POLICE REFORM IN DIVIDED TIMES

How can American policing be transformed into a more effective and egalitarian system of public safety? We need to start with four key facts. First, crime has devastating, disproportionate impacts on poor people and people of color, especially Black Americans. Second, police violence and other forms of abusive law enforcement also take a tragic and outsize toll on poor people and people of color, and here, too, Black Americans are particularly likely to be victimized. Third, improved policing has helped make crime far less common today than thirty or forty years ago, but some of the progress has been lost in recent years. Fourth, there have been successes over the past several decades in reforming police departments, but the victories have been partial and very often fleeting.

Fear of crime is often whipped up for partisan purposes, but the damage that crime inflicts on victims, as well as on their families and communities, is real and massive. Criminal victimization is also regressive, falling most heavily on those who are already disadvantaged. All of this is particularly true of the most extreme forms of violence—homicide, aggravated assault, and rape—which can fairly be called epidemic in the United States and which victimize African Americans, along with their families and neighborhoods, at greatly elevated rates.

There are between fifteen thousand and twenty thousand homicides annually in the United States—a rate of about five or six per one hundred thousand people in the country. Among Black Americans, though, the rate is much higher. African Americans die violently at seven times the rate of whites; for men the ratio is nine to one. Homicide is the third-leading cause of death among Americans aged fifteen to thirty-four; it is the leading cause of death among Black males under forty-five and the second-leading cause of death among Latino males under forty-five. Young Black men are fifteen times more likely than their white counterparts to be the victims of homicide. Violence is responsible for more lost years of Black male lives than cancer, stroke, and diabetes combined. Meanwhile more than a million Americans are hospitalized each year from attacks that do not turn out to be fatal, and African Americans are more likely than whites to be the victims of these attacks, too. Black people are also disproportionately represented among the several hundred thousand victims of rape each year in the United States.⁹

The failure to protect African Americans and other marginalized populations from crime is among the starkest and most damaging forms of racial inequality in the United States. No other wealthy country tolerates such extreme racial disparities in the risks of violent victimization.¹⁰

Moreover, beyond the lives that it cuts short, homicide and other forms of extreme violence can have tragic consequences for the families of victims and for the neighborhoods where it occurs. High rates of violence make fear a constant presence in people’s lives, affecting the material conditions of their daily existence in countless ways. It turns heat waves more deadly, for example, by making people afraid to leave their homes. Children living in neighborhoods with high rates of violence perform worse in school, reinforcing the cycle of disadvantage that keep families locked in intergenerational poverty. This is not just a matter of correlation: Black schoolchildren do dramatically worse on standardized tests in the days immediately after a local homicide than in the days just before. Exposure to lethal violence makes it hard for them to concentrate, and the effects appear to accumulate with each additional killing. Crime, especially homicide and other serious forms of violence, also depress property values, helping to maintain the gaping disparities between the household wealth of Americans of different races and robbing local governments of tax revenues, which in turn makes it harder for them to confront not only violence but virtually every other challenge they face. For communities as well as for individuals, exposure to criminal victimization, and in particular to homicide and serious assault, is a pillar of American inequality.¹¹

Much of the explanation for the racial disparities in rates of criminal victimization in the United States lies outside the criminal justice system: in the pervasive, interlocking disadvantages imposed on people of color, especially on African Americans. Discrimination perpetuates poverty, and poverty breeds crime, making potential offenders more desperate and potential victims more vulnerable.¹² But part of the explanation is inadequate policing, and more precisely the long history of police departments protecting white, wealthy neighborhoods more than poor neighborhoods disproportionately populated by people of color.¹³

If poor people and people of color in the United States have long suffered from inadequate protection against crime, they have also suffered from an excess of violence and abuse at the hands of the police. Police officers kill roughly a thousand Americans every year. Somewhere between half and eighty percent of the deaths, probably, are unjustified.¹⁴ And the victims of police killings are disproportionately people of color, with young Black people men especially at risk. Black Americans are fourteen percent of the population but more than a quarter of the people shot dead by the police.¹⁵ From 1980 through 2019, on an age-adjusted basis, Black people were more than three times as likely to be killed by the police as whites; Latinos were close to twice as likely.¹⁶ Between 2015 and 2019, an unarmed Black man was four times as likely to be fatally shot by the police as an unarmed white man.¹⁷ For Black and white males between the ages of twenty and twenty-four, the ratio was five to one.¹⁸

Deaths at the hands of law enforcement officers are the most extreme way, but far from the only way, in which the burdens of policing fall disproportionately on African Americans and other people of color. People of color, especially young Black men, are more likely to be stopped by the police. When they are stopped, they are less likely to be treated with respect, more likely to grabbed or struck, more likely to be searched, and more likely to be arrested.¹⁹

The harsh, often brutal treatment of African Americans and other people of color by police has ramifications far beyond the deaths, physical injuries, and indignities it inflicts. Stops and arrests are entry points into the carceral system. Excessively aggressive, discriminatory policing helps to sustain jail and prison populations that are bloated and racially lopsided. And the interactions that people have with the police reverberate through their communities, with lasting effects not just on attitudes toward law enforcement but on broader ideas about law, government, and society.²⁰ Unsurprisingly, African Americans consistently report less confidence in the police than whites.²¹ But mistreatment by the police often leads, also, to an enervating sense of disempowerment—a sense of physical vulnerability, lack of belonging, and alienation—not just in the immediate victim of the mistreatment but in friends, family, and neighbors as well.²²

The damage that American policing does to people of color and their communities has received more attention over the past three decades for several reasons. Part of the explanation is changes in law enforcement: the expansion of police forces since the 1980s; more aggressive use of stop-and-frisk; crackdowns on low-level, quality-of-life offenses; and the spread of militarized equipment and tactics, including through the proliferation and increased use of SWAT teams.²³ Increased public awareness of police violence has also played a role; the key contributors here have been the Black Lives Matter movement and the advent of smartphones and social media.²⁴ But some part of the reason that police violence and its disproportionate use against people of color has loomed larger may also be a success to which law enforcement agencies themselves contributed: the dramatic, transformational decline of crime in the 1990s. As threats of private violence became less omnipresent in poor neighborhoods of color, threats of police violence—which had always been there, in the background—became more jarringly inexcusable.²⁵

