How to Actually Fix America’s Police

Elected officials need to do more than throw good reform dollars at bad agencies.

An illustration of a police badge with "Freedom" and "Equality"
The Atlantic

George Floyd’s death is the latest in a long series of brutal encounters between the police and the people they are supposed to serve. Police abuse has targeted people of every race and class, but members of vulnerable populations and minority groups, particularly young black men, are especially at risk.

This is well known. The solutions are also well known. Prior tragedies have resulted in a string of independent, blue-ribbon commissions—Wickersham (1929), Kerner (1967), Knapp (1970), Overtown (1980), Christopher (1991), Kolts (1991), Mollen (1992), and the President’s Task Force on 21st Century Policing (2014)—to make recommendations for meaningful change that could address police misconduct. These groups have developed well-reasoned conclusions and pointed suggestions that are widely discussed and enthusiastically implemented—but only for a time. As public attention shifts, politics moves on and police-reform efforts wane. The cycle continues unbroken.

The problem America faces is not figuring out what to do. As an industry, American policing knows how to create systems that prevent, identify, and address abuses of power. It knows how to increase transparency. It knows how to provide police services in a constitutionally lawful and morally upright way. And across the country, most officers are well intentioned, receive good training, and work at agencies that have good policies on the books. But knowledge and good intentions are not nearly sufficient.

The hyperlocalized nature of policing in the United States is one factor here; the country has more than 18,000 police agencies, the majority of which (more than 15,000) are organized at the city or county level. Reforms tend to target single agencies. But it is not just the Minneapolis Police Department that needs reform; it is American policing as a whole.

What we desperately need, but have so far lacked, is political will. America needs to do more than throw good reform dollars at bad agencies. Elected officials at all levels—federal, state, and local—need to commit attention and public resources to changing the legal, administrative, and social frameworks that contribute to officer misconduct. As the University of Colorado law professor Ben Levin recently wrote, “Feigned powerlessness by lawmakers is common & frustrating. It reflects political cowardice or actual acquiescence in the violence of policing.” It’s time for that to change. Here is a blueprint for what they should do.

Federal Intervention

At the federal level, Congress should focus on three objectives.

The first is getting rid of qualified immunity. Qualified immunity is a judicial doctrine that protects officers who violate someone’s constitutional rights from civil-rights lawsuits unless the officers’ actions were clearly established as unconstitutional at the time. As the University of Chicago legal scholar William Baude has persuasively argued, the Supreme Court has provided multiple justifications for qualified immunity—including that it is the modern evolution of a common-law “good faith” defense, and that it ensures that government officials are not exposed to liability without “fair warning” that their actions are wrong—but neither the Court’s historical nor doctrinal justifications can bear the burden of scrutiny. Nevertheless, as the Court has described it, qualified immunity “provides ample protection to all but the plainly incompetent or those who knowingly violate the law.”

The problem is that the Court has taken an inappropriately narrow view of what it means for a constitutional violation to be “clearly established.” Essentially, a constitutional violation is clear only if a court in the relevant jurisdiction has previously concluded that very similar police conduct occurring under very similar circumstances was unconstitutional. The Supreme Court has, for example, applied qualified immunity in a case where an officer standing on an interstate overpass shot at a fleeing vehicle, something that not only contravenes best practices, but that the officer was not trained to do, a supervisor had explicitly instructed him not to do, and was unnecessary because officers under the overpass had set up stop strips and then taken appropriate cover. Nevertheless, because no court had previously reviewed such conduct and found it to be unconstitutional, the Court held that any violation was not clearly established and, thus, that the officer could not be sued for his actions. In another case, the Court held that qualified immunity protected officers who, contrary to their training, their agency’s policies, and long-standing police procedure, rushed into the room of a mentally ill woman who they knew had a knife and had threatened officers—but was no threat to herself—without bothering to wait for the backup officers they had already called. When the woman predictably threatened officers with the knife, something she would not have been able to do had they done what they were trained and expected to do, they shot her. Again, the Court found that because no court had yet explicitly held such conduct unlawful, a “reasonable officer could have believed that [such] conduct was justified.” This ridiculous standard means that qualified immunity does not protect all but the “plainly incompetent”; it protects even the plainly incompetent. And these are just two of many egregious examples.

As a judicially created doctrine, qualified immunity could be modified or eliminated by federal legislation. There is broad bipartisan support for doing so. The right-leaning commentator David French and the left-leaning UCLA law professor Joanna Schwartz have both made the case against qualified immunity. The American Civil Liberties Union, the NAACP Legal Defense Fund, the Cato Institute, and the Alliance Defending Freedom are among the groups that have filed amicus briefs or called publicly for the end of qualified immunity. The onetime Democratic presidential hopeful Julián Castro pledged to “end qualified immunity for police officers so we can hold them accountable,” and Representative Justin Amash, a former Tea Party Republican who is now a member of the Libertarian Party, recently introduced the Ending Qualified Immunity Act. With this scope of support, legislating the elimination of qualified immunity should be an easy first step.

