Afew months ago, New York Times columnist Pamela Paul made an argument that the police should aggressively crack down on subway fare evasion. To explain why this was necessary, she cited "broken windows theory," which she said progressives refuse to admit "works." She explained that allowing minor crimes "invites graver forms of crime," which is why we need to make sure laws against seemingly minor crimes are enforced. This is the core of the argument made in The Atlantic in 1982 by two political scientists, who argued that when a community allows small offenses (like broken windows) to go unpunished, soon the whole place goes to hell in a handbasket.

But the broken windows theory was a fraud. The writers of the original article did not produce evidence that it was true, and indeed there hasn't been evidence produced since to show that it's true. Joining us today is Bernard Harcourt of Columbia Law School, who in 2004 wrote the first book critical of broken windows policing, The Illusion of Order: The False Promise of Broken Windows Policing. At the time the book was written, "broken windows" was credited with having produced major crime reductions across the country. Today Prof. Harcourt joins to explain how this theory became so popular and ended up serving as the intellectual foundation for outrageously harmful practices like "stop and frisk."

Nathan J. Robinson 

Am I right that you would have hoped that your book had slipped into irrelevance by now?

Bernard Harcourt 

I would have thought, yes. It was a kind of full-blown critique of this mysterious theory, the broken windows theory, 20 years ago. And I would have thought that by now we wouldn't need to return to these issues because the theory had been buried and done with. But it seems to have a zombie life of its own. 

Robinson 

You mentioned the zombie life. We saw the theory bubbling to the surface most recently on September 5 in the New York Times in an op-ed by Pamela Paul, The Solution to New York's Transit Problem Is So Obvious Nobody Wants to Hear It, in which she calls for a massive police crackdown on fare jumping in the New York subway. And supporting her call for a massive police crackdown, she says, “Broken windows theory, as outlined by the social scientist James Q. Wilson and the criminology professor George Kelling in a 1982 essay in The Atlantic, holds that when minor or lesser kinds of disorder become more apparent, it invites graver forms of crime.” And then she says that “progressives are still loath to admit that broken windows policing works.” So we're going to talk today about broken windows.

Harcourt 

It’s surprising that passed the New York Times fact check.

Robinson 

I don't know if they do a fact check on the op-ed. 

Harcourt 

Listen, they do on opinions. I've written a few for them. The last one was on Alabama's new nitrogen gas asphyxiation method of execution, and they did an extraordinarily tough fact check. So I'm astounded that that sentence, the idea that the broken windows theory works, would be a factual statement. That actually should not have gotten through the fact check.

Robinson 

Well, I think the fact that it can still be stated as a "fact" speaks to the extraordinary success that this theory has had. In fact, it's hard for me to think of other examples where a theory that was published in a magazine, not even in a scholarly journal, but in a popular magazine became the most influential theory in American policing and criminology ever. Could you tell us first about just how influential this theory became?

Harcourt 

Yes, it is remarkable. What's remarkable is that the original article, which dated back to 1982 and was written by James Q. Wilson and George Kelling, was a narrative article. It was a story article. It wasn't a social science article. It didn't have social science facts to back it up. And in fact, when you read it, it's a fascinating short article that's all about the story of what happens to a neighborhood over time, this idea that the fences start breaking, the grass starts growing—there's a lack of upkeep, and eventually, tiny, minor forms of disorder start surfacing, and then all of a sudden, there's major crime in the neighborhood. It really was a story more than a social science hypothesis that had been tested with evidence.

It kind of played on our imagination about how neighborhoods might deteriorate. Now, it also played on a lot of racial stereotypes and moral stereotypes. In other words, James Q. Wilson, of course, was a brilliant political theorist, not just a public policy thinker. He was a brilliant political theorist and was drawing on the thought of his own mentor, Edward Banfield, who had this idea that there are morally backward societies and morally backward people. And in a way, this was building on this idea that there would be something like moral backwardness that would actually lead to serious crime, that it would have these really dangerous implications. Now, the theory was dormant for a long time, really, until the early 1990s when Rudy Giuliani picked it up with his first police commissioner, Bill Bratton, who would become the first police commissioner under Rudy Giuliani, and really ran on this theory of broken windows policing.

