter cities have seen backlash over the idea that police strategy would be driven by computer algorithms. Now, some police departments are promoting a “return to CompStat” as a more humanistic approach, said Andrew Guthrie Ferguson, a professor of law at American University and author of The Rise of Big Data Policing. These departments are not giving up on collecting data—that is only increasing—but they are making sure the analysis will be done by human beings. 

If the top brass of the NYPD is adamant about keeping CompStat, then why would the captain’s union call to end it? Ferguson wonders if it isn’t a scapegoat. “They are looking for justification for why they’ve treated the citizens of New York like statistics,” he said. “CompStat wasn’t looking at the individuals underneath it. They were looking for numbers—are your numbers going up, or down?” Now, with outrage over police abuse growing exponentially, the union is calling CompStat the “primary driving force in undermining police/community relations,” almost to say, “Hey, it’s not our fault—we were just following the numbers.”

In this way, CompStat—a mechanism created to hold police accountable—can also be used to skirt accountability. In 2015, Michael Bloomberg defended stop-and-frisk to an audience at the Aspen Institute: “One of the unintended consequences is people say, ‘Oh my God, you are arresting kids for marijuana that are all minorities.’ Yes, that’s true. Why? Because we put all the cops in minority neighborhoods. Yes, that’s true. Why do we do it? Because that’s where all the crime is.” 

For Bloomberg, you can’t debate the data. Yet Ferguson argues that data is not neutral. Numbers tell stories, but the story changes depending on what numbers you choose. “When you’re only focused on numbers,” said the professor, “you stop thinking about the context of how crime numbers arrive.” It has been argued time and time again that communities of color and the working class are locked in cycles of poverty and unemployment, conditions that “breed crime,” in the words of Patrick Langan. And likewise, that incarceration drives recidivism. Simply mapping reports of crimes does not tell you why those crimes are happening. And making more arrests does not tell you what impact you are having on the communities you police. 

“We didn’t have a number for that,” said Ferguson. “Community trauma numbers were never counted. The humiliation-of-stop-and-frisk-for-the-30th-time numbers were never counted.” 

There are initiatives to change what numbers police count as important, like CompStat for Justice, developed by the Center for Policing Equity, and CompStat 360, by the Vera Institute of Justice and the National Policing Center, which track parameters like police bias, community satisfaction, abuses of power, and officer well-being, though they are not yet widely implemented.

Just judging by the names of these projects, it is clear that CompStat is not going away any time soon. We long since crossed into the era of big data, where numbers drive the decisions of major institutions as much as they do cops patrolling the subway. Instead of laying the blame on CompStat, perhaps the captain’s union should ask themselves if they are counting the numbers that matter, and if there is more to fighting crime than putting cops on the dots.