What this covers

Rachel Harmon, a law professor, speaks with Tyler Cowen about police reform and the structural incentives that shape officer behavior. Rather than pursuing a single fix—better hiring, body cameras, or larger forces—Harmon argues that lasting change requires making the full costs and benefits of policing visible to the elected officials who set policy, then building accountability mechanisms and national standards that force those costs to be internalized. She contends that departments create officers, meaning culture, supervision, and incentive structures matter far more than recruitment practices in determining how officers behave on the street.

The conversation ranges across why officers coerce sex from vulnerable arrestees without explicit force; how the exclusionary rule reshaped police culture by teaching constitutional law; why stop, question, and frisk likely did not drive New York's crime decline despite its enormous social cost; and how federal funding often incentivizes the wrong behavior—rewarding arrests over prosecutions, for instance. Harmon presses the distinction between arrest and criminal charges, questions whether adding officers improves safety when we cannot measure harm, and explains why body cameras alone fail without clear standards and consequences. She advocates for internalizing the costs of policing through transparent indemnification data, national standards modeled on England's system, and a broader legal framework that weighs the four-part evaluation of policing interventions: effectiveness, benefit, efficiency, and fair distribution. Camden's reform under Scott Thomson illustrates what's possible when leadership, community buy-in, and the ability to rebuild culture align.

Sharpest takeaway

Harmon argues there is no silver bullet for police reform; the key lever is internalizing the full costs and benefits of policing through accountability, incentives, and national standards rather than relying on selection, body cams, or simply hiring more or different officers.

  • Departments create officers; culture, incentives, and supervision matter more than who is selected
  • Costs of policing must be made visible and internalized by the political actors who set policy
  • National standards could help even within a fragmented, locally-organized system

The argument · threads10 threads · 40 claims
0.85

Department culture, incentives, supervision, and discipline shape officer behavior more than selection or composition.

8 pointscentrality 5/5
  • We can screen out some bad officers (e.g. via state decertification, minimum age standards above 18) and education or diversity may help subtly, but we cannot select our way out of problems in policing because departments create officers—officers don't preexist a department—so the real determinants are department culture, incentive structures, supervision, and discipline, and good officers can be made from imperfect people.

    “departments create officers. The officers don’t preexist a department, really, so what you’re really looking at is the culture of the department, the incentive structures, the supervision, discipline. You can make good officers with imperfect people.”

  • Even an inherently cautious officer will respond to incentives: if required to complete every call in 90 seconds when facing someone with a knife in mental health crisis, the officer will tase rather than take time to de-escalate, illustrating that incentives about how a situation must be handled, not officer composition, govern use of force.

    “If you tell me that I have to complete every call in 90 seconds, and I face somebody who is holding a knife and clearly in a mental health crisis, I have two options. I can wait and talk and see if I can de-escalate the situation, or I can tase him and be done with it... If you rush her, she’ll tase them.”

  • The capacity to solve problems without force and the motivation to do so matter at least as much as composition or selection; reform requires giving officers alternatives, empowering them to use those alternatives, and providing the time and backup needed for less forceful responses, though evidence on de-escalation and implicit bias training is currently lacking.

    “the capacity to solve problems without force and the motivation to do so matter at least as much, if not more, and so what we have to do is give officers that capacity.”

  • Cameras alone won't change behavior because officers forget over time and many problems aren't intentional misbehavior; if a department doesn't care whether officers are polite, a camera won't make them polite—what's needed are standards, consequences, motivation, and the capacity to behave differently.

    “If a department doesn’t care if you’re polite or not, putting on a camera isn’t going to make you more polite. There would have to be standards. There’d have to be consequences. Officers have to have the motivation to behave differently, and they have to have the capacity to behave differently.”

  • You can't infer that women officers will be less violent from women's lower civilian violence, because police officers are armed and trained to use force, so they are situated differently with respect to force regardless of average size differences.

    “Police officers are armed, and they’re trained to use force, so there’s a difference. Even if women are, on average, smaller, there’s still going to be a difference in their preparedness to use force to enforce a command.”

