Police-Generated Killings: The Gap between Ethics and Law

The Priority of Protecting Life

The principle guiding my analysis is that, in their jobs, police have an ethical obligation to prioritize the protection of life. This idea enjoys broad acceptance, including in law enforcement, as evident from use-of-force policies. The U.S. Department of Justice (1995) emphasizes “the integrity and paramount value of all human life.” Likewise, a consensus use-of-force policy by various U.S. law enforcement organizations states that its overarching goal is to “value and preserve human life” (Association of State Criminal Investigative Agencies et al. 2020, 2).

The idea that officers have an ethical obligation, tied to their professional role, to prioritize the protection of life aligns with common intuitions about police work. When an elusive suspect flees but poses no threat to life, it is wrong for police to use deadly force to stop them. Such force is a disproportionate response, since preserving the suspect’s life trumps the government’s interest in catching them. Similarly, if police must choose between catching a thief escaping with stolen goods or saving a bystander from a deadly trap planted by the thief, they should save the bystander. Although police have a responsibility to enforce laws against theft, the value of life exceeds any piece of property, and thus protecting life takes priority.

A commitment to protecting life also entails working to prevent serious bodily injury and rape, not just deadly threats. The irrevocable nature of these harms explains why stopping them takes priority over, say, preventing property loss. When a car is stolen, it can be replaced, but it is impossible to restore a life taken or limb lost. Perhaps some property, like rare art, can never be replaced. Still, most without hesitation favor saving a life over rare art because of the former’s incomparable value (Rodin 2002, 43–48). This core value informs normative expectations for a wide range of human activity, and police work is no exception.

Police sometimes encounter threats that preclude the possibility of protecting every life, which complicates their obligation to prioritize the protection of life. These situations force police to prioritize certain lives over others. Such dilemmas are always regrettable, but in some cases our moral intuitions are clear on what choice police should make. If a hostage taker is about to kill their hostage and only one option will stop the threat—shooting the hostage taker—a police sniper should take the shot. Saving the hostage is morally preferable to saving the hostage taker because of the latter’s unjust threat. When police prioritize which lives to save, the lives of those who unjustly threaten others receive lower priority. That point by no means implies that the lives of suspected criminals lack value. Outside zero-sum scenarios, police should strive to protect the lives of suspects and bring them into custody without harm.

As these examples suggest, the obligation to prioritize the protection of life has implications for police tactics. Here is the most obvious implication:

Obligation to Choose Nonlethal Tactics: Police have an ethical obligation to use nonlethal tactics, unless deadly force is necessary to stop an unjust threat to life.

This obligation provides officers with valuable guidance and prohibits a range of actions, like shooting a thief just because they are elusive and might get away.

But though valuable, such guidance is limited. Officers often have various nonlethal tactics available, and the above obligation offers little guidance on which to choose. Officers’ general obligation to prioritize the protection of life has implications for those decisions. It specifically requires police to favor nonlethal tactics known to reduce the risk of deadly force, since they are most consistent with the goal of protecting life. We can express this principle as follows:

Obligation to Reduce the Risk of Deadly Force: If there is no threat to life requiring deadly force and nonlethal tactic X is available to police, known to reduce the risk of deadly force compared to other tactics, and generally as effective in enforcing the law as other tactics, then police have an ethical obligation to choose X.

This principle shows how a commitment to protecting life influences earlier stages of police work, not just split-second decisions on whether to shoot. Police have an obligation to choose nonlethal tactics that avoid needlessly creating situations that make deadly force more likely. ²

Importantly, this obligation concerns reducing risk for all parties in an interaction. Tactics that raise the risk of deadly force for suspects often raise that same risk for officers. If a likely armed suspect is not an imminent threat and an officer confronts them at close range, such action makes escalation of force more likely and puts both in greater danger. There is little margin for error, as any false move or misperception can precipitate deadly force (Pickering and Klinger 2016, 27–28). In comparison, use of distance and cover by police during the interaction (when feasible) represents a Pareto improvement: it reduces the risk of deadly force for one or more parties without raising that risk for others. Such Pareto improvements are what the obligation to reduce the risk of deadly force most clearly requires.

This obligation places an additional but modest constraint on law enforcement. It only requires police to choose the less risky tactic when it is generally as effective in enforcing the law as other nonlethal options. That caveat is important. Without it, an obligation to reduce the risk of deadly force could imply that police can do little to enforce the law. Many law enforcement measures raise the risk of deadly force, even if just slightly. Consider traffic stops. Although only a small portion of traffic stops end in deadly force, they increase that risk. If police had to abandon this and all other tactics raising the risk of deadly force, no matter how small, they would lack many tools to enforce the law and be less effective in achieving that goal. Such a requirement would be overly restrictive, which the above obligation avoids.

