The following was originally posted on JustSecurity.org. Just Security is based at the Reiss Center on Law and Security at New York University School of Law.

There are two huge problems with American policing today: We don’t know nearly enough about what works in a sound way, and what doesn’t — especially if one considers social costs, which usually get left out of the equation. And even when we do have a good fix on what works and its costs, there’s no real mechanism for identifying best practices and getting the word out.

Both of these problems were on display in Cincinnati recently, when 300 to 400 people attended the annual meeting of the American Society of Evidence-Based Policing (ASEBP). Never heard of it? That’s too bad: it was a terrific gathering of hardy souls, all deeply committed to the novel concept that policing practices actually should be based on sound science.

The meeting in Cincinnati served to underscore the low profile of evidence-based policing in the United States. Although 300 to 400 people may sound like a fair number, here’s some perspective: There are over 700,000 cops in the United States today; the International Association of Chiefs of Police’s most recent annual conference in the U.S. attracted about 18,000 officers and associated leadership, vendors, etc. Add to that the fact that ASEBP members teed up some basic questions about research and how to utilize it that have no adequate answers at present, and you start to see the issue.

There’s nothing magical about the tenets of evidence-based policing, a movement launched just over two decades ago by noted criminologist and policing educator Lawrence Sherman. It’s pretty much what you’d hope and expect. As Sherman describes it, “police practices should be based on scientific evidence about what works best.” Cynthia Lum, herself a noted criminologist, former cop, and director of George Mason University’s Center for Evidence-Based Crime Policy, explains how inescapably logical the movement is: “Why wouldn’t police tactics be based on what we know are effective strategies that reduce or prevent crime?”

The bad news is, for all its simple logic, evidence-based policing has not really caught on. Lum concludes her thought by saying, “There is little indication that most American police leaders and their agencies systematically or regularly use tactics that are evidence based.” The absence of more robust attendance at the Cincinnati meeting reflects this.

Evidence vs. `Experience’

What, you may be asking yourself, would the opposite of evidence-based policing even be? Who could be against such a thing?

The answer, at least as Lum and some of her co-authors tell it, is “experience,” which is to say that cops would much prefer to go on intuition, anecdote, something that happened to them once, or what travels by word of mouth, rather than what data and careful study would tell us. Seeking input from officers in one reasonably sized California agency, Lum and her researchers learned that cops don’t read much, and what they think is true can differ pretty widely from what research suggests. Add to this the often-held opinion among officers that the general public just can’t understand how uniquely difficult their jobs are, and you can see why they might overlook research in favor of this “experience.”

Take the bread-and-butter of policing: patrol. Over 40 years ago, criminologists studied the efficacy of random, motorized patrol in Kansas City, Missouri. One of those criminologists was the legendary George Kelling, the recently-deceased patriarch of — among other things – broken windows policing, the now-controversial theory that the way to fight crime is to eliminate disorder on the streets such as litter or broken windows.

They found that having police officers drive cars around the city serves neither to deter crime nor make people feel safer. There were methodological challenges to the study, but among those who follow policing research, there’s pretty much zero doubt that random patrol is a poor use of police resources. Still, it remains in wide use, and the cops in the study by Lum and her co-authors thought patrol and running from one 911 call to another (itself of dubious value) were the way to go. On the other hand, they didn’t think much of the one thing study after study has shown to reduce crime: hot-spot policing. Hot-spot policing involves intensive patrol where crime actually occurs. (We didn’t promise rocket science here.)

In case you weren’t already persuaded of the value of basing policing practices in actual evidence,  consider what happens when such techniques are not in use.

Start with facial recognition. The FBI and the Transportation Security Administration, along with local law enforcement agencies in Oregon, Michigan, and Florida, have already started using the technology. Immigration and Customs Enforcement (ICE) is too — according to a recent report in the Washington Post, FBI and ICE agents have scanned through millions of Americans’ photos in state driver’s license databases, all without their consent.

Yet, there has been no real quantification of the technology’s perceived benefits to policing, let alone the apparent costs. What we do know is troubling: in their current state, algorithms are more likely to misidentify people of color, women, and young people. Privacy concerns pose another huge social cost. Now, Congress is scrambling to catch up, trying to evaluate the costs and benefits as the horse is trotting out of the barn. Evidence of costs and benefits should have been weighed before the technology was ever deployed. And if the research was lacking, more should have been collected.

