This article first appeared on Verdict.

Once again, it is the story within the story that matters most.

Just over a week ago, researchers published the results of the largest, most sophisticated study to date on the effect of body-worn cameras (BWCs) on policing.

The research team conducted a methodologically rigorous, randomized trial involving officers with the Washington, DC, police department, one of the largest departments in the country.

They tracked officer behavior for 18 months, from the middle of 2015 to the end of 2016, comparing more than 2200 officers throughout the city. Roughly half the officers in the study were assigned a camera and half were not.

All the participants held a rank of sergeant or below, which helped guarantee that the study group included only those officers who have the most frequent contact with the public.

The researchers examined the effect of BWCs on four “outcome categories,” including the use of force, citizen complaints, routine patrol activity, and judicial outcomes. The entire study is worth reading in detail and can be accessed here.

This, however, is the bottom line (the emphasis is mine): “Across each of the four outcome categories, our analyses consistently point to a null result: the average treatment effect on all of the measured outcomes was very small, and no estimate rose to statistical significance . …”

Body worn cameras made no difference at all in how the police went about their job, to citizen complaints about police conduct, or to judicial outcomes—at least, no difference that rose to the level of statistical significance. Of particular note for those concerned about police violence, officers who wore cameras were no more or less likely to use force than those who did not.

Observers were shocked. In article after article, commentators and participants expressed astonishment at the results. The New York Times said the study “defies expectations.” The Washington Post similarly thought it “bucks early expectations about the impact of the devices.” Another journalist thought the results were contrary to “widespread expectations.”

The scholars involved in the project were likewise baffled. The results “kind of blew my mind,” said Yale political scientist Alexander Coppock. Even Peter Newsham, the Chief of Police in Washington, DC, who presumably knows his department as well as anyone, was “surprised.” “I think a lot of people were suggesting that the body-worn cameras would change behavior,” he said, but “there was no indication that the cameras changed behavior at all.”

GettyImages-642765652A Los Angeles police officer wear an AXON body camera on February 18, 2017 in Los Angeles, California.  David McNew/Getty

Newsham had expected in particular that the cameras would deter those officers who were inclined to “misbehave.” But in fact, there were about as many complaints against officers wearing the cameras as those who were not, and once again, the differences between the two totals were not statistically meaningful.

I don’t get it. The story here is not that the cameras had no effect. It’s that so many people were certain they would.

Those who express surprise at the results evidently believe that at least some officers were not doing what the public and their superiors want them to do, and that the cameras would have induced those officers to change their behavior. People may disagree about the number of officers in this category, but the surprised reaction makes no sense if everyone thought the police were behaving precisely as desired.

Suppose, for example, we randomized the dog owners in a city, assigning half to wear a body camera when they took Fido for a stroll and half to wander without. Because we expect a certain range of behavior—stopping at a bench, chatting with neighbors, checking email, sending texts—we are not at all surprised when the footage shows precisely this result.

We are only surprised when the footage shows something we don’t expect, like a random number of people whacking poor Fido across the head.

In the case of the body worn cameras, people believe the police misbehave in some statistically significant number of cases. Since no one wants to be caught on tape being abusive, they firmly believe the cameras will deter the misconduct. And because they are convinced the abuse occurs, they are at a loss to explain the results.

Of course, Chief Newsham offered one possible explanation for the non-results. Maybe, he mused, his officers “were doing the right thing in the first place.” The cameras did not deter misconduct because there was no misconduct to deter. This is certainly one possible explanation.

But even Chief Newsham doesn’t really believe this, or else he would not have been so surprised at the results. Apparently, no one has considered the possibility that the police in DC, like in most cities, are doing exactly what is asked of them and that it may be abusive.

That is, body worn cameras detect police doing their job. Most of the time, this involves innocuous interactions with members of the community. Occasionally, however, it involves a certain amount of hostility between the citizenry and the police, which often leads to hard feelings, sometimes to a complaint, and sometimes to the use of force.

The citizen thinks the officer has been abusive, and perhaps she has by the standards of the individual aggrieved or by those of the community of which the individual is a part. But if the officer is acting precisely as she has been trained and encouraged to act, she is not apt to alter her conduct regardless of whether she is being filmed.

The camera, therefore, does not change officer behavior precisely because the officer is doing exactly what is asked of her, even when her behavior is perceived as abusive. Thus, the study did not, and could not, detect the possibility that routine police interactions could unfold precisely as designed and nonetheless still be the source of friction and animosity between the citizen and the department.

The study looked for evidence that abusive behavior was anomalous , which meant it was blind to evidence that abusive behavior was inherent in the nature of police-citizen interactions and accepted by those who design and implement police practices.

I have written about body worn cameras before. As I have noted in the past, in this age of so-called criminal justice reform, they are the only change in police practice that has achieved widespread acceptance.

According to the results of a survey recently completed by the Major Cities Chiefs Association and Major County Sheriffs’ Association, nearly every large police department in the country plans to implement body-worn cameras. More than 95 percent of the departments are either committed to the cameras or have already begun to use them.

Only 4.5 percent said they either do not intend to use body cameras or chose not to go forward with the technology after completing a pilot program. Though there is a veritable explosion of promising and exciting research in policing, some of which I have written about before, the increasing use of body cameras is, so far as I can tell, the only reform of police practices that has achieved or will soon achieve widespread acceptance in the United States.

And there is a reason for this acceptance: the cameras do not alter police behavior. Oh, perhaps at the margins they might. They may deter the small number of officers who violate long-settled norms.

But contrary to what many people believe, in the great majority of departments, these are the outliers. These officers matter, make no mistake, but removing them will not solve the problem or relieve the tension between police and the communities they serve.

The routine strategies and behavior that most officers have been trained to employ, which involve no risk of adverse personal or professional consequences, are precisely what most departments want from their officers. This behavior cannot be deterred by a camera, even though it carries with it the inherent, inevitable risk of friction and resistance.

Though there is a veritable explosion of promising and exciting research in policing, some of which I have written about before, the increasing use of body cameras is, so far as I can tell, the only reform of police practices that has achieved widespread acceptance in the United States. And the reason is simple: they don’t compel the police to change anything.

I have written many times about my preference for narrowly focused, place-based policing strategies that shrink the blue footprint, and therefore limit the number of these routine police-citizen interactions.

The police must become part of a robust partnership involving the community and other municipal agencies. Collectively and creatively, the members of this partnership must target the very small number of people and places that account for the overwhelming majority of crime and disorder in a community.

So far, however, these collaborative, problem-solving, place-based strategies are the exception, not the norm. Until that changes, body worn cameras will simply capture precisely what we have long asked the police to do. That’s why most police departments like them. And that’s the problem.

“My fear is that this attention on body-worn cameras has sucked all of the oxygen away, and real money and real resources away, from more meaningful remedies that we know already work,” said Yu.

Joseph Margulies is a Professor of Law and Government at Cornell University. He is the author of What Changed When Everything Changed: 9/11 and the Making of National Identity (Yale 2013), and is also counsel for Abu Zubaydah, for whose interrogation the torture memo was written.