Now police can have an obligation to choose a tactic generally as effective as a more risky option, but perhaps marginally less effective. Consider how police conduct searches and serve warrants. A common justification for no-knock raids is preventing the destruction of evidence, especially drugs (Richards v. Wisconsin 1997). For this reason, no-knock raids may be marginally more effective in enforcing the law than knocking, announcing, and waiting fifteen to thirty seconds before entering. Yet preventing the small amount of evidence that can be destroyed during a standard warrant’s short waiting period pales in comparison with avoiding the deadly risks inherent in no-knock raids (see next section for discussion of those risks).

Discontent with policing today partly stems from concerns—voiced by protest movements like Black Lives Matter (see Lebron 2017)—that officers enforce the law without appropriate attention to protecting life, particularly of marginalized groups. Black Americans are more than twice as likely to be killed by police (Fagan and Campbell 2020). Those killed are also more likely to be unarmed compared with White victims killed by police, suggesting that racial bias affects officers’ decisions to use deadly force (Nix et al. 2017), though social scientists still debate that question (Fagan and Campbell 2020, footnote 34). What is clear from research on stop and frisk, traffic stops, SWAT team deployments, and arrests is that Black Americans come in contact more frequently with police, which cannot be explained by differences in crime rates (Epp, Maynard-Moody, and Haider-Markel 2014; Gelman, Fagan, and Kiss 2007; Mummolo 2018a; Weaver, Papachristos, and Zanger-Tishler 2019). More intensive policing inflicts many harms on Black Americans (Alexander 2010; Butler 2017), including greater risk of death, since police interactions always carry some risk of escalating and turning deadly.

Given racial disparities in law enforcement, bad police tactics pose avoidable risks that fall disproportionately on Black Americans. Because bad police tactics exacerbate the risks to Black lives, there are compelling racial justice grounds to take steps to end these tactics. And even if police enforced the law free from racial disparities, there still would be compelling reason to rein in bad tactics, since they undermine the protection of life. If the protection of Black and other lives is truly a priority, bad police tactics deserve our attention.

The Problem of Police-Generated Killings

Some killings by police are lawful and morally justified, like when deadly force is necessary to stop an unjust threat to life. Other killings by police violate the law and their ethical obligations, like shooting an unarmed fleeing suspect. Here the focus is on a third category: lawful but morally wrong killings. These killings involve police actions that, though legal, run afoul of our ethical intuitions. Such killings take life unnecessarily and violate the obligation to prioritize the protection of life. They undermine police legitimacy since those charged with upholding the law escape legal sanctions for blameworthy acts that harm others.

What I call police-generated killings often fall into this category of lawful but morally wrong killings. As understood here, the term police-generated killing does not refer to just any killing where police play a causal role in the outcome. Rather, it refers to a narrower category: killings that result from bad tactics known to raise the risk of deadly force during a police interaction. Tactics that lead to police-generated killings deserve moral blame not because they fall short of perfection, but because they are obviously bad. By raising the risk of avoidable harm, such tactics are negligent or reckless. Those responsible for police-generated killings choose risky tactics despite having safer options to enforce the law.

Police-generated killings come in two varieties. The first involves intentional deadly force:

Police-Generated Killing Involving Intentional Deadly Force: A killing by police where bad tactics by the officer(s) involved create a situation where deadly force becomes necessary or perceived as necessary.

This category excludes accidental killings that stem from morally justified tactics. If an officer engages in sound tactics when using deadly force necessary to stop an imminent threat to life, their action carries a small risk to bystanders and in rare cases may cause accidental death. Though the officer causes the death, bad tactics do not. ³ My focus is instead on cases where bad tactics are to blame. In such cases, deadly force may appear justified if we just look at the moment it is used. But that perspective proves incomplete and misleading, for it fails to account for prior police actions that escalated force unnecessarily and created a situation requiring deadly force. Such poor tactical decisions violate police’s obligation to reduce the risk of deadly force.

A police-generated killing involving intentional deadly force is the shooting of Tamir Rice. Only twelve years old, Rice was playing in a Cleveland park with a pellet gun when someone called 911, worried the gun might be real while noting it was “probably fake.” In the first of several errors, the dispatcher failed to tell officers the gun might be a toy. Officers Frank Garmback and Timothy Loehmann rushed to the scene, with Garmback driving and pulling their car directly in front of Rice. Loehmann jumped out and, perceiving a movement by Rice as a threat, shot him within two seconds of exiting the vehicle (Dewan and Oppel 2015).