A second thing Congress could do is pass legislation to further encourage better data collection about what police do and how they do it. For example, no one really knows how often American police use force, why force was used, whether it was justified, or under what circumstances it is effective. No one knows how many high-speed pursuits have been conducted or why they were initiated; how many fleeing drivers have been caught, or the number of collisions, injuries, or deaths that resulted. Only one state—Utah—requires agencies to report forcible entries and tactical-team deployments. Neither the police, nor anyone else, can tell us how many people have been injured when taken into custody, how many people have been arrested only to be later released without charges, or how many cases local prosecutors have refused to file for lack of evidence, constitutional violations, or police misconduct.

Moreover, no state or federal officials know how many publicly owned surveillance cameras police have deployed or privately owned cameras they can access, or where those resources are allocated. No state or federal officials know how many internal or citizen complaints of officer misconduct exist, whether people were dissuaded from making a complaint or their complaint was ignored or minimized, or the ultimate disposition of the complaint and whether the offending officer was disciplined. These data are not the administrative minutiae of policing; this is basic information about the everyday actions of government officials that is crucial to ensuring that such actions are properly regulated. Voluntary data sharing, such as the FBI’s current National Use-of-Force Data Collection efforts, is clearly insufficient. Congress gave the Department of Justice the power to require agencies to provide information about the use of force, but the DOJ has never exercised that authority. The federal government can require agency- and state-level data collection, coupled with a robust auditing system to ensure that accurate data are provided. This, too, is not a matter of partisan politics. Democrats tend to believe that policing suffers from systemic problems, the type that better data collection can help address, but that perspective is gaining support among Republicans, too. Tim Scott, a Republican senator from South Carolina, and Chuck Grassley, a Republican senator from Iowa, introduced the Walter Scott Notification Act, named after the man infamously shot in the back and killed by a North Charleston Police Department officer in 2017. Efforts like these are simply common sense.

The final thing the federal government should do is dedicate significantly more resources to supporting police training, local policy initiatives, and administrative reviews. Police agencies around the country regularly fail to meet what are generally recognized as minimum standards for use-of-force and arrest training, frontline supervision, and internal investigations. Some have a demonstrated pattern of violating the constitutional rights of their community members. Acting with legislated authority, the DOJ has intervened in a few of these agencies, mostly through consent decrees, assisted by an appointed monitor and enforced by a federal judge. While the DOJ cannot intervene in the actions of the more than 18,000 police agencies in the United States, Congress can instruct and empower it to offer technical assistance, identify conduct standards that can serve as references for courts in civil litigation, and provide a framework for responsive and democratically accountable community collaboration, opening additional avenues of reform. It cannot do any of that, however, if the presidential administration continues to seek to cut funding for such efforts. Congress could also better regulate the industry by requiring or encouraging clear, evidence-based conditions of accreditation, making them a prerequisite for federal funding and putting teeth into police-reform efforts by reducing or cutting off funding when agencies fail to meet those conditions.

State Intervention

State legislatures, which can often move much faster than the pace of national politics, have their own five objectives to focus on.

To begin with, 36 states have statutes that govern the use of both deadly and nondeadly force, while six states have statutes only for deadly force. More than three-quarters of the 58 total state statutes (some states have more than one) were adopted prior to or during the 1970s, and most have not been recently amended. In the absence of statutes, states regulate police use of force through judicial decisions. But even where state statutes do exist, the courts that interpret them unfortunately tend to rely on the Fourth Amendment law. This is a problem for two reasons. First, the Fourth Amendment regulates police seizures, but state law is supposed to regulate use of force, and not all uses of force count as seizures. (Several courts have held, for example, that an officer shooting at someone but instead striking a bystander does not constitute a seizure.) State law is supposed to be broader than the Fourth Amendment, which means that referring to Fourth Amendment doctrines in the interpretation of state law can provide less protection than state lawmakers intended. Second, and perhaps more important, those Fourth Amendment doctrines are a mess; they provide little meaningful guidance that officers in the field can use to determine when and how much force to use, and the guidance they provide to courts reviewing use of force is often flawed.

Worse, many of the state statutes and common-law doctrines are contrary to good practices. Some states allow officers to use force to make an arrest if they believe the arrest is lawful, even if it isn’t and their belief is unreasonable. Others are woefully outdated, and still provide a defense to officers who use deadly force to prevent the escape of a fleeing felon. And most states authorize officers to use “reasonably necessary” force, but do not bother to define what reasonable force is or explain how officers should determine that it is necessary. Very few states admonish officers to use appropriate tactics or punish officers for egregious mistakes that contribute to avoidable use of force.