As devastating a toll as crime now takes in the United States, it did far more damage thirty years ago. Between the early 1990s and the turn of the millennium, the national homicide rate dropped by roughly forty percent, and the decline was even larger in the neighborhoods and demographic groups hardest hit by crime. The rates of other crimes saw similar drops.²⁶ The sociologist Patrick Sharkey notes that for Black men, the homicide drop was the largest public health achievement of the past several decades, shrinking the racial disparity in life expectancy and preserving roughly one thousand years of life for every one hundred thousand Black men. Sharkey has also documented the ways in which the decline of crime changed the fabric of life in poor neighborhoods, allowing public spaces to be reclaimed and alleviating the constant, debilitating fear of violent attack. Combined with changes in law enforcement, the crime decline of the 1990s altered the nature of the physical insecurity experienced in poor communities of color, particularly by young people, “from the threat of violent peers to the threat of abusive police.”²⁷

The plummeting crime rates of the 1990s were followed by more modest reductions in homicides and aggravated assaults in the early years of the twenty-first century. Homicide rates began to rise, though, around 2014, and then surged in cities across the country during the COVID-19 pandemic of 2020 and 2021.²⁸ Some major American cities recorded more homicides in 2021 than in any prior year.²⁹ And just as the crime decline in the 1990s was particularly pronounced in poor neighborhoods and predominantly Black neighborhoods, fatal shootings have risen most dramatically in recent years in those same neighborhoods.³⁰ In Los Angeles, for example, Black Americans are nine percent of the population but constituted thirty-six percent of homicide victims in 2021; in New York City, the figures are twenty-four percent and sixty-five percent, respectively.³¹ Rates of other violent crimes do not appear to have risen as much as homicides, and the nationwide homicide rate in 2020 and 2021 remained well below its peak in the 1980s. Still, a significant amount of the progress made in reducing fatal attacks in the 1990s and early 2000s seems to have slipped away, at least temporarily and possibly for longer. Just as there was nothing unavoidable about the high crime rates of the 1980s, there is no guarantee those rates will not return.

The causes of the crime drop in the 1990s are still debated, and so are the explanations for the rising homicide rates of the past several years. Some of the credit for the crime drop, though, almost certainly should go to improvements in policing: either to the expansion of police forces in the 1990s, or to changes in how the police operated, or most likely to both factors. The evidence is threefold. First, a growing body of research links increased police presence to decreases in crime, especially homicides. Some of this research examines the effects of changes in the size of local police forces; others look at the effects of temporary surges in police presence because of, for example, terrorist alerts.³² Second, the crime drop during the 1990s was roughly twice as large in New York City as elsewhere in the country, and the most plausible explanation for the difference is the especially large changes in the quantity of and quality of policing in New York City during that period.³³ Third, there is strong evidence for the effectiveness of particular police strategies that became more widespread in the 1990s, especially tactics that focus on areas where a large number of crimes take place and other examples of “problem-oriented policing.”³⁴

American law enforcement didn’t just get better at controlling crime in the last decades of the twentieth century. It also improved in other ways, albeit unevenly, and too often transiently. At the beginning of the 1970s, for example, police departments in the United States were overwhelmingly white and overwhelmingly male. Many departments, particularly in big cities, grew more diverse in the 1980s and 1990s, often through hiring plans adopted in response to lawsuits. By the early 2000s, some large police forces were majority minority—this was true, for example, in Los Angeles, Detroit, and Washington, D.C.—and the percentage of female officers had grown as well. Smaller departments made less progress, though. Moreover, as court-ordered hiring plans have expired over the past twenty years, diversification has stalled even in larger departments, and some past gains have been undone.³⁵

Diversifying law enforcement agencies is not a panacea—there are no panaceas in police reform—but accumulating research suggests that minority and female officers are less likely to use unjustified force, especially against people of color.³⁶ White male officers partnered with minority or female officers also change their patterns of policing for the better. And diverse departments are less insular, more open to outside ideas, and better connected to the communities they serve, all of which makes them more likely to adopt other reforms.³⁷

One particular way in which police diversity has facilitated other reforms is by countering the strident hostility of police unions toward efforts to reduce police violence, increase police accountability, and combat racial biases in law enforcement. Police unions are not always implacable opponents of reform, but even today they do more to obstruct than to champion measures for making law enforcement fairer, more effective, and less lethal.³⁸ One reason for that is that the leadership of police unions remains considerably older and whiter than police officers overall. Organizations of Black and Latino officers have often championed reforms opposed by police unions, and the presence of officers of color may in some cases have led police unions to moderate their own positions.³⁹

The most important of those reforms over the past half century have been community policing and problem-oriented policing, both of which spread widely in the 1980s and 1990s, sometimes as actual programs and sometimes, unfortunately, just as slogans.⁴⁰ Virtually every police department in the country eventually claimed to practice “community policing,” in part because it became a routine requirement for federal grants. At its best, though, community policing was more than a slogan. It was a comprehensive reorientation of law enforcement away from a go-it-alone “warrior model” and toward a collaborative “guardian model” that relies on consultation and cooperation with the public and with other government agencies.⁴¹

Community policing had major weaknesses, some of which will be discussed below. And because community policing was implemented with varying levels of seriousness, it was hard to rigorously evaluate. It spread, though, because it often greatly increased public satisfaction with the police and made people feel safer. Decreased fear of crime itself probably led to real reductions in crime: when people felt safer they ventured out more, and streets and parks tended to become safer with more people around.⁴²

When implemented most fully, community policing also decentralized control within police departments, and it broadened officers’ focus beyond crime suppression, allowing them to address a range of other concerns raised by the communities they served. It therefore fit well with problem-oriented policing, which called on officers to work creatively, and on an ongoing basis, with other agencies and the public to address issues of particular local concern. Sometimes, but not always, problem-oriented policing involved reallocating resources to “hot spots” responsible for a disproportionate amount of crime; sometimes, but not always, the resources reallocated were patrol officers. Unlike community policing, problem-oriented policing—especially the focus on hot spots—lent itself to statistical evaluation, and a steadily increasing body of evidence credits these techniques with significant crime reductions.⁴³ Furthermore, there are indications that hot spots and other forms of problem-oriented policing have helped some cities defy the national trend and reduce homicides in 2021.⁴⁴

Beyond hot spots, community policing and problem-oriented policing frequently also involved increased attention to low-level, “quality of life” offenses, such as vandalism, loitering, and disturbing the peace. Part of the idea was that when these kinds of violations were left unaddressed, neighborhoods spiraled toward greater disorder, fewer people on the streets, and higher rates of serious crime; this was the famous, later infamous, theory of “broken windows.” Police initiatives focused on quality-of-life offenses could be empirically evaluated, too, and the results suggested that these programs were in fact effective at reducing serious crime, but not as dramatically as hot-spots policing and only when the programs involved community collaboration and carefully targeted particular problems in particular places. Aggressive, indiscriminate crackdowns on quality-of-life offenses, although frequently billed as community policing and problem-oriented policing, did not reduce crime.⁴⁵