Pulling the car so close to Rice was a tactical blunder criticized by experts of police force. Even if there were reasonable grounds to justify deadly force at the moment of the shooting, officers made critical mistakes beforehand, which created a situation where deadly force was perceived as necessary. When a potentially armed suspect is not an imminent threat, as in this case, best practices are for officers to maintain distance, talk to the suspect with the benefit of cover, and use de-escalation strategies to resolve the situation. Instead, officers pulled only a few feet from Rice, predictably increasing the risk of deadly force (Pickering and Klinger 2016, 28). It likely was an avoidable killing caused by bad tactics.

Other police-generated killings involve officers using bad tactics that, though not intended to be deadly, have that effect. These incidents comprise a second category:

Police-Generated Killing Involving Unintentional Deadly Force: A killing by police where bad tactics by the officer(s) involved have unintentional deadly effects.

Not all police actions resulting in unintentional death qualify as a police-generated killing, as defined here. Sometimes officers exercise diligence and engage in best practices, yet by accident their actions prove fatal—such as striking and killing a pedestrian after a tire on a well-maintained police cruiser blows. Police instead deserve blame for tactics known to increase the risk of deadly force and unnecessary to achieve the law enforcement objective at hand.

One of the most notorious examples of a police-generated killing involving unintentional deadly force is that of Eric Garner in 2014. Officers confronted Garner for allegedly selling untaxed cigarettes on the street in New York City. Video shows Garner’s resisting arrest, but without being violent. Officer Daniel Pantaleo used a chokehold to take down Garner, keeping his arm compressed around Garner’s neck for approximately fifteen seconds. While restrained, Garner told officers “I can’t breathe” multiple times (Guardian 2014). Shortly after, Garner lost consciousness and was taken to the hospital, where he was pronounced dead. The medical examiner ruled the death a homicide, identifying the chokehold and pressure to Garner’s chest as causes of death (Goldstein and Santora 2014).

There is no evidence that Pantaleo intended to kill Garner, which would have been excessive given that Garner posed no imminent threat. But Pantaleo still erred in choosing a tactic less safe than other options and known to carry lethal risks. Despite no law at the time banning chokeholds, the New York City Police Department (NYPD) had a longstanding policy prohibiting the tactic: “members of the New York City Police Department will NOT use chokeholds. A chokehold shall include, but is not limited to, any pressure to the throat or windpipe, which may prevent or hinder breathing or reduce intake of air” (New York City Civilian Complaint Review Board 2014, 11). Pantaleo’s action clearly fits the NYPD’s definition of a chokehold, as video shows him applying pressure to Garner’s throat with his arms.

The deaths of Rice and Garner highlight bad tactics associated with police-generated killings. Though not meant to be exhaustive, the following list identifies police tactics for which there is growing evidence that they unnecessarily endanger life.

(1) Chokeholds and other neck restraints. Neck restraints are especially risky because they can restrict breathing or cut off blood to the brain with fatal consequences. Garner is one of many victims to have died at the hands of police using such tactics. Because of their inherent dangers, chokeholds and other neck restraints should be treated as deadly force. Yet police continue to use these tactics in circumstances where deadly force is not justified (Matteis 2015).

(2) Failure to use distance and cover. As the fatal shooting of Rice illustrates, when officers confront at close range a suspect believed to be armed and potentially dangerous but not an imminent threat, they make the encounter more dangerous for all involved. Maintaining distance and cover gives officers more time to react and helps keep nonlethal options available (Pickering and Klinger 2016, 28; Stoughton, Noble, and Alpert 2020, 167–74).

(3) Rushing to confront a dangerous suspect alone. Research on killings by U.S. police finds that unarmed suspects are more likely to be killed by an officer who is alone. That finding makes sense: a single officer is more vulnerable and more likely to have to rely on deadly force when responding to an actual or perceived threat. So it is unwise for an officer to rush to confront a suspect rather than wait for backup when the suspect, though dangerous, poses no immediate threat (Stoughton, Noble, and Alpert 2020, 185; Zimring 2017, 59–61).

(4) No-knock warrants and raids. The 2020 killing of Breonna Taylor in Louisville, Kentucky, during the execution of a no-knock warrant highlights this tactic’s inherent dangers (Oppel, Taylor, and Bogel-Burroughs 2021). No-knock warrants and raids began in the 1970s as part of the war on drugs and involve police forcibly entering a premise without knocking and announcing. Fearing a home intruder, some surprised residents respond with deadly force. No-knock warrants and raids have resulted in many avoidable deaths to suspects, bystanders, and officers (Dolan 2019).

(5) 21-Foot Rule. This rule developed in the 1980s advises officers to use deadly force against aggressors with an edged or blunt weapon who get within twenty-one feet. The 21-Foot Rule has come under criticism, especially since police in countries like the United Kingdom have developed effective nonlethal tactics to disarm aggressors with an edged or blunt weapon. Nonetheless, the tactic continues to be influential among U.S. police (Stoughton, Noble, and Alpert 2020, 168–71; Zimring 2017, 100–2).