States can do better. In the past several years, for example, both Washington State and California have amended their statutory regimes, giving officers the authority to use force in the situations that require it while also providing meaningful guidance to officers and courts about what those situations are. California allows officers to use deadly force against “imminent threats of death or serious bodily injury,” and says that an “imminent threat” exists when “a person has the present ability, opportunity, and apparent intent” to cause such harm. Definitions like this, which draw from best practices in policing, give officers the leeway to protect themselves and others while also prohibiting them from acting on unfounded or purely speculative fears.

State legislatures can also amend law-enforcement officers’ bills of rights and the laws that govern the collective-bargaining rights of police unions. Most states permit or encourage collective bargaining for police unions—even states that, like Wisconsin, otherwise take a dim view of public-sector unions. Police unions do some good work; research suggests that officers at unionized agencies are, on average, higher paid and more professional than officers at nonunionized agencies. However, unions have leveraged the collective-bargaining process to create labyrinthine procedural protections that can make it exceptionally difficult to investigate, discipline, or terminate officers. Some of the limits on investigation—such as delaying interviewing an officer after a critical incident for several “sleep cycles”—are based on faulty reasoning and have been thoroughly debunked by credible scientific research. Too often, discipline is precluded by unnecessary or inappropriate procedural violations; in some cities, for example, civilians can file a complaint only during a limited period after an incident, sometimes as short as 30 days. When officers are disciplined, that discipline is subject to grievance and arbitration procedures; at one agency, a study found that arbitrators “routinely cut in half” the severity of disciplinary sanctions imposed by agency management. Officers should have a right to appeal disciplinary findings, but only when they are arguing that the agency’s decision was arbitrary and capricious or that the agency did not act in good faith. By protecting bad officers, collective-bargaining agreements and state laws contribute to misconduct.

Further, state legislatures can do a better job of certifying and, when necessary, decertifying officers. Currently, most states require most officers to be certified by a standards-and-training commission. Such commissions set minimum training requirements, but state law can impose specific training that the state commission has, thus far, omitted from the academy curriculum. Washington State, for example, now requires both violence de-escalation training and mental-health training, and the commission must “consult with law enforcement agencies and community stakeholders” in developing that training. And while most states allow for decertification—which prevents someone who has engaged in misconduct from continuing to work in that state as an officer—that authority can be tightly limited. In some states, an officer can be decertified only after a criminal conviction for a felony or serious misdemeanor. Even in states that have more permissive decertification regimes, decertification is often used only sparingly. From the 1960s until 2017, only about 30,000 officers were decertified, and three states—Florida, Georgia, and North Carolina—make up about half of those. As the decertification expert Roger Goldman has said, that isn’t because those states have a higher proportion of bad officers; it is because those states “have very active decertification programs.” States have good reason to strengthen their commitment to policing the police: According to a recent study, officers who are hired by another police agency after being terminated or resigning in lieu of termination from a prior agency are more likely than other officers to engage in future misconduct.

A persistent culture of secrecy regarding personnel matters has not helped. Many states have sharply limited the public’s right to access officers’ disciplinary files or agency use-of-force investigations. Although there is, and must be, room for certain employee information to be kept confidential, an officer’s actions while dealing with members of the community and the steps that an agency takes to investigate those actions are clearly matters of public interest. The states that have passed broad sunshine laws, such as Florida, have taught us that public access can be a crucial component of police accountability without impeding proper police action. States that allow agencies to shred disciplinary records after a set period, sometimes as short as six months, are effectively making patterns of misconduct by problem employees significantly more difficult to detect. States should follow the lead of Florida and, more recently, California in passing public-records laws ensuring that disciplinary records and reports pertaining to critical incidents such as police shootings or other serious uses of force cannot be hidden.

Finally, states can rethink their approach to criminalization. “Overcriminalization” has been broadly discussed; there are so many laws that violations are ubiquitous. If everyone is a criminal, officers have almost unfettered discretion to pick and choose which laws to enforce and whom to stop, frisk, search, or arrest. And, as the saying goes, when all you have is a hammer, every problem looks like a nail. For too long, the hammer of criminal law has been used against a wide array of social ills. The result is police over-involvement in matters that would be far better left to other government institutions and social-service providers, including school discipline, poverty, homelessness, and substance abuse. The opioid crisis remains a stark reminder that the United States cannot arrest its way out of addiction. The troubling discrepancies between how police have been cast as soldiers in the War on Drugs—a war that, despite almost identical drug-use rates between white and black Americans, is fought mostly in poor and minority communities—and how police have been seen as an adjunct to the public-health authorities addressing opioid abuse in suburban middle- or upper-class neighborhoods should be a stark warning for state legislators to rethink the scope of criminal law.