  • In a community where everyone is nice, police violence would fall through several channels—prevention of underlying problems via education, economic development, and mental health services; re-identifying and tolerating minor disorder; and citizen trust that makes people comply with mere requests—and officers in such a town would also be more likely to wait out noncompliance and seek alternatives.

    “If people trust the police in your nice town where people believe that the police are there to help, not to hurt... they might be far more likely to respond to just a mere request... So I think niceness could help, but thinking about how it could help is really what matters.”

  • Popular culture has shifted from Officer Friendly to Training Day, conveying a message about policing that can distort both public perception and police perception of what the norms are, just as media distorts perceptions in other contexts.

    “we used to have Officer Friendly, and now we have Training Day... it certainly portrays a message of policing that can distort both public perception and police perception of what the norms are”

  • Officer health and well-being must be prioritized because overly stressed or physically/mentally hurt officers are more likely to hurt others; officer suicides are a huge problem, peer services often work better than mandated therapy, but states must balance encouraging officers to seek help against removing officers who are risky to others.

    “we have to prioritize the health and well-being of police officers because police officers who are overly stressed or who are hurt physically or mentally are more likely to hurt others.”

0.74

Criminal justice should contract by decriminalizing low-level offenses and redirecting social problems to non-police responders.

5 pointscentrality 5/5
  • There is not much of a right to a fair trial today; the system is clogged with offenses never meant to be prosecuted, using criminal justice to address problems better handled elsewhere, with tons of misdemeanors dropped on arrival in court and severely limited defense counsel—so reducing the scope of the system would improve it.

    “No, no, there’s not much of a right to a fair trial... we clog the system with a lot of offenses we never mean to prosecute... I think we could do a lot to improve the system by just reducing the scope of the system.”

  • Some tasks now dumped on police—crises involving substance abuse, mental health, or homelessness—would be better handled by trained social workers, and providing more mental health services and social workers responding to crises (with or without law enforcement) could prevent crises and yield more effective policing.

    “There are some aspects of what police departments are expected to do now that would be better handled by social workers who are trained to deal with people in crisis... There is no police department in America who wouldn’t say what we need is more mental health services, because mental health services would prevent the crises.”

  • The decision to arrest someone and the decision to sanction them criminally are not the same decision; many behaviors we dislike can be addressed without arrest, as shown by Harvey Weinstein—a violent felon allowed to self-surrender with pre-negotiated bail—so arresting every low-level offender (e.g. marijuana possession) imposes large aggregate harm without producing public safety.

    “the decision to arrest somebody and the decision to sanction them criminally aren’t the same decision... We didn’t arrest Harvey Weinstein. He was allowed to self-surrender with a pre-negotiated bail package... This was a violent felon... And we didn’t arrest him.”

  • Harmon rejects the premise that reducing arrests causes bedlam: New York dramatically dropped misdemeanor arrests (e.g. for simple marijuana possession) without a substantial increase in crime, and arrests should be evaluated like any intervention—on effectiveness, benefit (costs vs benefits), efficiency (no less harmful alternative), and fair distribution—tests that many arrests fail because they impose unnecessary harm.

    “the decision to arrest somebody and the decision to sanction them criminally aren’t the same decision... We didn’t arrest Harvey Weinstein. He was allowed to self-surrender with a pre-negotiated bail package... This was a violent felon... And we didn’t arrest him.”

  • The assumption that low-level offenders won't appear in court is unfounded: most people who get traffic tickets show up or pay, court appearance is highly malleable, and merely reminding people raises appearance rates by about 10 percent, while low-level theft penalties are not severe enough to make people flee.

    “most people who get traffic tickets show up or pay their ticket. And we know that’s totally malleable. That is to say, all we have to do is remind you, and then the percentage of people who show up goes up 10 percent.”

0.63

The U.S. lacks coherent national police data infrastructure despite fragmented local agencies, hampering reform and accountability.

2 pointscentrality 5/5
  • The United States is bad at gathering police data because it has over 12,000 (formerly cited as 18,000) highly diverse local law enforcement agencies, and there has been no national will to mount a coherent, standardized data-collection effort.

    “we have more than 18,000 law enforcement agencies in the United States, and that’s a lot of data to collect. They’re very diverse, and so it’s hard. We’d have to have a national effort to do it in any sort of coherent way, and we haven’t had the will to do that.”

