Broken Windows Turns Forty-Plus, and What Wilson and Kelling Actually Wrote

The most quoted essay in the history of American policing is also one of the least read. People who love it and people who loathe it cite it constantly, and both camps tend to describe a document that never existed: a manifesto for arresting everyone who jaywalks, or a license for stopping young men on the street by the thousands. Neither is what two professors wrote for a magazine in the spring of 1982.

The essay is more than forty years old now, which is old enough to read as history. So let us read it as history: what the authors saw, what they argued, what New York did with the idea, what the courts said about the part that went wrong, and what the record supports. A conservative should want to know, because the theory is routinely attributed to us, and because getting a theory right is the first step toward judging it fairly.

The essay and the authors

"Broken Windows: The Police and Neighborhood Safety" appeared in The Atlantic Monthly in March 1982. Its authors were James Q. Wilson, a political scientist at Harvard who had written widely on crime and on government, and George L. Kelling, a criminologist who had studied police patrol and worked on a major experiment in Newark, New Jersey. Wilson died in 2012 and Kelling in 2019.

The essay is a piece of argument, not an experiment. It offers a theory, supports it with observation and a few studies, and draws policy conclusions. It was written for a general readership, and that is why it survived. It is short and vivid and it has an image.

The Newark experiment that inspired it

The idea began with a puzzle. In the 1970s, in Newark, the Police Foundation, a research organization, tested foot patrol. In some neighborhoods the department put officers on foot, walking beats, rather than leaving them in cars. The Newark Foot Patrol Experiment, which ran from 1978 and was reported in 1981, found something that surprised the researchers. Foot patrol did not reduce the measured crime rate. But residents in the foot-patrol areas felt safer, were less afraid of crime, and thought better of the police, and the officers themselves reported greater satisfaction.

Kelling and Wilson wondered why. If the officers had not reduced crime, why were people less afraid? Their answer was the heart of the essay. The officer on foot dealt with disorder: the panhandler who was aggressive, the drunk in the doorway, the group of rowdy teenagers, the abandoned building. These were not serious crimes, but they were what the residents feared and what shaped how they used their streets. The officer did not make the neighborhood crime-free; he made it feel governed.

What the essay argued

Here is the argument in paraphrase, in five steps.

1. Disorder is a signal. A building with one broken window that is left unrepaired tells passersby that no one is looking after it. Sooner or later more windows get broken, because the first one proved that no one cares. The authors cited an experiment by the psychologist Philip Zimbardo in 1969, in which a car left on the street in the Bronx was stripped within hours and a car left in Palo Alto, California, sat untouched for over a week until the researcher himself smashed it, at which point others joined in. Later writers have disputed how clean that comparison was.
2. The signal changes behavior. In a neighborhood where public disorder goes unchecked, law-abiding residents withdraw: they stay inside, avoid the street, and stop watching each other's property. Informal social controls weaken. People who are inclined to offend learn that no one will intervene.
3. Serious crime follows. The authors proposed that, in that setting, serious crime was more likely to take root. They described this as a possible sequence and not a law of nature. They were careful to say that it was a hypothesis about how disorder and crime might be linked, and that evidence on the causal link was thin.
4. The police have a role in order, not only in arrest. Police departments, the authors argued, had drifted from the old role of the patrol officer, a figure who knew his beat and enforced the unwritten rules of a community, toward a role of responding to calls and making arrests for serious crime. The essay said the police should return to the business of order maintenance: helping neighborhoods enforce their own norms about public behavior.
5. Order maintenance must be exercised with judgment. This is the part that gets lost. The authors insisted that the officer's authority was to support the community's standards, and that he must be trained and supervised to avoid becoming the agent of prejudice or of arbitrary power. They worried openly about the danger that police discretion could be used against people simply for being unconventional or for belonging to a minority group, and said that the problem of controlling it was serious and unsolved.