In sum, various bad tactics lead to police-generated killings. The next section looks at whether these avoidable killings and the bad tactics behind them merit legal sanctions.

Accountability for Bad Tactics

The most challenging police-generated killings to evaluate are those where, though bad tactics precede deadly force, such force is necessary to stop an unjust threat to life. When officers choose tactics that raise the risk of deadly force, their error does not justify any response by a suspect. Officers should use distance and cover when engaging armed suspects who are not imminent threats. But if officers instead confront a suspect at close distance, command them to drop their weapon, and the suspect responds by threatening deadly force, officers face an unjust threat to life. Officers who do not use deadly force in this scenario jeopardize their lives and possibly others’ too. In these cases, it may seem unfair to sanction officers. If we can show that legal sanctions are appropriate even in these cases, it is safe to conclude that they also are appropriate for other police-generated killings where the exculpatory factors are weaker.

Potential insights for this task come from ethical and legal thinking on self-generated self-defense, where the person who engages in self-defense started the trouble (Leverick 2006, 109–29). Here is an example:

The Bar Fight: While at a bar, Sam provokes Hank by punching him once in the jaw. Hank responds disproportionately by pulling a knife. Faced with an imminent threat to his life appearing to require deadly force to stop it, Sam shoots and kills Hank. Though Sam ends the fight with intentional deadly force, he did not intend to kill Hank at the start of the fight.

Sam intentionally uses deadly force only after Hank makes an imminent and unjust threat against his life. Normally, that threat would justify deadly force, but Sam’s initial assault puts his self-defense claim in doubt. After all, if Sam never punched Hank, he likely would have never ended up in a situation requiring deadly force. Similarly, in police-generated killings, deadly force likely would not have occurred if officers had avoided tactics that unnecessarily escalated force.

Some may object to the analogy between self-generated self-defense and police-generated killings on the following grounds: police often respond to disturbances caused by others rather than start the trouble, like Sam does in the bar fight. But even if someone else starts the trouble, an officer intervening still has an obligation to choose tactics that reduce the risk of deadly force (assuming deadly force is unnecessary). When an officer violates that obligation and causes a police-generated killing, they are morally blameworthy in much the same way Sam is. Both the officer and Sam escalate force unnecessarily and play a causal role in the resulting death.

Cases of self-generated self-defense, like the bar fight, raise thorny questions about culpability (Leverick 2006, 109–29; Robinson 1985; Sangero 2006, 310–39). Broadly speaking, there are three different ways to treat this case:

None of these approaches enjoys a clear consensus, which is evident in law (Robinson 1985). The rationale behind (1) is straightforward: a common criterion for justified self-defense is that the person making the claim did not provoke the difficulty requiring force, and failing to meet that criterion precludes a claim to self-defense (e.g., State v. Moore 1975, 276). Arizona’s statute on self-defense reflects this approach. It only allows an initial aggressor to regain a self-defense justification by withdrawing or communicating their desire to end the conflict (Arizona State Legislature 2020). Other states like Iowa opt for (2). Its statute treats deadly force by an initial aggressor as justified when in response to force “grossly disproportionate to the provocation,” which they reasonably believe places them in “imminent danger of death or serious injury.” In line with the Model Penal Code, Iowa’s statute denies a self-defense justification to those who provoke an attack with the intention of having an excuse to kill, but potentially allows such a justification to those who provoke an attack without murderous intent (Dubber 2015, 164–65; Iowa Legislature 2021). Option (3), imperfect self-defense, emerged as a compromise position in the common law (Moreland 1952, 87–92). The Supreme Court of North Carolina and other courts recognize this doctrine and the penalty of manslaughter as appropriate for deadly force by someone who provoked the conditions that made it necessary (State v. Bush 1982, 159).

The goal here is not to resolve longstanding debates over whether (1), (2), or (3) is correct. But from these approaches we can draw two important conclusions for police-generated killings. First, the most lenient option still outlines sanctions for self-generated self-defense. Most readily recognize that Sam commits a blameworthy act and deserves some sanction for starting the fight leading to Hank’s death. Likewise, police-generated killings involve a clear wrong: failing to prioritize the protection of life by choosing tactics that unnecessarily raise the risk of deadly force. In this way, police violate an ethical obligation tied to their profession. Now not all ethical violations merit legal sanctions, which in many cases would fail to advance a legitimate state interest. But for police-generated killings, the state’s interest in protecting life gives it compelling reason to sanction ethical violations that cause such killings.

Second, penalties for self-generated self-defense apply to civilians who often lack any training in use of force or de-escalation. Such laws communicate the normative expectation that all should avoid actions that unnecessarily risk precipitating deadly force. That expectation should be even stronger for professionals trained in use of force and de-escalation. For this reason, democratic institutions have strong reason to prohibit bad tactics and impose meaningful penalties on officers who use them—especially when they cause an avoidable death.