At the other end of the spectrum from quality-of-life policing, problem-oriented policing sometimes took the form of programs in which the police worked with community groups and other government agencies to address particular groups of people responsible for a disproportionate share of a city’s gun violence. These programs—the first and most famous of which was Boston’s Ceasefire initiative in the 1980s and 1990s—were sometimes called “focused deterrence” because their most prominent component was often threats of heavy penal consequences targeted at the individuals and groups driving violence in a particular area. But the programs also included offers of social support to the same people, and some, more recent versions of this approach emphasize peer-to-peer counseling more than policing. Evidence is growing that these programs, when done right, significantly reduced gun violence.⁴⁶

At their best, moreover, community policing and problem-oriented policing provided their greatest benefits in poor neighborhoods. These programs were redistributive, and not just because they were resource intensive, requiring lots of time from lots of officers in neighborhoods hit hardest by crime. Community policing and problem-oriented policing required police to adopt what the criminologist James Q. Wilson had called the “service style” of law enforcement—the kind of policing typically found in relatively well-off suburbs—and to move away from the types of law enforcement traditionally pursued in poorer areas—what Wilson had called the “legalistic style,” which prioritized arrests, and the “watchman style,” which emphasized order maintenance.⁴⁷ (Conversely, when community policing and problem-oriented policing were reduced to “zero tolerance” campaigns against quality-of life offenses, they doubled down on the watchman style.) In addition, community policing and order-maintenance policing frequently required officers to arrange for crime-ridden neighborhoods to receive a range of services other than law enforcement: trash cleanup, rodent abatement, streetlight repair, and so on. Officers might complain that they had not signed up to be social workers, but it turned out they often were good at getting other municipal agencies to pay more attention to marginalized neighborhoods.⁴⁸

For several different reasons, community policing and problem-oriented policing have lost much of their luster over the past two decades. The terrorist attacks of September 11, 2001, led to calls for more aggressive forms of law enforcement, and state and local budget shortfalls in the early 2000s led many departments to view community policing and problem-oriented policing as unaffordable luxuries. Degraded forms of these programs, like the zero-tolerance crackdowns on turnstile jumping and “squeegee men” in New York City, helped to sour community activists against them and, eventually, in many cases, against the whole project of police reform.⁴⁹ And it became increasingly clear that, even at their best, community policing and problem-oriented policing had some glaring flaws.

One was that these programs paid little attention to police violence. This was not inherent in the philosophy of either set of reforms. Departments could have collaborated with the public and agencies outside law enforcement to reduce police violence. But they rarely did. Police reformers deemphasized the problem of police violence, especially police killings, because they did not appreciate the gravity of the problem. This was partly because the victims were generally members of marginalized groups (poor, young, and Black, Latino, or Native American), partly because federal government did not (and still does not) collect reliable data on police killings or other forms of police violence, and partly because, until the advent of body cameras and smartphones, officers’ accounts of these episodes were difficult to challenge. It took those technological developments and the Black Lives Matter movement to give the issue of police killings the attention it had long deserved.⁵⁰

Nonetheless, lethal police violence against Black Americans and Latinos has declined significantly over the past half century. Controlling for the age of the victims, the risk of being killed by a police officer in the United States fell during the 1980s, and the drop was particularly dramatic—around fifty percent—for Black Americans and Latinos. The decline in police killings over the course of that decade appears to have been mainly attributable to new policies restricting the use of lethal force against fleeing suspects.⁵¹ In the following three decades, in contrast, the age-adjusted rate of police killings of Black Americans and Latinos remained roughly constant nationwide, while the rate for white Americans increased; as a result, the age-adjusted figure for Americans overall also rose somewhat. In some places, though, rates continued to drop. From 2013 through 2019, police killings rose in rural areas and suburbs but declined by thirty percent in the thirty largest American cities, probably because of new restrictions on officers’ use of deadly force.⁵²

Police violence in some cities has dropped especially sharply. In Los Angeles, for example, significant uses of force by the police appear to have been cut in half between 2006 and 2019, and police shootings appear to have declined by forty percent.⁵³ Shootings by LAPD officers increased alarmingly in 2021 but remained far lower than in past decades.⁵⁴ In Cincinnati, on the other hand, uses of force by the police have dropped by fifty percent over the past fifteen years, but the rate of police shootings has not shown a similar decline. Police shootings in Cincinnati rose sharply between 2006 and 2011, plummeted in 2012 and 2013, spiked again in 2014 and 2015, and then gradually dropped over the next several years, returning by 2019 to roughly the 2006 level.⁵⁵ Los Angeles and Cincinnati have been sites of major, sustained efforts at police reform over the past two decades, so the statistics regarding uses of force by police in these cities over the past decade are encouraging in some respects (significant reductions in uses of force in both cities and significant reductions in police shootings in Los Angeles) but disappointing in others (no overall progress on police shootings in Cincinnati).

The statistics also underscore the great variation in patterns of police violence, as in crime rates, across the United States. The Police Scorecard Project, led by the activist and data scientist Samuel Sinyangwe, compared the number of uses of deadly force per ten thousand arrests for hundreds of American police departments from 2016 to 2020: the results ranged widely, from fifteen down to zero.⁵⁶ If the United States as a whole had New York City’s rate of police killings, almost 600 fewer Americans would have been killed by the police in 2019.⁵⁷ (If America had New York City’s homicide rate, 4,400 fewer people would have died violently that same year.⁵⁸) Police reformers can succeed, and have succeeded, but the victories have been slow, partial, and often fleeting.

By the middle of the second decade of the twenty-first century, a small but growing number of activists and scholars were losing faith in the project of police reform and were calling for the police to be abolished altogether.⁵⁹ A milder version of this idea, “defund the police,” spread widely in 2020. The argument was straightforward. If officers victimize poor people and people of color and also fail to protect them from crime, why not replace police departments, partly or entirely, with something less harmful and more effective?