Notice what the essay does not say. It does not say "arrest people for minor offenses as a matter of course." It does not say "stop and search large numbers of people." It does not mention quotas. It talks about officers who know their neighborhoods, about informal rules, and about the community's own definition of order. The leap from that to a program of mass enforcement of minor offenses was made by others, later.

New York, 1990 to 1994

The theory entered practice in New York, and the story is worth telling in order.

In the late 1980s New York City's subway was widely regarded as a place of disorder and crime. The transit system had been trying to deal with graffiti on subway cars since the 1980s, with a program of cleaning cars quickly so that vandalized cars did not stay in service. In 1990 William Bratton became chief of the New York City Transit Police. He applied the broken windows idea in a concrete form: the Transit Police went after fare evasion, which they treated as both a minor offense and a source of information, since many people stopped for jumping a turnstile turned out to be carrying weapons or to be wanted on warrants. They also went after aggressive begging and other conduct that made riders feel unsafe. Transit crime fell over the following years.

In January 1994 Rudolph Giuliani became mayor and appointed Bratton as police commissioner. Bratton brought the approach to the NYPD as a whole. That year the department also adopted CompStat, a management system, associated with Deputy Commissioner Jack Maple, in which commanders were required to appear regularly before the department's leadership with up-to-date crime statistics for their precincts and a plan to reduce crime. CompStat is often confused with broken windows, but it is a separate idea about accountability and data: the commander is responsible for results in his precinct, and he is expected to know what is happening there and to act on it.

Crime in New York City fell sharply through the 1990s, and the city's decline was among the largest in the country. That is a fact about the period. What is disputed is why.

The part that went wrong

Broken windows and stop-and-frisk are two different things, and the confusion of the two has caused more trouble than any other error in this debate.

Order maintenance is the enforcement of public-order norms: addressing aggressive panhandling, public drinking, vandalism, fare evasion and similar conduct, ideally through the officer's presence and discretion, with arrest as a last resort.

Stop-and-frisk is a different tool, grounded in a Supreme Court decision, Terry v. Ohio (1968), which held that an officer may briefly stop a person on reasonable suspicion, based on specific facts, that the person is involved in crime, and may pat down the outer clothing if he reasonably believes the person is armed and dangerous. It is a Fourth Amendment doctrine about suspicion of crime. It is not a doctrine about disorder.

In the 2000s the NYPD's use of street stops grew enormously. Whatever the department's intentions, the practice became the subject of a federal lawsuit, Floyd v. City of New York. In August 2013, Judge Shira Scheindlin of the Southern District of New York ruled that the way the NYPD had been conducting stops violated the Fourth Amendment, because many stops lacked reasonable suspicion, and also the Fourteenth Amendment's guarantee of equal protection, because the department had implemented its stops in a way that targeted people by race. She ordered reforms, including the appointment of an outside monitor. The Second Circuit stayed her orders and reassigned the case, a controversy of its own, and the city, under a new administration, later dropped its appeal and accepted reforms.

I want to be careful about what that ruling means, and what it does not. It was a trial-court decision, and it concerned one department's practice of stops, not the underlying theory of order maintenance. But it is consistent with the essay's own warning. The authors of the essay had said that the exercise of police discretion over minor matters was dangerous if not carefully constrained. A program that evaluates officers on the number of stops they make, rather than on whether their stops are justified, converts the officer's discretion into a numbers game. That is the opposite of the community-rooted judgment the essay described, and defenders of the theory, including Kelling, have argued that mass stop-and-frisk was a misapplication of it.

The conservative lesson is the one we apply to every government program: the quality of a policy depends on the quality of its administration. A good idea carried out by a quota system turns into a bad one. And a conservative who cares about the Fourth Amendment has no reason to defend stops that lack reasonable suspicion. Probable cause and reasonable suspicion are not technicalities. They are the line the Founders drew between a free people and a police state.

What the evidence shows

Now the hard question. Did it work?