He called it the "quality of life" initiative. It came under different flavors, but the idea was, we need to attack things like turnstile jumping, minor disorder in neighborhoods, drug use, sex work, but also abandoned buildings and whatnot. The idea was, we need to pay attention to these minor forms of disorder because they're going to result in serious crimes: homicides, murders, rapes, and robberies. He pretty much ran on that platform, and he was extraordinary with the media. So, once he came into office, and once his police chief, Bill Bratton, seized on this idea of broken windows policing, they went kind of gangbusters on that. There was a massive drop in crime throughout America. But Giuliani, who was the master of the media, was able to kind of—

Robinson 

Back then. The younger people might not remember when Rudy Giuliani was impressive.

Harcourt 

Yes, back then. Back then, he was at the top of his game, really. And he would seize credit for the crime drop, which was happening in New York, and it was happening everywhere else in the country. He would seize credit and would claim that it was due to broken windows policing. So it was that effect. And then, of course, Malcolm Gladwell played a big role with his book Tipping Point, which was a New York Times bestseller. And one of the central examples that Gladwell used in his book Tipping Point was the broken windows theory. He was kind of trying to interpret it for the layperson, this idea that there was a tipping point, that if you had a little disorder in the neighborhood, at some point, eventually it's going to pass the tipping point and turn into murders, robberies, assaults, and rapes.

Robinson 

And so by the point, by the late '90s and early 2000s, as you document in your book, the media is referring to the now famous broken windows essay as the “bible of policing”—the blueprint for community policing. And it was stated as a fact, just as Pamela Paul stated that broken windows has helped reduce crime in cities around the nation and has had undeniable successes. The American Bar Association Journal was saying there's little dispute that the theory works, that it has sparked, as the Christian Science Monitor put it, “a revolution in American policing.” You just quote headline after headline about how this undeniably changed American policing, as if we'd figured out how to stop crime.

Harcourt 

Right. And you have to understand, of course, at the time, we had mass incarceration, which had started in this country about 1973. People need to understand that we were in a context here where a lot of people were trying to figure out how to fight crime, and one of the major ways in which most states were addressing crime was through massive arrests and incarceration programs, which resulted in extraordinary numbers of persons going to prison and jails. So in ordinary terms, we used to have about 150,000 persons in jails and prisons, constant from like 1920s to the 1960s, and starting in 1973, we get mass incarceration. By 2008 we've got 2,000,300 people behind bars, predominantly young African American men. And you've got a rate of incarceration in this country that reaches, in 2008, one percent of the adult population.

So you had a very Draconian policy of imprisonment, starting in the late 1960s when crime became a big issue, and with Richard Nixon, that reached this crescendo of incarceration. And so everybody on the progressive side was trying to find a different answer. And so at the time, conservatives are pushing for imprisonment and for broken windows policing, but progressives are looking for something—folks like Bill Clinton grabbed on to broken windows policing as well, because it's viewed on the progressive side as an alternative to mass incarceration. So basically, everybody's coming together. It's like a kumbaya. You've got the conservatives, like James Q., who was pretty conservative. You got the progressives who love this idea because it means we're not going to have to imprison everybody. And everybody's coming together. It's like, this is the manna from heaven—this thing drops from nowhere. And you've got Giuliani, who's touting it like a mad man because, of course, reducing crime is the big way to get elected, and everybody's on board. 