  • We collect crime data effectively (uniform crime reports for nearly 100 years with standardized definitions), but there is far less interest in collecting data on police coercion such as stops and uses of force, because political demands have focused on public safety effectiveness rather than the harms policing imposes.

    “We do actually collect a lot of data from police departments, pretty effectively, with respect to crimes... There’s just a lot less interest in collecting data on the coercion that police officers do, like arrests... but stops, uses of force, and the like.”

0.53

Better accountability through transparent cost tracking and community input into policing standards is the most promising reform lever.

3 pointscentrality 4/5
  • Harmon favors indemnification (the city paying judgments against officers) over personal officer liability because, if payments are made transparent without nondisclosure agreements and tracked as data, communities can pressure their governments to train, supervise, and discipline officers better—the goal being to internalize and fairly distribute the full costs of policing among the political actors who set policy.

    “I’m actually pro-indemnification, because I think if the city pays, then as long as we make that payment transparent so we don’t have nondisclosure agreements, then we can act, and we also collect data on it.”

  • Better accountability—where officers are responsive to standards, must explain their behavior, and face consequences for failing them—is the most promising lever, but the standards must reflect the full costs and benefits of policing, which requires knowing those costs by hearing from the communities on whom the costs of policing are concentrated.

    “Those standards should reflect the full costs and benefits of policing so that we set rules for policing that actually get what we want out of policing... that means hearing from the members of the community on whom we tend to concentrate the costs of policing.”

  • If able to make one lasting change, Harmon would establish national standards—beyond federal effectiveness funding, lawsuits, and prosecutions—to ensure policing is harm-minimizing, fair, and authorized by local communities, modeled on England's approach, which could help substantially even while policing remains organized in many local departments.

    “setting up some national standards so that we not only have a federal intervention in making police more effective... but national standards for ensuring that policing is actually beneficial, that we’re harm minimizing, that it’s fair, and that it’s authorized by local communities.”

0.53

Police sexual assault is vastly underreported and underinvestigated, requiring mandatory officer duty to report and supportive reporting cultures.

3 pointscentrality 4/5
  • Unlike many uses of force whose legality is ambiguous, sex between a police officer and someone he has just arrested is unambiguously wrong, so officers must have a duty to report each other's misconduct and departments need an internal culture that facilitates such reporting, since the danger is officers being alone or with peers who won't report.

    “There’s no question that sex between a police officer and somebody he’s just arrested is wrong. That is just pure bad, and so we should never have a situation in which another officer has information or evidence or has witnessed such a thing and is unwilling to report it.”

  • Sexual assaults by police are far more common, less likely to be reported, and harder to prove than the public realizes; complaints come in sporadically as 'he said, she said' from victims with credibility problems (felons, prostitutes, runaways), so departments and even the Justice Department repeatedly close cases, yet thorough investigation often reveals a single officer committed dozens of prosecutable incidents.

    “I don’t think they realize how common sexual assaults by police are, how unlikely they are to be reported, and then how difficult they are to prove once they are reported... when you would look at these cases and start actually seriously investigating, you’d realize that actually, this officer had done it dozens of times.”

  • Officers don't need force or explicit threats to coerce sex because they control the situation—the person is arrested, the officer is armed and acts invincible—and abusers carefully choose victims with criminal histories who believe they cannot resist, which exposes a legal gap because proving a federal felony required proving force.

    “you would have to prove force to prove a felony, but officers don’t have to use force, and they don’t even have to explicitly threaten force. They control the situation. The person is under arrest. They’re wearing a gun, and they express and act as if they’re invincible.”

0.49

Actual reform requires sustained local knowledge and resistant cultures can outlast reform-minded leaders without deep structural change.

2 pointscentrality 4/5
  • Because police institutions are local, you have to know a great deal about what is happening on the ground in a community to predict whether a department will actually change, which is why on-paper formal reforms (as in Minneapolis) can fail to generate real change even after high-profile incidents.

    “One of the things about police institutions is that they are local institutions. So you have to know a lot about what’s going on in a community to know how likely they are to change.”