Here is the case for the theory. Disorder really does affect how people feel about their neighborhoods, which is what the Newark study found. Public order is a legitimate public good, and a city that cannot keep its subway or its parks usable has failed at a basic function. The best-designed disorder-policing programs target specific places and specific problems, with the officers working with residents and businesses, and some research reviews have found that this kind of approach is associated with crime reductions. A systematic review of studies by Anthony Braga and colleagues, published in 2015, found that disorder policing strategies that were community-oriented and problem-solving tended to show crime-reducing effects, while strategies built on aggressive misdemeanor arrests did not show the same effects. The review is public, and you should read it before you quote it.

And here is the case against. The theory's central causal claim, that visible disorder produces serious crime, has been hard to prove. A well-known study of Chicago neighborhoods by Robert Sampson and Stephen Raudenbush, published in 2004, argued that the causal chain from disorder to crime is weak, and that collective efficacy, the cohesion of residents and their shared expectations for social control, does more to explain differences in crime than visible disorder does. Critics also note that crime fell in many American cities in the 1990s, including ones that did not adopt broken windows policing, so the drop in New York cannot be credited to the theory alone. Economists and criminologists have proposed other explanations for the decline, from the growth in police numbers to changes in the drug markets to the economy to the incarceration rate, and none has been shown to account for it entirely. For the economics of police numbers, see What Economists Found When They Counted Police Officers and Crimes.

My reading of the record is this. The strong claim, that fixing minor disorder is the main reason crime falls, is not established. The modest claim, that police who deal with disorder in a targeted, community-based way can improve safety and the public's sense of safety, has reasonable support. And the claim that mass enforcement of minor offenses is an effective crime-fighting tool is the one with the weakest support of all. The authors never made that claim.

What a conservative should take from it

Three things.

First, order is a conservative value. Edmund Burke wrote of the "little platoons" of society, and a neighborhood is one of them. A community that can enforce its own norms, with the help of an officer it knows, is the kind of mediating institution conservatives prize. The essay's best idea was to give the officer a role in supporting that, not replacing it.

Second, discretion has to be disciplined. A power that is vague enough to be used at will is a power that will be abused. The essay saw that. The remedy is training, supervision, written standards and measurement that rewards good judgment rather than volume. That is exactly the sort of accountability CompStat was meant to supply, and it is why the data a department keeps should include the justification for stops, not merely the count.

Third, the officer on foot is worth the money. The Newark experiment's finding that foot patrol made people feel safer and improved relations with the police still matters. A city that wants order needs officers on the street who know the neighborhood, and that takes headcount, a theme I take up in Why Cities Cannot Hire Enough Police Officers, and What Would Fix It. Cities that talk about order maintenance while cutting their police forces are describing a policy they have not funded, a point that also applies to the experiments discussed in Defund the Police Was a Slogan Cities Tried and Then Quietly Walked Back.

For how the courts treat the officer's decision in the moment of force, which is a separate matter from order maintenance, see Graham v. Connor and the Officer's View in a Split Second.

The case in brief

1. The 1982 Atlantic essay by James Q. Wilson and George Kelling grew out of the Newark foot-patrol experiment, which found that officers on foot made residents feel safer without clearly reducing measured crime.
2. The essay argued that unchecked disorder signals neglect, weakens informal social controls and may invite serious crime, and it urged police to support neighborhoods' own norms, with discretion constrained to avoid prejudice.
3. New York applied the idea in the Transit Police from 1990 and the NYPD from 1994, alongside CompStat, a separate accountability system, and crime fell sharply, though the causes are disputed.
4. Stop-and-frisk is a different doctrine, rooted in Terry v. Ohio, and the NYPD's use of it was held unconstitutional in Floyd v. City of New York (2013), a result consistent with the essay's own warning about unchecked discretion.
5. The record supports targeted, community-based attention to disorder, does not establish disorder as the main cause of crime, and gives little support to mass enforcement of minor offenses.

What you can do: read the original essay in The Atlantic's archive, then ask your local police department how it measures its officers' work on public order and what training governs stops.