Some may object to individual sanctions in these cases on the grounds that police-generated killings stem from system failures, not just errors by particular officers. When such killings occur, various factors typically are to blame—such as breakdowns in communication, inadequate training, and institutional pressures—similar to how plane crashes stem from multiple errors (Sherman 2018). This objection makes a critical point, but can err in minimizing officers’ role. For instance, Lawrence Sherman (2018, 439) identifies various factors leading to Tamir Rice’s death—miscommunication by the 911 dispatcher, the Cleveland Division of Police’s failure to contact departments where job candidates previously worked, and the pellet gun’s missing the standard orange mark to indicate it is a toy. Those observations are all true, but Sherman omits a crucial point. Even with all those factors, Rice likely would still be alive if officers at the scene chose better tactics. ⁴ Many factors can contribute to an outcome without eliminating the moral responsibility of individual actors (see Miller 2016, 138–57).

It is important to pair greater accountability with additional measures. Many police-generated killings result from bad tactics found in departmental training and policies, which highlights the need for systemic change. In a well-functioning system, there are rules on bad police tactics to avoid, regular training on alternative tactics that reduce the risk of deadly force, resources to employ those alternatives, and incentives to encourage their use. These institutional features aim to reduce police-generated killings and come with higher expectations for officers. Notably, mechanisms of accountability for individual officers help contribute to systemic change, since they create incentives for police administrators to implement training and policy aimed at avoiding tactics that prompt legal sanctions.

As to how to hold officers accountable, the law offers several options: (1) criminal penalties, (2) decertification, and (3) civil damages. All have a role with their own strengths and weaknesses. In response to high-profile killings by police involving bad tactics, the focus often is on (1) for understandable reason. Of the three sanctions, criminal penalties represent the most severe option and communicate most directly the gravity of the offense.

Though criminal penalties are important, it would be a mistake to solely rely on them to ensure accountability. One reason why is the high bar that must be met to impose criminal penalties. Criminal law is set up to err on the side of not convicting the guilty rather than convicting the innocent, and this principle applies across the board—including to officers whose bad tactics jeopardize life. Even in a fair and well-functioning criminal justice system, its high burden of proof would leave advocates of police accountability frustrated on occasion.

For this reason, there is value in pairing criminal penalties with other sanctions that require a lower burden of proof and can provide a meaningful floor of accountability for police-generated killings. Police decertification, which revokes an officer’s law enforcement license, is the most natural option to serve that function. Officers responsible for police-generated killings commit a grave violation of their professional obligations, and license revocation imposes a severe professional penalty in response. In contrast to just firing an officer, this sanction prevents them from serving in law enforcement elsewhere. ⁵ And since administrative law frequently relies on a lower standard of proof than criminal law—like preponderance of the evidence—it faces less formidable hurdles for ensuring accountability (Goldman 2003, 139).

Admittedly, lowering the burden of proof raises the risk of imposing sanctions on officers who were justified in using deadly force. Such increased risk is intolerable in the criminal context, where mistaken sanctions imprison the innocent, but is less problematic in the context of license revocation. In contrast to criminal penalties, mistakenly revoking a license—though regrettable—does not deprive an officer of a basic right like their liberty. Such mistakes are justified if they are an unavoidable part of an overall system that improves police accountability and conducts reliably accurate (but not foolproof) investigations of alleged misconduct.

In addition, professional penalties for police-generated killings can have indirect effects that improve accountability in the criminal system. If administrative law bans specific tactics, it becomes easier to show that officers should have known to avoid such tactics. Professional penalties thus are relevant for criminal convictions that require showing an officer acted recklessly or negligently (e.g., manslaughter).

Like police decertification, civil suits against officers require a lower burden of proof than what criminal law requires. When courts award civil damages to victims of police misconduct, local governments pay these damages (Schwartz 2014). In this way, civil suits hold democratic institutions accountable for failures in their role as oversight bodies of the police. Such suits also serve an important function that criminal penalties and decertification do not—providing restitution to those harmed most directly by bad police tactics.

The sanctions of criminal penalties, decertification, and civil damages together have the potential to provide meaningful accountability for bad police tactics. Current law, however, falls woefully short of ensuring such accountability, as the next section explains.

Where the Law Falls Short

This article opened by discussing the fatal shooting of Walter Scott, an unarmed fleeing suspect who presented no immediate danger. Deadly police force in such circumstances violates the law and can result in the sanctions listed above: criminal penalties, decertification, and civil damages. But that was not always true. If the shooting occurred in the 1970s, in many jurisdictions it would have been “lawful but awful.” The category of lawful but awful police force can shift, which reminds us that current failures to sanction bad tactics are far from inevitable.

