In This Issue
Winter Issue of The Bridge on Complex Unifiable Systems
December 15, 2020 Volume 50 Issue 4
The articles in this issue are a first step toward exploring the notion of unifiability, not merely as an engineering ethos but also as a broader cultural responsibility.

Policing as a Complex System

Friday, December 18, 2020

Author: Brendan O’Flaherty and Rajiv Sethi

Policing in the United States is decentralized. There are more than 17,000 law enforcement agencies, many with overlapping jurisdictions. They each have their own organizational culture, their own protocols for selection and training, and even their own procedures for recording and reporting data. They interact with local prosecutors and judges. They exchange personnel—officers dismissed or disciplined at one location can often find employment at another. And they operate under different laws and guidelines concerning the use of force, including deadly force.

Historical Context

The relationship between police and African American communities has historically been fraught. As Randall Kennedy (1997, p. 26) observed in Race, Crime, and the Law, for centuries it was a crime “for blacks to do all sorts of things deemed to be permissible or admirable when done by others,” including learning to read, defending themselves from abuse, or fleeing enslavement. The police were called on to enforce laws that were oppressive in the extreme, resulting in cycles of mistrust, as Terrence Cunningham, then president of the International Association of Chiefs of Police, acknowledged and apologized for in 2016 (Kennedy 2016).

Even as explicitly discriminatory laws were repealed, police practices continued to contribute to mistrust. In 1966 James Baldwin observed that black neighborhoods were “policed like occupied territory.” And “since [the police] know that they are hated, they are always afraid…a surefire formula for cruelty.”

Over the past few years, the proliferation of smartphones, security cameras, dashcams, and bodycams has vividly brought many instances of fear and cruelty to light. The killing of Philando Castile and the non-fatal shooting of Levar Jones revealed unwarranted fear, and the killing of George Floyd callous indifference to human life. The latter, in particular, appears to have marked a watershed in American discourse on police use of force. The mass protests in June 2020 were on a scale unseen in recent history, with millions of people joining demonstrations in more than 500 cities.

When egregious incidents of deadly force come to light, a natural response is to seek accountability. The focus is on the officer’s conduct and on the response of the justice system. However, criminal prosecutions of officers involved in deadly force incidents have been rare, and convictions rarer still.

Under a federal standard established by a 1985 ruling in Tennessee v. Garner, police officers may use deadly force if they believe that a suspect “poses a significant threat of death or serious physical injury to the officer or others.” But fear of death is a subjective state of mind, not easily verified, and prosecutors and jurors have been reluctant to second-guess claims that such fear was present and reasonable under the circumstances. Since a criminal conviction requires unanimity among jurors that a defendant is guilty beyond a reasonable doubt, an increased willingness to prosecute officers is unlikely to have much impact on the incidence of deadly force.

Criminologist Lawrence Sherman (2018, p. 434) has argued that, in addition to focusing on the officer -directly responsible, police homicides should shine a spotlight on “the complex organizational processes that recruited, hired, trained, supervised, disciplined, assigned, and dispatched the shooter before anyone faced a split-second decision to shoot.” In his view, policing is a system with interactive complexity and tight coupling, much like air traffic control, which similarly leaves little room for error and can give rise to catastrophic failures. He notes that “historical evidence from other complex systems suggests that their collateral death rates decline more substantially by re-engineering their social and technical systems than by increasing the certainty, speed, or severity of punishment” (Sherman 2018, p. 422).

Reengineering Policing

How might policing be reengineered to achieve a substantial reduction in the use of deadly force? And can this be done through piecemeal reform, or does it require agencies to be reconstituted root and branch?

To begin with, training needs to focus not only on how an officer responds to threats but on how to prevent interactions from evolving in a manner that results in a dangerous situation. Attempting deescalation, calling for backup, and assessing conditions from a distance can all help.

Patience is especially appropriate when a suspect shows signs of mental illness or is armed with a knife. In many jurisdictions, police are trained to shoot individuals brandishing a knife if they approach to within 21 feet. There is no scientific basis for this rule. For instance, no American officer was killed with a visible knife over the period 2008–13 (Zimring 2017). Meanwhile, more than a hundred civilians with knives are killed by police annually. They can be disarmed without recourse to lethal force.

Even if a shooting is justified under the legal standard, fewer shots will result in fewer fatalities. Police can be trained to fire one or two shots and reassess the situation. In 1991, 41 shots were fired by four plainclothes officers at Amadou Diallo, who was unarmed and innocent of any crime. And after a shooting, officers need not simply wait for an ambulance to arrive. They can offer medical assistance, including the application of hemostatic bandages, and transport victims to hospitals in their own vehicles.

More generally, police homicides could be investigated not just to determine individual culpability but to examine procedures and training in much the same manner as regulators respond to airplane crashes, or hospitals to preventable deaths, or the military to battle-field accidents, with a view to identifying all contributing factors so that such incidents are less likely to be repeated.

Police agencies vary enormously in the propensity to use force. Consider, for instance, those in Phoenix and Philadelphia, cities of comparable size. Over 2013–18, Phoenix Police Department officers were responsible for 103 civilian deaths, while Philadelphia officers were responsible for 34. In fact, the number of police homicides in Phoenix over this period even exceeded the 61 fatalities in New York City, which has five times the population.

Differences across these cities in the rate of violent crime per officer, or the number of officers per capita, are too small to account for such disparities in the use of deadly force. Though hard to quantify and measure, differences in organization culture are a significant factor. Rather than bad apples, the problem may be one of bad orchards (O’Flaherty and Sethi 2019). In this case, agencies with the most serious problems may be the most reluctant to change course, and attempts at piecemeal reform will be ineffective.

But neither bad agencies nor good practices can be reliably identified if police homicides are not accurately counted. Mayors and departments can’t be held responsible unless somebody knows how many police homicides are occurring, what the trend is, and how it compares with good practice. Right now, there is no official count of police homicides, or even an official definition. A number of media outlets and crowd-sourced projects have admirably helped to fill the gap, but they don’t have agreed-upon definitions or a regular source of funding.

If you can’t count, you can’t manage. If Black lives matter, then Black deaths—and the deaths of other police homicide victims—must be properly counted.


Baldwin J. 1966. A report from occupied territory. The Nation, Jul 11.

Kennedy M. 2016. Head of police chiefs group apologizes for “historical mistreatment” of minorities. NPR, Oct 18.

Kennedy R. 1997. Race, Crime, and the Law. New York: Random House.

O’Flaherty B, Sethi R. 2019. Shadows of Doubt: Stereotypes, Crime, and the Pursuit of Justice. Cambridge MA: Harvard University Press.

Sherman LW. 2018. Reducing fatal police shootings as system crashes: Research, theory, and practice. Annual Review of Criminology 2018(1):421–49.

Zimring FE. 2017. When Police Kill. Cambridge MA: Harvard University Press.

About the Author:Brendan O’Flaherty is a professor of economics at Columbia University. Rajiv Sethi is a professor of economics at Barnard College, Columbia University and an external professor at the Santa Fe Institute.