Robinson

Yes, it is quite extraordinary. It appeals to conservatives' desire to eliminate disorder and troublesome people, and liberals’ desire, as you say, that maybe we could just go after the squeegee men and graffiti, and if we could keep the windows from getting broken, then we can have an alternative to mass incarceration. It serves both their interests so well that nobody bothers to go back to check the article for whether there was any actual evidence in it. And I went back and read this article recently. I already read this thing—I remember being excited in college, and I hadn't noticed the first time I read it the gaping hole where the evidence should be. As you say, they tell this story where a piece of property is abandoned, and then litter accumulates, and then people start drinking, and then people start getting into fights, and then eventually someone's murdered. This is a story about a hypothetical fictitious city, but if you look for the proof that the piece of property being abandoned in the real world actually initiates a causal sequence with a murder at the end of it, they cite this one experiment by Philip Zimbardo. I think they completely misstate the facts of that experiment.

Harcourt 

Right. They had no evidence. It was really just this idea. And actually, what's remarkable is, in the article, James Q. said, this is just a hypothesis. This is just an idea. Actually, we don't have the proof for it, but that was kind of buried in the article. And the one thing they talked about was that Zimbardo experiment, which was they had left a car in an abandoned area, and after a few days, they broke a window of the car. And then after that, supposedly, in the experiment, people went in and started to steal things from the car and rip it apart and whatnot. But actually, in the study itself, it didn't really confirm major crimes like homicides or rapes or assaults. It was that minor disorder might trigger minor disorder or something like that. It wasn't a theory that these forms of disorderliness—which heaven knows what are they exactly?

For me, that became a big issue because it was this assumption that we all know what disorder is, which is not the case. We can talk about that. And of course, if you just use common sense, you would start to think, well, wait a minute—even if we knew what disorder was—maybe it's not disorder that causes homicides and rapes. Maybe both have similar antecedent conditions. So maybe a neighborhood that has a lot of disorder, defined by James Q. here, might also have a lot of violent crime in it, and maybe these things come together. Maybe there was a common cause that caused them both, but the idea that one caused the other, there really was no evidence for that, nor was there any evidence that police reducing the disorder would actually reduce serious crime as well.

Robinson 

One of the things that you do in your book is go through some studies purported to confirm the broken windows theory. You pointed out that there was a kind of methodological mistake in finding correlations without considering whether something could have caused both. But I just wanted to dwell on that Zimbardo experiment. Looking at the original Zimbardo experiment, and in fact, in the article, they misrepresented it. So the way they portray it is that Zimbardo left a car in the Bronx, which is a poor neighborhood, and people smashed up the car very quickly. So then he went to a rich neighborhood of Palo Alto, and nothing happened to the car until he smashed a window, at which point people descended on the car and started breaking it. That's how they portray the experiment. And they use that to say, well, if you smash just one window, you initiate a cycle that will, in rich bourgeois Palo Alto, turn everyone into barbarians. The thin veneer of civilization will crack, and things will go to hell in a handbasket. But in fact, Zimbardo and his students, he says, left the car and broke the window. Nothing happened in Palo Alto. He then took the car to the Stanford campus, and he and his students began smashing the car, and other students joined in with their professor in smashing the car. And what Zimbardo proved from the study, what he says he proves, is just that violence is kind of contagious and enjoyable, nothing to do with a community descending because of a broken window. And so the thin piece of empirical evidence that they had is, in fact, just as fraudulent, I think, and should have fallen foul of the Atlantic's fact-checkers in the same way that Pamela Paul's claim should have fallen foul of the New York Times' fact-checkers, but it didn't. They never checked the original study.

Harcourt 

Right. And for some reason, that replicates over time. So not only was there the problem with the Zimbardo study, but the next study, in a way, tries to prove broken windows in a statistical analysis of about 40 neighborhoods done by a sociologist, Wesley Skogan, where he shows that neighborhoods with high disorder are neighborhoods with high crime. That's all he shows in his study, and it's touted as being the proof. So when Giuliani picks it up from James Q. Wilson—James Q. Wilson was using Zimbardo—he picks it up in '93 or '94, and he then relies on the Wesley Skogan study, which is a little piece of Skogan’s book. It turns out that study also didn't pass the fact-checkers.