Like the details of “community policing,” the precise meaning of “defund the police” could be hard to pin down. Asked what a world with defunded police would look like, Representative Alexandria Ocasio-Cortez said in June 2020 that it would look “like a suburb.”⁶⁰ That suggested that defunding the police might involve not taking money away from law enforcement but instead bringing the service style of policing to poor and marginalized neighborhoods—much as community policing had aimed to do in the 1980s and 1990s. More commonly, though, advocates of police defunding sharply disassociated themselves from the agenda of community policing: their principal goal was not “better” policing but less policing. Activists argued that “[t]he more cops you have, the more encounters our communities have with law enforcement, the more violent encounters there are.”⁶¹ Ocasio-Cortez herself later criticized proposed budget cuts at the New York Police Department for not going far enough. “Defunding police means defunding police,” she explained.⁶²

It is hard to blame “defund the police” for the rising rates of homicide during the pandemic or the spikes some localities saw in property offenses because average police budgets were not substantially leaner at the end of 2020 than when the year began, and many cities restored funding to police in 2021.⁶³ That was partly because the slogan never was embraced by a majority of Americans, regardless of race or party, and support for it steadily declined as the protests of 2020 grew more distant, the pandemic dragged on, and worries about crime deepened. By the end of 2021, nearly half of American adults, including thirty-eight percent of the Black population and forty-six percent of the Latino population, wanted more spending on law enforcement.⁶⁴

The argument for police defunding deserves to be taken seriously. A significant minority of Americans, including more than forty percent of Black adults, a quarter of Latino adults, and many activists and scholars, continue to favor cutting police budgets.⁶⁵ And the checkered history of police reform gives ample reason to wonder whether, as activists suggest, it is time to try something different.⁶⁶

But defunding the police is not the answer, for three reasons. The first reason is political, the second is practical, and the third is conceptual.

To begin with the politics: it matters that most people oppose defunding, including and often especially in the nonwhite neighborhoods hit hardest by both crime and abusive forms of policing. And it matters even more when, as today, politics are highly polarized. Eric Adams, a former police captain, campaigned successfully for mayor of New York in 2021, opposing calls to defund the police and promising to reform the NYPD instead. His key support came from Black, Latino, and low-income voters in the outer boroughs.⁶⁷ In November 2021, a year and a half after George Floyd died brutally at the hands of Minneapolis police officers, voters in that city rejected a measure to replace the police department with a new Department of Public Safety. The measure failed in part because it was opposed by a majority of voters in the city’s largest Black neighborhoods.⁶⁸ In the fall of 2020, I helped supervise a group of law students at Stanford who met with members of the African American Mayors Association to identify responsibilities that might usefully be shifted away from the police. What surprised the students most was how opposed the mayors were to removing police officers from schools. (The students wound up recommending that step anyway, for reasons that will be discussed below.)⁶⁹

Remaking public safety takes time. This is particularly true of approaches like community policing and problem-oriented policing, which aim to turn police departments into learning organizations committed to continual improvement.⁷⁰ Progress in reforming the police has been frustratingly slow in part because so many reforms, including programs of community policing and problem-oriented policing, are abandoned when new police chiefs are appointed, political winds shift, crime rates rise, or budgets are tightened.⁷¹ Successful transformation of public safety systems requires reforms that can be sustained over the long haul.⁷² But the weak public support for police defunding and the shifting politics of crime mean that even when police budgets are trimmed, the cuts are likely to be reversed. This is in fact precisely what happened across the United States in 2021, as cities restored funding they had taken from police departments in 2020 and in many cases gave them additional money.⁷³ Police defunding is not, to put it mildly, a politically resilient strategy, especially in a period marked by extreme polarization.

Sometimes radical change is necessary despite its lack of political support. But the problems with police defunding go beyond its unpopularity. As a practical matter, voters are right to be skeptical that taking money from law enforcement will reduce police abuse, let alone improve public safety. Calls to defund the police draw inspiration from the movement to abolish prisons; in both cases, abolition is understood both as a utopian endpoint—a horizon to aim for—and a path for practical, incremental progress toward a better society.⁷⁴ This makes sense for prisons, because the grotesque hypertrophy of the American carceral system may be its signal vice. There are so many people in prison who plainly don’t need to be there that it would take years of aggressive decarceration before we got to difficult cases. For the foreseeable future, any decreases in the number of people the United States locks up will very clearly be changes for the better, reducing the harms of mass incarceration without endangering public safety.

But the benefits of incremental reductions in the scale of policing are less obvious. Smaller police departments, with fewer contacts with the public, are not necessarily less lethal, and larger, better-funded departments are not, for that reason, more abusive. Community policing, problem-oriented policing, and better systems of police oversight all cost money; this is particularly true of the first two, which require freeing up officers’ time. Cutting funding for police departments means making these reforms more difficult. It also can mean officers who are overworked, overstressed, and cut off from the people and neighborhoods they are supposed to be protecting, all of which can lead to more, rather than fewer, violent encounters. After the 2008 financial crisis forced the city of Vallejo, California, to reduce the size of its police force by close to forty percent, police killings skyrocketed.⁷⁵ In contrast, Camden, New Jersey, significantly reduced police violence through reforms that ultimately involved more officers, more contacts with the public, and a larger police budget.⁷⁶

Vallejo and Camden are only two cities, and in different ways their experiences are exceptional. But it is difficult to identify any city that has reduced police violence by shrinking its department, or any department that has been successfully reformed on the cheap. The suburbs that Representative Ocasio-Cortez called models of police defunding often spend more money and employ more officers per capita than large urban departments. The disparity is even greater when the large number of private security guards supplementing public police in many American suburbs is taken into account.⁷⁷

Advocates of police defunding often stress that they are calling not just for cuts in police budgets but for allocating the savings to schools and social services.⁷⁸ What happened in Vallejo, they would say, is not a fair test of their agenda. On the other hand, Camden—which disbanded its police department and created a new one, without a union contract—has sometimes been cited as a good example of what it truly means to “defund the police.”⁷⁹ But most advocates of police defunding make clear that they really do mean taking money away from law enforcement and reducing the number of contacts between police and the public.⁸⁰ The opposite happened in Camden.

Furthermore, when money is taken away from police, there is no guarantee where it will be spent or if it will be spent at all. Advocates of mental health deinstitutionalization in the 1970s called for closing mental hospitals and replacing them with community mental health services; they succeeded in closing mental hospitals.⁸¹ To quote the criminologist Cynthia Lum, who began her career as a patrol officer in East Baltimore in the late 1990s, America’s poorest neighborhoods suffer from “under-everything.” They need more resources for every kind of municipal service, including the police. And Lum doubts that other social service providers “will suddenly care more about these communities and do so without disparity if given the responsibilities previously delivered by the police.” That is wishful thinking, she says, contradicted by the experience of “any police officer who has tried to get social services to respond to an abandoned or neglected child at three in the morning.”⁸² In fact, some of the best documented successes of community policing came when officers helped neighborhoods secure help from other city agencies. It should not take a police officer to get more social services in an impoverished neighborhood—or better streetlights, or rodent abatement—but often that was precisely what it took.