Local Intervention

Local agencies, for their part, have much they can do. To get started, they should focus on five specific improvements.

Many agencies have accountability systems—so-called early-warning or early-intervention systems—that look great on paper but are neither followed nor audited. Since the 1980s, these systems have had the potential to identify officers before they engage in misconduct and allow supervisors to step in to prevent bad outcomes. Unfortunately, many agencies ignore their own protocols—the early-warning system becomes a meaningless administrative task—or supervisors assume that officers do not need any intervention unless they are flagged by the early-warning system. Neither error is acceptable, and both can be corrected.

The hyperlocalization of policing in the United States has resulted in many agencies either creating their own policies and training from scratch, often without the benefit of research or broad experience, or simply purchasing them from private vendors. Agency policies and training should do more to incorporate industry best practices and generally accepted principles. For example, a study of use-of-force policies at the 50 largest agencies in the country—agencies that have the time, resources, and depth of experience to get it right—shows that they are all over the board. Some merely repeat the constitutional standard laid out by the Supreme Court. Others add little more than an interpretation of constitutional law and an aspirational instruction to safeguard the sanctity of life. But it is entirely possible to adopt policies that touch on tactics and provide meaningful guidance for officers to follow in the field; we know because some agencies have done exactly that. Some have adopted policies that instruct officers to use the least amount of force that can be safely employed, and others have provided specific tactical guidance for officers making traffic stops, effecting arrests, or interacting with people who are mentally ill.

Another crucial objective is that officers must also be trained—meaningfully so—on their agency’s policies. The “read and sign” approach is an unfortunate reality; officers are expected to acknowledge that they have received new policies, but many agencies do nothing to ensure that they understand those policies. Sometimes agencies attempt to use technology to increase efficiency by, for example, having a command staff member read a policy aloud, posting the video online, having officers click the video link, and calling that “training.” It is no surprise when such “training” is ignored. We have all read too many depositions in which officers testify that they were not familiar with the content of an essential policy. Policy manuals are too lengthy for anyone to realistically expect officers to memorize the whole thing—an entirely separate issue—but when it comes to use of force, emergency driving, and a few other areas of low-frequency, high-risk activities, a more robust effort is required.

Of course, the best policies and training in the world will not mean a thing if they are not enforced. When the Phoenix Police Department adopted a body-worn camera system, for example, it had a broad mandatory recording policy that required officers to activate their cameras for almost all civilian interactions. A month after deployment, officers were recording only 42.2 percent of the incidents they were supposed to record; a year later, that number fell to 13.2 percent. The agency had the equipment and the policy; what it lacked was adequate supervision. Police reform lives or dies with first-line supervisors, and agencies need to ensure that corporals, sergeants, and lieutenants are doing the jobs they are paid to do. This, of course, requires agencies to train supervisors—a great officer does not always make a good supervisor—and to audit their decisions. In the same vein, agencies should invite external oversight. As government institutions in a democracy, police agencies must be responsive to community concerns, especially in the context of high-risk activities like the use of force and emergency-vehicle operations.

Police agencies also need to be much more transparent in the aftermath of high-profile incidents. Although certain information, such as body-worn-camera footage, may need to be withheld for a certain period to avoid contaminating crucial witness interviews, there is no legitimate justification for denying public access for months or years. The perception that police agencies are hiding embarrassing or inculpatory information is particularly destructive when agencies have readily shared video of interactions that reflect positively on the agency; nothing destroys public trust faster than a perceived double standard. As the authors of a book about police homicides, all officers or former officers, wrote: “Law enforcement agencies simply must find better ways to release more data, and to release it earlier.” At a minimum, agencies can adopt policies that presumptively mandate the release of video or other information a set amount of time after an incident, as the Los Angeles Police Department has done with its 45-day commitment. Many members of the public may see that as too long, and perhaps it is, but having a certain date will help prevent perceptions of a police cover-up.

Perhaps most important, agencies need to create a culture that understands and values the importance of peer support and intervention. Officers, like everyone else, behave the way they think their colleagues and co-workers expect them to behave. Few things are more important to weeding out misconduct and creating a professional culture than peers sending the message that misbehavior is simply not acceptable. Agencies must put professionalism, including peer intervention, at the center of police culture. Following the example of the New Orleans Police Department’s Ethical Policing Is Courageous (EPIC) program is an obvious first step that can protect officers and the public alike.

Meaningful police reform is possible, but it will require a coordinated effort from federal, state, and local government. It will require sustained pressure from the public to push elected officials to take action. This will not be straightforward, nor will it be fast. And as the protests of George Floyd’s tragic, predictable death continue, it should be patently obvious that the country has no patience for the same old apologetic and half-hearted attempts at reform that we have seen previously. George Floyd, and all of us, deserve better.