A Spike in Police Shootings in Phoenix

The recent outcry over policing in Phoenix offers another example. Even before the release of the viral video showing officers shouting obscenities and pulling guns on a couple and their young children, the Phoenix Police Department had drawn heat for an alarming spike in shootings by officers last year: 44, more than any other police department across the country, and nearly double the department’s average in recent years.

The department admirably tried to better understand what was happening, and brought in the National Police Foundation to dig into the numbers. After a six-month study, the Police Foundation couldn’t pinpoint any single cause of the increase but made several recommendations, including that the department change its training and reconsider the way officers respond to people in mental health crisis. The police chief and mayor have promised to do so.

What is frustrating, however, is that the Phoenix Police Department had the same data the Police Foundation did — numbers on police shootings, other uses of force, crime, arrests, and so on — but maintained that information for “administrative purposes and not research interests,” according to the Police Foundation’s report. In other words, the evidence that helped inspire these changes by the department was already there. It just wasn’t being used in the most effective way. That’s precisely what evidence-based policing aims to change.

Before police can use research in their decision-making, however, there must be more of it to draw upon. There’s a serious shortage of research on policing, and too little that makes the leap from the academic world into practice. Academics have their research projects about policing, but police chiefs regularly complain that when their agencies agree to act as guinea pigs, they don’t get much out of it in a timely fashion. Scholars study some aspects of policing over and over (like hot-spot policing) and yet there are enormous gaps on coming issues like predictive analytics, cybersecurity, no-arrest policies for misdemeanors, facial recognition, and much else. We operate in the dark far more than we should.

Even when we do research, it tends to focus on “what works,” without nearly enough attention to the “social costs” of policing, i.e., the impact of police tactics on those who actually are policed.  You can’t really draw conclusions about stop-and-frisk without asking what it does to the people stopped, and to community trust and cooperation with the police. You can’t talk about facial recognition and ignore privacy costs or racial disparities. We’ve seen what happens when these very policies and tools were adopted without consideration of the social costs: public outcry, lawsuits, and loss of trust.

No Storehouse of Knowledge

Even when we do learn something that seems firm and knowable, the U.S. has no central repository of that knowledge. At the ASEBP meeting in Cincinnati, participants asked the simplest of questions: Where can we go to learn what research tells us? The embarrassing answer to this sound question is: pretty much nowhere.

The National Institute of Justice runs crimesolutions.gov, which was a great idea for collecting information but seems overly rigorous in what can be included, and is underused generally. Lum’s Center has its own cool resource, an evidence-based policing matrix that visually categorizes studies police tactics based on the nature of the target, level of specificity, and whether the strategy was proactive or reactive. The goal is to translate this research into practice, and give police departments evidence to use in developing or assessing their own strategies. But scanning it only underscores the paucity of studies about the methods and impacts of policing.

What’s really remarkable is the absence of any entity that actually delves into the research, identifies best practices, and then disseminates them to the profession. The American Law Institute has a project, Principles of the Law: Policing, that is trying to do this at a very high level (disclosure, one of us is part of that project), but nothing at the nuts and bolts level that the police actually can use. It’s no wonder cops overlook what research is available if there are no formal mechanisms through which to share new, evidenced-based findings and make recommendations as to practice.

Contrast this, for example, with the U.K.’s College of Policing, whose mission is to make sure “policing practice and standards are based on knowledge, rather than custom and convention.” The College of Policing then uses this evidence-based knowledge to set standards for officers and departments. The College is far from perfect. Importantly, it doesn’t seem to focus on the social costs of policing much either; the entire discussion of cost-benefit analysis on its website neglects this crucial piece. Still, this is a sensible way to move policing into the 21^st Century.

For now, evidence-based policing seems to be a niche approach struggling for approval. Kudos are due to the small segment of American policing, present in Cincinnati, that seems to care about what research can tell us. And to the institutions that are trying to actively support and nourish the radical notion that research and experimentation are what are needed to make us safe. Such institutions include Arnold Ventures, one of the few large funders and boosters for evidence-based policing, as well as the Police Executive Research Forum and the National Police Foundation. They seem to take research knowledge seriously.

Many more people should care. We should seriously be considering a U.S. College of Policing. It’s well past time that we expand our research and evaluate police work based on evidence. Until we better analyze police strategies, policies, and technologies, and learn, in a quantifiable way, what works and what doesn’t, we are not truly advancing public safety. Until we insist on policing based firmly in evidence, we are neglecting the first job of government: keeping the citizenry safe.