  • Camden's improvement under Scott Thomson shows reform succeeds through innovative leadership, broad buy-in from communities and officers (e.g. a progressive use-of-force policy based on ALI principles), and the power to rebuild a department from the ground up, because resistant department cultures can otherwise 'wait out' reform-minded leaders who are likely to be fired within a year.

    “He did that by getting everyone to the table and getting buy-in from everyone involved, and that’s a lot more likely to have success, lasting success. So leadership matters. Buy-in matters.”

0.44

Very small police departments lack accountability mechanisms, and federal funding often incentivizes harmful policing rather than effective or fair practice.

8 pointscentrality 2/5
  • Reducing police salaries too far (as in small or poor departments) lowers quality by attracting officers with misconduct records, but there is no evidence that raising salaries 50 percent or doubling them would yield radically higher-quality officers.

    “when we reduce police salaries a lot... then police quality goes down because you start getting officers with records of misconduct and the like... But I don’t know that if we doubled salaries or raised them 50 percent that we’d get radically higher-quality police officers.”

  • About half of the roughly 12,000 local departments have fewer than 10 officers, and in such tiny offices accountability mechanisms like use-of-force policies and disciplinary structures don't matter, so some consolidation would be valuable—though consolidation has costs because diverse communities may not get the policing they want.

    “About half of those are fewer than 10 guys. Now, we talk about police accountability and having use-of-force policies and disciplinary structures, but in an office of 10 guys, none of that matters.”

  • Most federal money to local police funds effectiveness or national priorities rather than accountability, and programs are often designed in ways that incentivize harmful policing—for example the Violence Against Women Act arrest program rewards arrests even above prosecutions, encouraging pro-arrest policies even though arresting people doesn't necessarily protect public safety and does increase harm.

    “That program specifically incentivizes arrests, and it incentivizes arrests even above prosecutions... it encourages departments to adopt pro-arrest policies and states to adopt pro-arrest laws. But arresting people doesn’t necessarily protect public safety, and it does increase harm.”

  • The truth law students are least likely to learn is that policing is not all about constitutional law; they learn the Fourth Amendment and Miranda and think they know the law of the police, but the full law relating to policing requires weighing all the harms and benefits of policing, with law playing a critical role in setting standards.

    “When you go to law school, what you learn about policing is the Fourth Amendment and Miranda, and you think you’ve learned the law of the police... we have to consider all the harms and benefits of policing, that law plays a critical role in setting standards”

  • Civil asset forfeiture is, in Harmon's view, simply all bad.

    “Unless you ask me about civil asset forfeiture, and then I’ll just tell you it’s all bad.”

  • Some research suggests adding officers improves public safety even when arrests don't rise—officers act as a sentinel discouraging crime—but we don't know whether adding officers also increases harm, so we cannot say whether doubling a force makes things better or worse; it depends entirely on what those officers are directed to do.

    “if you add officers to a department, public safety gets better, even if arrests don’t go up. So the presence of officers as a sentinel to discourage misconduct or to discourage crime actually works pretty well. But what we don’t know is whether adding those police officers also increases harm”

  • A promising future legal scholar shows two traits: being 'a dog with a bone'—a set of fundamental questions that bothers them enough to keep returning to it—and choosing a question that is a rich vein to mine, likely to yield results over many years.

    “one is, I want to see someone who’s a dog with a bone... Then... I want to make sure that bone is a rich vein to mine. I want to know that that’s going to yield results over years and years.”

  • The number of law schools could be reduced, which wouldn't be a bad thing, because some law schools are effectively diploma mills producing lawyers who lack a good shot at JD-required or JD-valued jobs and therefore cannot pay off their loans; schools ranked 150-200 likely should not survive, while those ranked 30-40 should.

    “Some law schools are effectively diploma mills that produce lawyers who don’t have a good shot of getting a JD-required or JD-valued job, which means they can’t pay off the loans that they take out to go to law school.”

0.43

Officers face almost no consequences for lying, and exclusionary-rule suppression of evidence is costless to dishonest officers.