Tennessee v. Garner (1985) illustrates this point. Here the Supreme Court intervened and had perhaps its greatest success in restricting excessive police force. Garner prohibits police from shooting fleeing suspects who pose no significant threat of death or serious bodily injury to others, and deems such force an unreasonable seizure in violation of the Fourth Amendment. Before the decision, numerous police departments had already scrapped the fleeing felon rule, which allowed officers to use deadly force to prevent the escape of felony suspects, regardless of whether they were dangerous. Killings by police decreased without increases in officer deaths, as more departments banned deadly force against nondangerous fleeing suspects (Sherman 2018, 425–28). These findings contributed to a growing consensus that the fleeing felon rule was inconsistent with best practices and unnecessary to enforce the law. The court cited this shift as a reason for its ruling (Tennessee v. Garner 1985, 10–11, 18–19). What resulted was a specific rule to guide police on when they could use deadly force, and evidence suggests that the ruling further helped reduce killings by police (Tennenbaum 1994).

The effectiveness of Garner lies in providing concrete guidance for police to avoid a specific bad tactic. But the same reason for why Garner is effective—its specificity—also limits it as a panacea for bad tactics. Many police tactics beyond what Garner addresses also endanger life unnecessarily. Before deciding whether to shoot, police make various tactical decisions that can have lethal consequences, and for those decisions Garner offers limited guidance.

In Graham v. Connor (1989), the Supreme Court put forth a more general rule for evaluating police force. According to Graham’s “objective reasonableness” standard, what matters in determining the constitutionality of police force is not an officer’s intentions, but whether the force in question would be justified from the perspective of a reasonable officer in the same situation. This principle has the virtue of offering guidance for a broader range of action than Garner does. Yet that apparent advantage comes with a downside: Graham offers a principle whose guidance is far less clear. How exactly a “reasonable” officer would act has sparked much disagreement since Graham (Alpert and Smith 1994; Harmon 2008). That is especially true in the decentralized context of U.S. policing, which consists of nearly eighteen thousand different law enforcement agencies that lack uniform policies and training (Reaves 2011, 2).

Language from Graham further adds to the confusion over how to apply its objective reasonableness standard. An oft-quoted passage from Graham reads:

The “reasonableness” of a particular use of force must be judged from the perspective of a reasonable officer on the scene, rather than with the 20/20 vision of hindsight. . . . The calculus of reasonableness must embody allowance for the fact that police officers are often forced to make split-second judgments—in circumstances that are tense, uncertain, and rapidly evolving—about the amount of force that is necessary in a particular situation. (Graham v. Connor 1989, 396–97)

Notably, this passage echoes language from Justice Sandra Day O’Connor’s dissent in Tennessee v. Garner (1985, 23) and has become a common justification by the court for bad police tactics—a development that complicates Garner’s legacy (Obasogie and Newman 2018, 1476). By incorporating concerns from O’Connor’s earlier dissent, Graham signaled the court’s reluctance to second-guess tactical decisions by police. Certainly, any evaluation of police force must consider the stress and uncertainty experienced by officers at the time. The problem with Graham, however, is that its emphasis on split-second decisions in police work can discourage analysis of deliberate tactical planning that often comes prior. ⁶

Lower courts have split on how to apply Graham, with some considering prior tactics when evaluating the reasonableness of police force and others restricting their focus to the final frame when force is used (Noble and Alpert 2010, 486–87). Courts also have been inconsistent in incorporating the latest understandings of best police practices into their analyses of reasonable action (Garrett and Stoughton 2017). Despite such confusion, the Supreme Court has doubled down on Graham’s approach of eschewing specific rules against bad tactics. That is evident in Scott v. Harris (2007), which deems it constitutional for police to terminate a high-speed chase in response to a minor traffic violation by ramming the suspect’s car—an action that left the suspect paralyzed. The ruling explicitly rejects attempts to formulate “an easy-to-apply legal test” to evaluate specific police tactics (Scott v. Harris 2007, 383).

The Supreme Court’s embrace of the doctrine of qualified immunity exacerbates the failures of constitutional law to rein in bad police tactics. This doctrine says that police officers and other government officials are protected from civil damages for constitutional rights violations unless “clearly established” law prohibited their actions. Pearson v. Callahan (2009) even allows courts to skip their analysis of whether an officer’s actions violated another’s constitutional rights and go straight to deciding whether the officer broke clearly established law. That wrinkle stunts the law’s development. When there is no analysis of whether police tactics violated someone’s rights, this approach forecloses the opportunity to establish that unconstitutional actions are in fact unconstitutional, and bad tactics continue without sanctions (Schwartz 2017, 65–66).