The most fundamental problem with the strategy of police defunding, though, does not have to do with its weak political support or with concerns about its practical operation. The most fundamental problem is conceptual: a lack of clarity about what is to be defunded and what should replace it. Advocates of defunding often say that communities themselves can provide public safety better than the police. But what counts as the community, and what counts as the police? Do the police stop being “the police” if they pursue problem-solving or public-health approaches to community safety? Are they defined by their uniforms? Their guns? Their badges and legal powers? If wealthy communities replace public officers with private guards, does that count as defunding the police?

The sociologist Egon Bittner defined the police by their unique social role: their responsibility for responding in emergencies where force may need to be used because “something that should not be happening is happening and someone must do something about it now!” What is special about the police, he said, is their ability to threaten force, and their responsibility to do so, in situations that require it.⁸³ One reason this is a helpful way to define “the police” is that much of what is problematic about law enforcement—so many of the pathologies of police work—can be traced to officers’ ability to use or threaten force and to the circumstances in which, for that reason, they are called to respond.

Understood in this way—defined in terms of their distinctive social function—the police will be with us in one form or another for the foreseeable future. The key questions about the police are how they will be managed, how they will be held accountable, and what tasks they will be given beyond their core responsibility of responding to situations where force may be needed.

It is not much of an answer to say that resources should be invested in the community instead of in the police, or that the community rather than the police should keep neighborhoods safe. Communities act through groups, organizations, and governments—including police departments. Moreover, communities do not have clear, natural boundaries, and however they are defined, they contain people with varying points of view. Communities are not monolithic. What needs to be decided is not whether the community should be empowered, but how: what kinds of agencies should provide public safety, how they should be configured and controlled, and how conflicting values and expectations will be reconciled.⁸⁴

Advocates of police defunding who imagine communities to be organic, naturally defined, and homogeneous are repeating a key mistake of the backers of deinstitutionalization in the 1970s, who assumed that communities would come together to take care of people released from mental hospitals.⁸⁵ The community policing movement made a similar mistake; that is how it wound up blind to police violence. Backers of community policing often supposed that residents of a neighborhood would speak with one voice. Young or old, working or unemployed, conforming or rebellious, they would share common values. Because of this simplifying assumption, community policing programs often relied on business owners and older residents to express the sentiments of the community, and little effort was made, in particular, to solicit the views of young people.⁸⁶ That was a critical error, in part because it is young people who tend to be the victims of police violence.⁸⁷

Community policing would have been more successful and more “durable” (to borrow Patrick Sharkey’s term for policies that can be sustained over the long term)⁸⁸ had it paid more attention to the complexity of communities.⁸⁹ But instead of working to incorporate the concerns of all members of the public, community policing relied too often on the small, unrepresentative groups who showed up at neighborhood meetings. Those groups were unrepresentative in part because departments often worked hardest to involve community members whose views they found congenial; as a result, the voices elevated in community policing were often less critical of the police, and more supportive of aggressive law enforcement tactics, then the public as a whole.⁹⁰ All of this meant that community policing, like deinstitutionalization in the 1970s, was critically undermined by a certain naivety about actual communities.⁹¹ Arguments for defunding the police often have the same fault.

A key element of the extraordinarily punitive turn of American criminal justice in the 1980s and 1990s was the rejection of rehabilitation: the abandonment of any effort to make prisons places that would help people convicted of crimes instead of warehousing them and making them suffer. This reimagination of the prison, making pain and containment its defining objectives, was commonly justified by the claim that rehabilitation was a proven, unmitigated failure. Studies of rehabilitation programs, it was said, had exhaustively demonstrated that “nothing works.” That was an egregious misreading of the research, distorting even the article from which the phrase “nothing works” had been lifted. As the sociologist David Garland has written, the aggressive insistence that “nothing works” was less a careful assessment of the evidence than an “emotive overreaction” to the excessive faith that had earlier been placed in rehabilitation, a cathartic expression of “the righteous anger of the disillusioned.”⁹²

If police reformers feel disillusioned and angry today, it is hard to blame them. It is easy to understand why many scholars and activists now say community policing made things worse, not better, by legitimizing the police and expanding their footprint.⁹³ But police reform has had successes as well as failures. Deep, long-lasting change will require recognizing what has gone right as well as what has gone wrong.

The successes and failures of police reform over the past half century offer three lessons. The first is that successful reform requires broad public support. Transforming law enforcement takes time and persistence, usually over a decade or more, and reforms opposed by significant parts of the public are likely to be abandoned too quickly, typically after only a few years. The second lesson, connected to the first, is that even in the smallest communities, people have conflicting aspirations about policing and public safety; they often disagree about how to balance crime control and order maintenance with the interests of privacy, liberty, freedom from harassment, and protection against state violence. There is no single, unified “people,” no homogenous “community.” The third and final lesson grows out of the first two. Because police reform requires balancing contradictory values and mediating between competing groups, reformers need models. Big successes, when they come, are built on earlier, smaller successes.

New efforts to remake public safety should heed each of these lessons. Police reform stands the best chance of working when it is durable, pluralistic, and experience-based.

Defunding the police fails these tests. It isn’t durable, because it lacks broad public support or even support by a narrow majority. Most Americans, regardless of race, oppose defunding the police. Defunding the police isn’t a pluralistic agenda, partly because it is opposed by large portions of the American public—indeed, by a majority—and partly because it relies on a simplified view of communities as organic, well-defined, and homogenous, with clear and uniform ideas about law enforcement. And defunding the police isn’t experience-based: we don’t have models of where it has succeeded in making policing more effective, more equitable, or less harmful.

On the other hand, broad majorities of the American public do support reforming the police. This is partly because, across lines of race and party, Americans share some of the core beliefs motivating calls to defund the police: that law enforcement is frequently too violent, that officers are insufficiently accountable, and that many of the functions carried out by the police could be better carried out by other agencies or community organizations, working separately or in conjunction with the police.⁹⁴ These areas of consensus open up avenues of police reform that are politically feasible, have long-term resilience, respect the diversity of communities, and build on past successes.

A durable, pluralistic, and experience-based agenda for police reform should be centered around four elements: a revival and strengthening of community policing; an expansion of problem-oriented policing, with a particular focus on programs targeting serious violence; a renewed commitment to diversity in police hiring and promotion; and improved mechanisms of oversight and accountability.

The central idea of community policing—that law enforcement should consult with and work in partnership with communities to pursue the community’s goals for public safety rather than operate as freestanding, largely autonomous crime fighters—remains broadly popular with Americans of all races.⁹⁵ And for good reason: the successes of community policing in the 1980s and 1990s were real.