So I actually got my hands on the data in that study, and it was the messiest data I'd ever seen. But in the study, there were actually five crimes that they had tried to identify to see if they were linked to major crimes. One of them was burglary, one of them was small robberies, one of them was sex work, assault, etc. Of those five, there was only one of the five that went in the right direction for the study, and it turned out that was the only one that they used. So they had actually discarded four other small time crimes because they didn't come to the right results.

One of them, actually, which is really interesting in this context, was sex work. Actually, red light districts tend to be more orderly, in part because commercial sex operators want to have a neighborhood that feels sufficiently safe that people are going to go there and drop their pants. So actually, there's an interesting way in which that form of "disorder," which is commercial sex, is actually tied to more institutionalized orderliness. But in any event, that data was simply eliminated from the study because it didn't go with the right result, and there were just all kinds of problems with that study, which had been touted as being the study that proved the broken windows theory. In fact, it didn't, and I spent a good chapter of the book going through that study carefully and picking it apart. 

Robinson

It's extraordinary. You quote at the beginning that Giuliani and Bratton, I think in a policy paper, said Wesley Skogan has found that disorder is the first step in the downward spiral of urban decay. And then you go through and realize that he hasn't found that at all. In fact, it's kind of academic dishonesty. It's bad. It's really a kind of intellectual malpractice to discard the pieces of your data that contradict the theory.

Harcourt

Not only that. There were so many problems with this study, but maybe just to take a step back, a quantitative study is going to tell you a static—in other words, it's a one-time shot. It's one moment about neighborhoods that have a lot of disorder and have a lot of crime, or neighborhoods that have little disorder, little crime, or neighborhoods with some disorder and whatever. A one static snapshot of neighborhoods is never going to be able to validate a causal theory that operates over time. In other words, even if—and it doesn't—the study showed that those lined up perfectly so that only minor disorder neighborhoods had no major crime, and major disorder neighborhoods had major crime, even if it had lined up perfectly across different crime variables, even assuming it had been perfect, it doesn't prove anything. It still doesn't prove anything. Because it doesn't show that one causes the other or that the other causes the other. It's static. It's static. It's not telling you what the causal mechanism is!

Robinson 

Their whole thing was a story about something that occurs over time. You start with the broken window, you end up with a murder. 

Harcourt 

Right. You can't prove it by simply saying, this is a high crime neighborhood, and it's also got a lot of disorder. So one caused the other. It's not how you go about explaining a theory that has temporal causation in it. The study itself, even if it hadn't had all those defects, couldn't prove in any way or couldn't non-falsify the theory. 

Robinson 

So the correlations didn't consistently match up, but even if that correlation had consistently matched up, they're also consistent with a theory that, I don't know, poverty causes both disorder and crime, rather than the broken windows. And the reason this is important is that broken windows is a theory that unless you crack down on very small things, and in that they include things—and this is another remarkable thing, when you go back to the original article, they include things that are not illegal. So this category of disorder includes squeegee men—Giuliani really didn’t like the squeegee men in New York. Giuliani revived, I think, the anti-unlicensed dancing ordinance from the Prohibition era. It includes drug use, but it also includes loitering, just standing around in the wrong place. So let's get into a little bit of this idea of disorder, this thing that they said you had to eliminate if you didn't want rape and murder and robbery.

Harcourt 

I would say that from a kind of theoretical perspective, that's the most interesting question: what is order, actually, and what is disorder. Now, they had identified certain things. It was very interesting to see their little list. So, for instance, one of the things that they identified as being disorderly was hanging out on the stoops. Okay, something you do in certain neighborhoods in New York City, probably in New Orleans as well. People hanging out on the stoops was considered a form of disorder. Now what's interesting is, who does hang out on stoops? What neighborhoods have people hanging out on the steps? It turns out that it's a very cultural and racially identified practice. It's very common in what we used to call Hispanic neighborhoods in America—Dominican, Mexican American, or Puerto Rican neighborhoods in New York, where the people would hang out during the day on the steps and kind of stand there and watch the neighborhood.