So despite Garner’s success in prohibiting a bad tactic and reducing police killings, the Supreme Court has resisted that approach in other cases. When confronted with many of the bad police tactics discussed above, the court has avoided taking action to ban them. In City of Los Angeles v. Lyons (1983), the court sidestepped the question of whether chokeholds were unconstitutional, allowing them to remain police practice throughout many parts of the country. In Richards v. Wisconsin (1997), the court ruled that no-knock raids by police were constitutional. And in City and County of San Francisco v. Sheehan (2015), the court granted qualified immunity to officers who shot a woman with mental illness holding a knife in an encounter precipitated by poor tactics. These and other cases suggest a troubling conclusion: instead of serving as a bulwark against bad tactics, constitutional law too often empowers police to use them—especially against the most vulnerable and marginalized (Butler 2016).

There are, of course, tools beyond constitutional law to promote accountability for bad tactics. Constitutional law sets minimum standards that police nationwide must uphold, and states can go beyond those requirements by prohibiting additional tactics that unnecessarily endanger life. Many states, though, fail to pursue this option. In fact, an analysis of state criminal statutes on police deadly force finds that many outline standards less restrictive than what Garner requires (Stoughton, Noble, and Alpert 2020, 81–86). Such statutes, often enacted before Garner, represent more than just dead laws superseded by federal rulings. They continue to have pernicious effects by entrenching outdated standards in police policy (e.g., Pittsburgh Bureau of Police 2021, 3) and limiting the ability of prosecutors to obtain convictions of officers responsible for avoidable killings (e.g., Palmer 2019). In addition to these deficiencies in criminal law, state administrative law remains underdeveloped as a tool for revoking the licenses of officers who engage in bad tactics (Goldman 2016).

All these factors hinder meaningful accountability for police-generated killings. Consider the shooting of Tamir Rice. In this case, the prosecutor argued against criminal charges and commissioned two reports that drew heavily on Graham to claim that the officer’s deadly force was objectively reasonable (Crawford 2015; Sims 2015). The grand jury then followed the prosecutor’s recommendation to not charge the officers (Williams and Smith 2015). Having escaped legal sanctions, the officer who made the critical error of driving immediately in front of Rice received only a five-day suspension (Associated Press 2018a). The officer who shot Rice was fired, but for a reason unrelated to the shooting—providing false information on his job application (Calamur 2017). Since his license was not revoked, another department was able to hire him (Associated Press 2018b). Without clear case law or statutes prohibiting the tactics that led to Rice’s death, accountability proved elusive.

One way to ensure greater police accountability is for legislators at the local, state, and federal level to take a more aggressive approach to outlining and prohibiting in law specific bad tactics. Some may question this approach on the grounds that any list of bad tactics will prove incomplete and require revision in light of new research. Certainly, the law will require updates as new evidence builds that specific tactics raise the risk of deadly force unnecessarily. But such work is the inevitable price of meaningful police oversight in a democratic society.

This approach offers distinct advantages to excessive reliance on something more general like Graham’s objective reasonableness standard, which fails to provide officers concrete guidance. As Egon Bittner (1970, 38) notes, it “smacks of . . . perversity” to empower police to use force without clear guidelines. Rules prohibiting specific bad tactics can save lives, evident from the reduction in police killings after the ban on deadly force against nondangerous fleeing suspects.

The Case and Tools for Reform

Despite the dangers of police-generated killings, some still may see them as rare events that do not merit significant attention. There are several reasons to reject this skepticism. First, police-generated killings may occur more often than many assume. Each year in the United States, police kill hundreds of suspects who have a weapon less lethal than a firearm, like a knife or club, or no weapon at all (Washington Post 2021). Policing in other countries suggests that many of these killings can be avoided. In the United Kingdom, police kill individuals at a far lower rate. That is partly due to fewer handguns in the population, but police tactics also play a role. Since many officers do not carry firearms, U.K. police have had to develop nonlethal tactics to disarm suspects without firearms, such as repositioning, use of protective shields, and not rushing to resolve incidents (Police Executive Research Forum 2016, 88–105). These tactics do not appear to sacrifice officer safety, given the low numbers of police in the United Kingdom who are killed (Zimring 2017, 88).

Second, regardless of their frequency, police-generated killings can have outsized effects. Not every killing by police has the same impact. Killings involving questionable force are more likely to generate protests, erode trust in and cooperation with the police, and hinder public safety. Indeed, public trust in the police tends to drop after high-profile incidents of questionable police force, especially among groups most directly affected (Weitzer 2002). After the killings of Breonna Taylor, George Floyd, and other Black lives in 2020, a Gallup poll found that only 19 percent of Black Americans had confidence in the police (J. Jones 2020). The frequent lack of accountability in response to police-generated killings only compounds this distrust and undermines the legitimacy of government institutions more broadly.