The older versions of community policing had critical weaknesses, though, that newer versions should avoid repeating. Most glaringly, community policing programs in the 1980s and 1990s typically failed to address the issue of police violence. This failure was connected to a larger deficiency: a narrow, simplistic understanding of “the community.” Reformers tended to assume that communities had unambiguous boundaries and clear, consistent ideas about public safety. Divides within communities were often ignored, or—worse—departments worked to elevate only the voices within the community that they valued. Discussions of community policing also tended to proceed as though there were three distinct groups of people in any given area: the “community,” the police, and the people from whom the community needed protection. The community generally wasn’t thought to need protection from the police themselves. And the people from whom the community did need protection—in particular, young people—generally weren’t understood to be part of the community. Community policing fell far short of its potential in part because it ignored the divides within communities, and because it tended to treat young people as threats, not as stakeholders and partners.⁹⁶

In a similar way, community policing programs often erred by writing police officers themselves out of the community. As originally conceived, community policing aimed in part to tap into the idealism and initiative of line officers. This was true to an even greater extent of problem-oriented policing. But as community policing turned from an idea into a movement and then into something of an orthodoxy, it became more and more top-down. Less effort was spent drawing on the experience of the rank and file and enlisting their enthusiasm. In this respect, community policing continued a long tradition in American law enforcement. Patrol officers are necessarily given broad discretion; there is no way to regulate the vast majority of the decisions they make on every shift. But police departments have rarely made serious efforts to draw on the experience of line officers in ways that are systematic and publicly accountable or to involve them in the ongoing enterprise of remaking law enforcement. Police unions, with the implacable opposition they have often shown to reform, have done little to bolster the case for empowering the rank and file. But across the country there are hundreds of thousands of officers who want to work collaboratively with the public in confronting both crime and police misconduct. Community policing, like the older forms of law enforcement it aimed to supplant, took too little advantage of those officers.⁹⁷

These weaknesses were not inherent, though, in the basic idea of community policing. There were versions of community policing that took seriously the diversity of aspirations within any given community, the disputed boundaries of the community, and the need to protect members of the community from police violence and other forms of official abuse. There were versions of community policing that involved structured discussions among people with conflicting ideas about how public safety should be pursued in a particular locality—about what tactics the police should use, how police resources should be allocated, and how officers should be held accountable.⁹⁸ There were departments that made conscious efforts to bring together parts of the public that disagreed about policing tactics, including those highly critical of the police.⁹⁹ There were opportunities, too rarely pursued, to strengthen community policing by treating young people as members of the community with distinct concerns and distinct contributions to offer in pursuing public safety. There were versions of community policing and problem-solving policing that drew heavily on the experience and enthusiasm of the rank and file, and there were initiatives to rein in police violence that did the same.

In Lowell, Massachusetts, for example, police brought together residents of a largely white neighborhood and a poorer, heavily Cambodian neighborhood to discuss which neighborhood should be the site of a new police precinct that both wanted; the result was an overwhelming consensus that the interests of the Cambodian neighborhood should prevail.¹⁰⁰ When residents of Fremont, California, complained to police about a low-income apartment complex they thought, erroneously, was a hot spot of drug dealing and sex work, police brought them into the complex to encounter the actual tenants.¹⁰¹ Some beat meetings at the heart of Chicago’s community policing program successfully employed structured discussions among residents with conflicting ideas about public safety, sometimes aided by professional facilitators; Chicago also used organizers to boost participation at these meetings by poor people and people of color.¹⁰² A community policing initiative in Riverside, California, involved officers reaching out to, and working to address concerns of, not just seniors and other groups they thought would welcome partnership with law enforcement but neighborhood activists who had long been harsh critics of the police.¹⁰³ More recently, the Community Safety Partnership in Los Angeles—which has successfully reduced violence and increased public satisfaction with the police at four public housing projects over the past decade—grew out of meetings at which police officers sought out, listened to, and apologized to residents who were distrustful and sharply critical of the police.¹⁰⁴ Police in Boston reached out to Black clergy to secure the participation of young people in an anti-violence initiative.¹⁰⁵ And the most successful programs of community policing tended to be the ones that drew significantly on the ideas and initiative of rank-and-file officers; that was also true of a promising program to reduce police violence that was developed, and prematurely abandoned, in Oakland in the 1970s.¹⁰⁶

Recommitting to community policing, with or without a broader and more sophisticated understanding of community, does mean rejecting one key objective of many advocates of police defunding: reducing how often officers interact with the public. Community policing can and should be reconfigured to address police violence, to respect the diversity of communities, and to broaden who is treated as belonging to the community. But extensive contact between the police and the public is central to the very idea of community policing; there is no other way for the police to consult with and partner with the community.

Community policing is fully consistent, though, with shifting some responsibilities away from the police to other agencies or community organizations. Indeed, part of the idea of community policing is that the police do not and should not have a monopoly on the provision of public safety, and key parts of that job can and should be performed by others. And—jumping ahead—part of the idea of problem-oriented policing was that some kinds of crime and disorder are best addressed by agencies and organizations outside law enforcement, alone or in coordination with the police, and even when the police are involved, arrests and criminal sanctions should be used only when necessary.¹⁰⁷

So, a revival of community policing and problem-oriented policing can and should include a reexamination of which functions, currently carried out by police officers, could be better performed by others.¹⁰⁸ Advocates of defunding the police have suggested shifting responsibilities for mental health response, traffic enforcement, and school security to agencies outside law enforcement, and those suggestions deserve careful consideration. Much of the work in these categories does not require officers authorized to employ “situationally justified force”¹⁰⁹—the core, distinguishing feature of the police. Others may be able to perform these functions more skillfully and with less potential for violence. Roughly a quarter of the people killed by the police show signs of mental illness, for example.¹¹⁰ A growing number of cities dispatch medical and behavioral health professionals, rather than police officers, in response to 911 calls involving people in mental health distress, and the evidence suggests that these programs defuse situations that could otherwise escalate into violence.¹¹¹ Perhaps ten percent of police shootings originate in traffic stops, and these stops also serve as principal point of entry into the criminal justice system for many people of color. Many of these stops could safely be carried out by unarmed, non-police personnel or replaced with violation notices sent in the mail.¹¹² Similarly, police officers in schools are a key element of the “school-to-prison pipeline” that draws many children of color into the criminal justice system, and there is little evidence that replacing them with educational personnel would make schools less safe.¹¹³