Now, it turns out that actually is a form of social control. It's actually people in the neighborhood taking control in the neighborhood, and knowing who's in the neighborhood and who's not, identifying people who don't belong, and identifying the regulars. It's a form of social control in the sense that, actually, it's a way in which in some neighborhoods, people are trying to create order. Okay, but it didn't fit with the cultural norms of the James Q. Wilsons. And so all of a sudden, this form of behavior, which was a form of orderliness, was being described as disorderly, but in a way, it was identifying a minority community more than identifying disorder. And a lot of these forms of disorder, actually, when you start to kind of scratch at the surface, start to have racial correlations or ethnic correlations, and what we're talking about all of a sudden are Latino or Black neighborhoods, rather than "disorderly" neighborhoods.

And one of the most fascinating is some social scientists, Danielle Wallace and others—Raudenbush, Rob Sampson, and others—did some studies of how people identify disorder. This was in Chicago. They did this incredible—they videotaped neighborhoods, and they had people code and had people identify disorder, and it turned out that with the same amount of disorder in a neighborhood, people saw more disorder in Black neighborhoods than in white neighborhoods. So with the identical factual disorderliness, as defined by these metrics of disorder, they would identify more disorder in minority communities than in white communities, which goes a little bit with the way in which there's an ideological dimension to disorder. It's simply not the case that all of these forms of disorder are what everybody would call disorder.

Robinson

What's critically important is that they're talking about disorder as distinct from crime and harm, because things that even are crimes or harms don't necessarily fall in the Wilson-Kelling category of disorder. In fact, they even say that, well, the police might need to do some things that aren't strictly legal in order to eliminate disorder, which is anti-social behavior. And you point out that corporate crime, like wage theft, wouldn't fall into the category of disorder, even though that's crime and harm. But Wilson would cite things like eccentric clothes and unusual hair, I think he says at one point, might be some signs of disorder.

Harcourt

In fact, the word he used was conk rags. That's what James Q. Wilson referred to as being kind of an indicator that things were spiraling, which was a racially identified way of referring to a particular way of wearing your hair. Probably the most interesting and tricky thing about disorder is in the broken windows essay, they talk about the fact that cops might have to use a little disorder themselves, in other words, and that's what's so fascinating. So what happened in New York City is they started their broken windows policing in '94, and two to four years later, crime has gone down. Crime had gone down in New York City, and had gone down even more in a place like Los Angeles, which didn't even have broken windows policing, and in other cities throughout the country, but it's gone down. But what's gone up in New York by 60 percent are complaints of police misconduct. And so you've got a situation in New York where crime is going down, but complaints of police misconduct are going up. And the question is, well, is that order or is that disorder? And right there you've got the whole problem. You've got the whole kind of made-up-ness of the theory. Because, wait a minute, if complaints of police brutality are going up 60 percent in New York City, maybe disorder has gone up? And there's the trick. How do you define disorder?

Well, of course, Giuliani and Bratton were not defining it by complaints of police brutality. They were identifying certain other things. And of course, maybe we would say, well, disorderliness has actually gone up. So what you end up realizing is that disorder and order are two of the most malleable terms we have in society.

Robinson 

One person's order is another person's disorder. 

Harcourt 

Exactly right. 

Robinson 

Hence, the title of the book, The Illusion of Order. You're creating a situation that we call order because you harassed and arrested a bunch of squeegee men. In doing so, you may have used illegal and unconstitutional violence, but we're going to put that in the bucket of order. And the thing is, it becomes clear in other points in their essay, They talk about this police officer in Newark and how he used to do community policing, and often it entailed basically taking the side of the respectable citizens against those who are kind of mentally put in the bucket of the disorderly people. That's drunk people, but also, they say that he would always take the side of the business owner in a dispute with the customer. So it's clear that order means whether a certain segment of the population is happy with the way things are.