4 pointscentrality 3/5
  • The exclusionary rule has profoundly changed police culture by giving departments an incentive to teach officers constitutional law and officers an incentive to learn and follow it; before the rule officers knew nothing about constitutional law, after it every officer treats the Fourth Amendment as part of his manual—illustrated by a corrupt cop who still wouldn't fabricate probable cause because that was the part subject to the rule.

    “the effect of the exclusionary rule, in large part, was to give police departments an incentive to teach officers the law and give officers an incentive to learn that law and follow it, at least in large part.”

  • There are almost no consequences for police officers who lie: under the exclusionary rule the worst outcome of an officer lying at a suppression hearing is the evidence being suppressed—exactly what would have happened had he told the truth—so lying is a costless gamble, while internally, departments often fail to act and officers who admit lying learn that admission is more punished than the lie itself.

    “A police officer engages in an illegal search. Then he lies at trial, taking the chance that he might get it in. At worst, the only outcome is what would have happened if he had told the truth. So I don’t think that’s a very effective incentive.”

  • Police should not be able to open a person's smartphone on the street, but with a warrant they may, since phone contents can be important evidence for convicting someone of a serious crime.

    “I think the police should not be able to open your smartphone when they see you on the street, but with a warrant, sure.”

  • There is nothing that focuses your attention more than someone trying to hit you in the face—it is terrific at screening out all distraction, to the point that you forget who you are when you box.

    “there is nothing that focuses your attention more than someone trying to hit you in the face. It is absolutely terrific at screening out all distraction. You forget who you are when you box.”

0.39

Body cameras and diversity recruitment produce minimal behavioral change without accompanying standards, consequences, and cultural shift.

3 pointscentrality 3/5
  • Studies on gender show a slightly better effect than studies on racial diversity in policing, and communities value departments that reflect them, but diversity should be pursued as an end for other reasons rather than treated as a solution to problems in policing.

    “The studies on gender show a little bit better effect than the studies on racial diversity in policing... but I don’t take diversity in policing to be a solution to problems in policing. I think it’s an end that we might pursue for other reasons.”

  • Studies of body cameras—measuring whether they reduce complaints or uses of force—are at best mixed, and the evidence that they do good in those respects is pretty weak; body cams currently function more as anecdote than data, though machine learning may make them genuine data over time.

    “the results of studies of body cameras, at least in terms of if you’re studying whether they reduce complaints or whether they reduce uses of force, the studies are, at best, mixed. I would say the evidence that they do good in these respects is pretty weak.”

  • Proactive 'stop, question, and frisk' and broken-windows policing probably did not drive New York's murder-rate decline, because cities that didn't adopt those strategies also saw substantially lower crime and murder rates today; policing had some effect, but stop-question-frisk as carried out in New York was so vastly costly that its benefits must be weighed very cautiously.

    “was it proactive policing? Was it “stop, question, and frisk” and broken-windows policing that drove the murder rates down? And the answer to that is probably no. I think cities that didn’t engage in those strategies also had substantially lower crime rates and murder rates than they do today.”

0.35

Federal DOJ pattern-or-practice lawsuits and Justice Department enforcement matter symbolically and practically but are expensive and inconsistently durable.

2 pointscentrality 3/5
  • The Justice Department should retain the power to sue police departments for patterns or practices of unconstitutional conduct; such suits are important symbolically and practically and have reformed big problematic departments, but they are very expensive, work unevenly, and their reforms don't always last—so they are not a panacea.

    “The lawsuits by the Justice Department for pattern or practices of unconstitutional conduct by a police department are important, both symbolically and practically... I don’t want to suggest that they are somehow a panacea to problems in policing”

  • The substantive (non-headline) difference between the Obama and Trump administrations on policing was that under Sessions the DOJ effectively shut down new pattern-or-practice lawsuits (trying to exit the negotiated Baltimore and Chicago consent decrees, succeeding in Chicago), reversed restrictions on the 1033 military equipment program, and rhetorically advocated more aggressive policing (stop-question-frisk, letting suspects bang their heads on car doors), signaling that the administration valued aggressive over constitutional or harm-efficient policing.

    “When Sessions came in as Trump’s first attorney general, he effectively shut down the pattern-or-practice lawsuits that the Justice Department was carrying out... no new investigations or almost no new investigations.”