Shifting responsibilities away from law enforcement is different than defunding police departments, let alone abolishing them. For one thing, most of what the police do falls outside the categories of mental health response, traffic enforcement, and school security. Estimates of the percentage of 911 calls that involve people in mental health distress range from roughly one percent to around twelve percent; the true figure is likely somewhere in the middle of that range.¹¹⁴ Traffic enforcement takes up a larger share of officers’ time, but probably under twenty percent.¹¹⁵ In 2016 there were twenty-seven thousand police officers assigned to schools; that was roughly four percent of the six hundred fifty thousand police officers in the United States.¹¹⁶ Almost certainly there are other jobs that could and should be transferred away from the police, but many, maybe most, calls to the police are calls for someone with the defining attribute of the police, the authority to use force. And when other agencies or organizations take on responsibilities for responding to mental health emergencies, enforcing traffic laws, or keeping schools safe, they will sometimes need to call the police for backup—albeit less commonly than might be expected.¹¹⁷ It is possible, moreover, that the best results will be obtained by creating teams in which police officers work alongside paramedics and mental health professionals; this approach has been adopted in Dallas, Texas, with promising results.¹¹⁸

Shifting responsibilities from the police, or leveraging police resources by having the police work alongside other professionals, will often free up funds that can be shifted, too. It is hard to say how much of a constraint law enforcement budgets place on the ability of local governments to fund other services. Nationwide, about one percent of state expenditures and six percent of local government expenditures go to the police, but per capita spending on law enforcement varies widely from place to place.¹¹⁹ Poor neighborhoods generally will benefit most if new services are provided without cuts to the police budget. But even when the only way to fund new services is to shift expenditures away from law enforcement, the focus should be on providing the new services, not on cutting police budgets. Otherwise, police defunding could wind up looking like mental health deinstitutionalization: program cuts without the promised reinvestments.

Community policing, done right, can make police more responsive and accountable to the public, it can make people feel safer, and it can help groups with conflicting goals and concerns work together on issues of public safety. It is less clear whether it can decrease crime, although there is reason to expect that when people feel more secure, they will spend more time on the street and in other public spaces, which in turn will make those places safer. There is better evidence for the effectiveness of problem-oriented policing in reducing crime, particularly certain forms of problem-oriented policing, when they are integrated with strong programs of community policing. The types of problem-oriented policing with the strongest records of success are programs that concentrate law enforcement resources on specific, crime-plagued locations (“hot spots”) and programs that focus on particular groups of people responsible for a disproportionate share of a city’s gun violence.

The most famous program of the latter variety was Boston’s Ceasefire initiative, which dramatically reduced youth homicides by interrupting cycles of retaliatory gang violence. Ceasefire identified a relatively small number of groups responsible for the bulk of youth shootings in Boston and targeted their members with threats of criminal enforcement along with offers of economic support and social services if they refrained from gun violence. The program relied on consultation and coordination between the police department, a range of other municipal agencies and nonprofit groups, and inner-city clergy. Ceasefire cut youth homicides in Boston by roughly fifty percent. Those gains began to disappear in the early 2000s when the program was discontinued, then were recovered when the program was restarted.¹²⁰ A variation on the Ceasefire approach, implemented in Oakland, California, reduced both homicides and nonfatal shootings by roughly half. The Oakland program focused on adult shooters rather than juveniles, reflecting differences between homicide patterns in Oakland and Boston; it also reduced (but did not eliminate) the role of the police and expanded the role of peer-to-peer counseling.¹²¹

Ceasefire programs are not cure-alls: both Boston and Oakland have seen increases in gun violence during the pandemic. More importantly, the experience in both cities demonstrates the importance of developing programs in close consultation with community groups, members of the public, and other municipal and county agencies, in part to tailor the program to local circumstances and local concerns. This process often requires several iterations of the program and can take several years. Moreover, even successful programs, like Boston’s, are difficult to sustain over the long term without broad public support. For all these reasons, Ceasefire initiatives work best, and are most likely to have a lasting impact, when they are paired with strong programs of community policing. That is also true of problem-oriented policing more generally.¹²²

Programs like hot-spots policing and Ceasefire initiatives can themselves become abusive forms of law enforcement, even—or maybe especially—when they enjoy broad public support. The theory of hot-spots policing calls for patrol officers to be concentrated in high-crime areas, but it leaves open the question how they should patrol: in particular, how aggressively they should employ their arrest powers or their authority to conduct investigatory stops and frisks. The simplest way to answer those questions, though, is with lots of stops, lots of frisks, and lots of arrests, and unfortunately this is sometimes what hot-spots policing has become.¹²³ Ceasefire initiatives, for their part, can wind up looking like systems of arbitrary and draconian punishment for people the police think, but cannot prove, are responsible for neighborhood violence.¹²⁴ The best versions of hot spots and Ceasefire have minimized these risks of abuse, but the risks are real, and cities employing these strategies need to keep them constantly in mind.

Community policing and problem-oriented policing both rely on community engagement, and that engagement is easier to obtain when police forces reflect the diversity of the communities that they serve. Police diversity is important for other reasons, as well. A growing body of evidence suggests that female officers use less force than male officers and that officers of color use less force than white officers, especially against African Americans.¹²⁵ Male officers partnered with female officers, and white officers partnered with officers of color, are also likely to use less force.¹²⁶ And diverse police forces are less insular and parochial. They are more open to outside ideas—including ideas that grow out of efforts at community policing and problem-oriented policing—and they have more-vibrant internal discussions. Furthermore, they are more likely to have strong, vocal organizations of Black officers, Latino officers, or female officers. Those groups, in turn, can serve as bridges to identity-based groups in the communities the officers patrol, and they are often effective champions of reforms that are opposed by police unions and other workplace organizations dominated by white, male officers.¹²⁷

The ranks of American police forces grew more diverse in the closing three decades of the twentieth century, but unfortunately that progress, which was far from complete,¹²⁸ has since stalled and in many cases is beginning to be reversed. Much of the progress was the result of affirmative action plans required by court orders or adopted under the threat of litigation; many of those plans have since been abandoned as the law has grown less supportive of race- and gender-conscious remedies for past discrimination.¹²⁹ Partly for that reason, American police forces on the whole look less like their surrounding communities today than they did twenty years ago. Federal data indicate, for example, that two-thirds of police departments employing a hundred or more officers grew whiter relative to their surrounding communities between 2007 and 2016.¹³⁰

The problem is particularly acute with regard to Black officers. Black officers hired in the 1970s, 1980s, and 1990s are now retiring, and new Black officers are not joining in anywhere near the numbers that would be required to make up for the retirements. The NYPD, for example, has fourteen percent fewer Black officers today than in 2008. In Philadelphia there are nineteen percent fewer Black officers than in 2017. The drop in Chicago has been twelve percent since 2019.¹³¹

Improving police diversity is often something of an orphan cause, opposed by people on the right who oppose affirmative action in general and viewed warily by people on the left who fear that tokenism will substitute for real reform. But community policing, problem-oriented policing, or any other effort at transforming law enforcement is unlikely to succeed, especially in the long term, unless it is coupled with a renewed commitment to diversifying the demographics of American law enforcement.