Harcourt 

Exactly, and not surprisingly, it usually lines up with race and class and ethnicity. That's one of the things about this that is so disturbing. It created in our minds this association of disorderly, often with minorities, in New York City. And you got to understand, another reason that it became so popular was that for a lot of liberal New Yorkers at the time, the panhandler, the guy on the corner asking for money, the homeless person on the street, it was very disturbing. It's not what you want to see when you're walking home from the office. You might be liberal, you feel bad, you don't know what to do. Well, all of a sudden, that person has become the problem that you have to get rid of. And so you can kind of try to sweep those people away with police sweeps, and you're no longer going to have that moral nuisance—you're not going to have to feel bad about the panhandler or the homeless person, about not giving them money, or about walking by and not doing anything because all sudden, they've become the problem. So, the police have to take them away.

And I think in New York, that played a big role as well in making it so popular. It kind of turned a nuisance into something that the police had to get rid of. And so a lot was swept under the rug that way. Now, what's also particularly interesting about this theory is that it came at a particular time, the mid-1990s, when New York was going through forms of gentrification and real estate development that, in part, were transforming New York, but that needed some justifications. And the broken windows theory was the perfect justification, so Times Square was completely redesigned at this time. Now it's important to understand, of course, those plans for Times Square had started in the 1970s. Those plans that had been laid down in 1974 under Mayor John Lindsay and other mayors were for a radical redevelopment of Times Square, which used to be the red light district in New York and would become kind of the Disneyland that it is today, with Nike shops and pedestrians and whatnot. But it was a time when that kind of real estate redevelopment and redistribution needed to be justified, not simply as, well, we're going to build new buildings and a lot of real estate developers will make a lot of money. It also needed to be justified politically, and the broken windows theory served that role perfectly by saying, look, we're getting rid of the triple-X adult stores down here in Time Square, and we will reduce crime as a result. So it also served as this mask for real estate redevelopment and large-scale wealth redistribution.

Robinson 

Because you could say, we have to do it. There's no argument because unless you deal with disorder, you are going to have terrible violence. And so we have no alternative. There's not really a debate about whether the squeegees should be legal. Yes, it's annoying to come and have someone spray your car when you didn't want them to, they leave a mess, and then they want money from you. But not only is it annoying, but we need to arrest the squeegee man because if we don't, it's literally a slippery slope down towards complete chaos and dysfunction, which the more I think about it, is a theory that I would want to see extraordinary empirical support for. Because I know the story sounds compelling in one way, but also it sounds really counterintuitive in another. That is to say, well, really? If we don't crack down on illegal dancing, that's what police should be spending their time on, rather than actual murder, violence?

Harcourt 

Exactly, what kind of resources are we putting into this? Where are we putting our resources? Now, those would have been all the questions that should have been asked, but none were really, in part, because of the real grip that Giuliani had on the media. You have to understand this. In '94, '95, '96, and all the way to 2001 when Giuliani became America's mayor—he really is at his height, tragically, following the 9/11 terrorist attack on the Twin Towers. And in fact, it's really at that time that he, again, is touting broken windows. He's actually running for election at that time of 9/11. He was trying to extend his mayoral mandate, actually, on the basis of the great job that he's done as mayor of New York and seeing us through the Twin Towers.

Robinson 

There's also an element that people like a counterintuitive idea. You said it appears in the Malcolm Gladwell book. Like when they pitch this to the Atlantic: what if? It's one of those, everything you know about policing is wrong, and there's an incentive that publishers have also because we need to have original ideas. The old idea is that police should just do the boring work of going out and investigating murder. But what if, actually, the broken window—and it's an easy, quick fix, too.

Harcourt 

Right. It had everything going for it, except the fact—

Robinson 

Except any empirical support whatsoever.