So there are compelling reasons to pursue reforms tackling the problem of police-generated killings. If successful, such reforms can save lives and foster greater trust between police and communities. Democratic institutions have several tools to advance those goals:

(1) Criminal statutes on police deadly force that prioritize the protection of life. In the fight against bad tactics, some emphasize that police administrators should lead through reforms to departmental policy (Zimring 2017, 219–38). That strategy is worth pursuing—police administrators should do everything in their power to prevent unnecessary killings. But the reality is that, without the intervention of lawmakers, many agencies never adopt reforms (Gilbert 2017). Criminal law serves as a tool for implementing stronger and more uniform sanctions for bad tactics. Currently, state statutes permit police deadly force across a range of circumstances where it is unnecessary to protect life against an imminent threat. Many statutes also fail to prohibit egregious tactics like chokeholds. Such broad permissions show up in police policies, training, and practice, as well as create obstacles for obtaining convictions after police-generated killings. Recently, a few states have taken steps to address these problems through enacting criminal statutes that narrow the justification for police deadly force and ban chokeholds (Chabria 2019; Ferré-Sadurní and McKinley 2020). More states should do the same.

(2) License revocation for officers who engage in bad tactics. State administrative law has the potential to define and ban bad police tactics that unnecessarily risk life, as well as impose meaningful sanctions for their use. Though some states list unjustified deadly force as a reason for license revocation (Goldman 2012, 152), they need to go further and specify bad tactics that would result in loss of license, especially for repeated violations or violations that result in police-generated killings. To provide a meaningful floor of accountability even when other mechanisms fail, states should be able to pursue decertification independently of whether an officer is criminally convicted or terminated following arbitration outlined in a union contract (Goldman 2012, 150–51; Rushin 2019, 586–87). These reforms would incentivize Peace Officer Standards and Training (POST) commissions—which in most states have authority over police training requirements and revoking licenses (Goldman 2003, 122)—to train officers to avoid bad tactics banned by law. States also can pass laws detailing training requirements.

(3) Requirements that a portion of patrol officers not carry a firearm. In the United Kingdom, most officers do not carry a gun and kill at far lower rates than in the United States. Not surprisingly, compared with the United States, U.K. police have shown greater urgency in adopting nonlethal tactics that can help resolve encounters with suspects who are unarmed or have a weapon less lethal than a firearm. Such tactics are critical for officer safety when they cannot rely on a gun as a crutch to make up for bad tactics. Though advocated by some (Vitale 2017, 25–27), removing guns from U.S. officers has remained a fringe proposal due to worries that it would endanger officers given the higher number of guns in the U.S. population. But even if a higher proportion of U.S. officers need to carry firearms compared with their U.K. counterparts, steps can be taken to remove guns from officers on less dangerous patrols. For these officers, it would be imperative to learn nonlethal tactics when encountering aggressors with a weapon less lethal than a firearm, which could help spur training in tactics currently neglected in the United States. Local governments and universities can advance this goal by requiring a portion of their department’s officers to not carry guns.

(4) Enhanced constitutional protections against bad tactics. In Garner, the Supreme Court banned a bad tactic by ending the blanket permission for police to shoot fleeing felons to prevent their escape. This intervention came after numerous states and law enforcement agencies had abandoned the fleeing felon rule. Looking forward, if more states and law enforcement agencies abandon other tactics that unnecessarily threaten life, future litigation can make the case that these tactics no longer are reasonable and therefore violate the Fourth Amendment. This long-term strategy can bolster protections against bad police tactics and ensure that they apply nationally (see Garrett and Stoughton 2017). In addition, the doctrine of qualified immunity hinders courts from recognizing constitutional violations caused by bad tactics. Congress has the power to end this judicial invention, a step that recently has gained support (Sonmez, Kane, and Colvin 2020).

Together, these recommendations call for higher police standards. They place additional restrictions on what tactics officers can use and thus will meet opposition from some police organizations. When the Police Executive Research Forum called on police to voluntarily implement tactics to reduce the risk of deadly force, both the Fraternal Order of Police and International Association of Chiefs of Police denounced the idea (Jackman 2016). The above reforms go further than ones already rejected by some of the most powerful police organizations.

Despite formidable challenges facing reform efforts, history does offer reason for hope. Some of the same organizations resisting reforms today also opposed ending the fleeing felon rule. Before Garner, the membership of the International Association of Chiefs of Police passed a resolution supporting the fleeing felon rule by a four-to-one margin (Kleinig 1996, 114). In this case, dire warnings that ending the rule would increase crime and endanger officers proved unfounded. A combination of research, voluntary police reform, state legislation, and Supreme Court intervention ended the fleeing felon rule and saved lives. Similar reforms at all levels are now needed to address police-generated killings and the moral hazards they pose.