Community policing and problem-oriented policing lost credibility in part by worrying too little about police abuse, and especially about police violence. For new programs of community policing and problem-oriented policing to succeed over the long term, they need to be durable, and they will not be durable unless they are accompanied by effective strategies for addressing violent, discriminatory, and otherwise abusive forms of policing. Diversifying police departments will help in this regard but is not sufficient. Policing needs effective mechanisms of oversight and accountability.

The most effective mechanisms will be adapted to local circumstances, and they should be continually improved, capitalizing on successes and correcting for failure. The “problems” addressed by problem-oriented policing and community policing can and should include police violence and other forms of abusive law enforcement. But decades of experience have already helped to identify a range of best practices for controlling abusive policing and making police more accountable. Among them are the following:

(1) Strong civilian oversight. Most large police departments in the United States now have some form of institutionalized public oversight, whether from a commission, an independent auditor or inspector general, or both. But many smaller departments still lack anything along these lines. Community policing, which involves departments consulting with and partnering with the public, is not a substitute for systems that place law enforcement under the formal supervision of civilian officials.¹³²

(2) Meaningful judicial review. Courts, by themselves, are notoriously weak tools for guarding against police abuse.¹³³ But transforming public safety requires the availability of judicial forums for pressing claims of police misconduct, not least to make clear that the police, like everyone else, are not above the law. Doctrines of qualified immunity that can make it practically impossible to sue the police also reinforce sensations of disempowerment and habits of disengagement that eviscerate efforts at meaningful reform.¹³⁴

(3) Transparent systems of internal discipline. Even without qualified immunity, most instances of police misconduct will not wind up in court. This is one reason that law enforcement agencies need their own, robust systems for detecting and remedying official abuse. And it is difficult for these systems to be strong if they are opaque. Experience has shown that civilians involved in police discipline are no tougher than command staff on rank-and-file officers—if anything, quite the opposite—but over the long term the involvement of civilians helps to keep the system honest.¹³⁵

(4) Independent investigation of police killings. While it is important that police departments maintain their own, robust systems of internal discipline, public confidence requires that the most serious cases—when a police officer kills a civilian—be handled, to the extent possible, by investigators and prosecutors independent of the officer’s own agency and the prosecutorial office that works regularly with that agency.¹³⁶

(5) Tracking of problem officers. Abusive policing is not just a problem of “bad apples”; systems and occupational cultures are to blame, as well.¹³⁷ Still, a disproportionate share of police violence and other forms of official misconduct can be traced to officers who break the rules repeatedly. Just as programs of community violence reduction have often focused, successfully, on the small number of individuals responsible for an outsize share of violent crime, tackling police abuse is partly a matter of identifying and dealing with problem officers. This requires reliable systems for tracking the officers within a department with unusually high numbers of civilian complaints, disciplinary infractions, or episodes of violence, but it also requires ways to track these officers between departments, so they do not simply move from agency to agency, leaving their disciplinary records and risk-assessment flags behind them.¹³⁸

(6) Mandatory use of dash cameras and body cameras. Dash cameras and body cameras, like every tool of police reform, are not cure-alls. Early expectations were that officers would be less abusive if they knew their actions were being recorded, but the evidence on that so far is equivocal.¹³⁹ Moreover, there are serious, unresolved questions about the best rules for handling the recordings these cameras generate and under what circumstances officers should be allowed to turn them off.¹⁴⁰ But dash cameras and body cameras can powerfully counteract one of the greatest obstacles, historically, to holding officers accountable: the dearth of reliable, objective evidence about what happened in interactions between citizens and the police.¹⁴¹

(7) Bright-line rules. Even the most robust systems of after-the-fact review, whether in the form of civil lawsuits, criminal prosecutions, or internal disciplinary proceedings, will have limited effectiveness if the guidelines under which police officers operate are vague. In particular, it can be difficult for judges, jurors, and disciplinary boards to second-guess an officer’s decision about what constitutes “reasonable” use of force in a certain situation. So it is important to supplement vague standards of this kind with more specific, per se bans on practices—such as choke holds and no-knock drug raids—that experience has shown are deadly and unjustifiable.¹⁴²

None of these mechanisms of police accountability, nor any of the other reforms sketched in this article, can succeed in isolation. Decades of experience have shown that innovative approaches to public safety, no matter how promising they seem for reducing crime, controlling police abuse, or strengthening communities, are unlikely to last longer than a few years, even in a political environment less polarized than today’s, unless the reforms enjoy broad public support. And that kind of support materializes only when a comprehensive package of reforms—drawing on the lessons of community policing and problem-oriented policing, embracing a commitment to police diversity, and including robust systems of oversight and accountability—is hammered out and continually refined by the police working in genuine collaboration with a broad range of government agencies, community organizations, and advocacy groups. Cities that have sustained police reform over the long haul—Los Angeles, Cincinnati, New York—all have well-established systems of civilian oversight, and all have benefited from reform processes pushed along by judicial orders and monitoring. Los Angeles, Cincinnati, and New York all have seen homicide rates spike over the past two years, and all three cities—especially Cincinnati and Los Angeles—continue to have unacceptably high rates of police violence. But each of these cities has better policing today than it did thirty, twenty, or even ten years ago.

Most of the chief elements of effective police reform enjoy broad, bipartisan support, and most are also strongly endorsed by people of color, especially African Americans, who are the most burdened by crime and by police abuse. This is true of community policing and problem-solving policing, and it is true of the leading proposals for increasing accountability and oversight of the police—including mandating body cameras, requiring independent investigations of police killings, banning choke holds and no-knock warrants, and facilitating civil liability for police illegality by stripping officers of qualified immunity—and it is true of suggestions to shift certain responsibilities, like mental health response, away from the police.¹⁴³ Evidence is weaker regarding the level of public support for efforts to expand the recruitment, hiring, and promotion of minority officers and female officers, but those programs are unlikely to elicit strong opposition as long as they avoid the use of quotas or other rigid preferences.¹⁴⁴ There is therefore reason to hope that a comprehensive program of police reform can be fashioned that will cut crime, rein in police violence, reduce some of the nation’s worst forms of inequality, and—not least important—enjoy the kind of broad public support that will allow it to last, even in times as divided as ours.