Harcourt 

Except any empirical support. And so the social science studies that come in to try and corroborate it are just fraught. George Kelling does a few later on, but they just don't hold water. And then, of course, just the crime statistics don't hold water either. Because, although it's true that crime came down sharply in New York during the time 1994-98, it's also true that it fell sharply in other cities. It fell dramatically across the United States. There was just a remarkable crime drop during that period, starting in '91 actually—in New York, it started in '91 under Dinkins, and it wasn't as a result of the broken windows policing. Crime kind of already started to go down because the crack epidemic had reached its peak in 1991.

So you see it across the country, but you also see it in towns and cities and municipalities that do the exact opposite of broken windows policing, that have a kind of like, let's reduce arrests, approach. You see this in San Diego, for instance, that had a very different policy. Of course, Los Angeles at this time had no functional policing. This was between Rodney King at the beginning of the '90s and the Rampart scandal at the end of the '90s. The LAPD was completely dysfunctional during this time, was not engaged in broken windows policing, and they saw equivalent drops and sometimes greater drops in robberies and whatnot. So when you just compare what's going on with overall crime as well, there was no real indication that it was working, but Giuliani had the pulpit and the media at his fingers, and he was able to say, the drop in crime here is broken windows. 

Robinson 

Well, that raises the question, what do we know about effective ways to reduce violence in society if broken windows is a discredited theory? 

Harcourt 

Right. So this is a big source of debate, and I think we tested a lot of different policing approaches. We've tested different interventions, particularly drug treatment interventions, trying to reduce dependency on drugs, trying to deal with different drug trade activities. Legalizing licensing has been a way of addressing some of these things. So, for instance, the broken windows goes against sex work. Well, there are certain places that have been working on having licensing of commercial sex. That's the case in some counties in Nevada. And so there have been a lot of other efforts that we've seen that have been able to address things that we would like to see addressed, like addiction behaviors, and they've proven to be much more successful in addressing those than kind of cracking down on them.

The other thing is that the crackdowns have often caused unnecessary violence and harm, and I think we're experiencing that right now in New York. There are these reports of a recent police killing in New York involving someone who ran a turnstile and who was chased down and shot fatally. There are many other ways. Even on turnstile jumping, if it is a problem, you can construct turnstiles that you can't jump. So there are just so many other ways to deal with this than through police action. And that was the point in part. Another big question is, even if you believe in the broken windows theory, that doesn't mean that the answer is aggressive arrests for misdemeanors. In other words, if graffiti is the problem, you can have the sanitation department deal with it. If turnstile jumping is the problem, you can have the MTA deal with it. You don't have to throw the cops at every problem. And that was another big source of it.

Robinson 

That's right. It actually stands separately as a causation theory of how violence occurs from the answer of what you do about it. There's this kind of assumption, this kind of leap to, well, then you definitely have to arrest the panhandlers. You need the agents of the state to go drag these people off the street. But you could do a million things.

Harcourt 

Right. You've basically got two theories in one, two explanations in one: You've got the explanation that disorder causes crime, but then you also have this separate explanation or theory that you throw the police at the disorder and arrest people, and that's going to reduce crime. That's another whole piece of it that we haven't even talked that much about—a whole other piece of this theory that there isn't good evidence for that.

Robinson 

It has done a lot of harm, the theory that you have to stop the disorder, which is minor things that may not even be illegal and that are often proxies for poor people, essentially. And when you go back to this guy Banfield's writings, it's even more explicit in there that the things poor people do are the problem with the city. Adopting that has caused so many young Black men to be stopped and frisked in New York, and not just men. People have their constitutional rights violated, with so many violent police encounters. I assume the harm of this one theory were a big part of why you were motivated enough by it to write an entire book on why it was wrong.

Harcourt 

Yes, there was a lot going on in New York at the time. There was the Diallo assault, terrible stuff going on. So it just had a way of feeding a kind of aggressiveness in policing that obviously can often go off the rails. And I think we saw it then. I think we just saw it in New York again a couple of days ago.

Robinson 

I think that this book might be due for a 20th anniversary edition. I was very depressed reading it.

Harcourt 

An updated preface or something. 

Robinson 

To look at how long ago, because at the time you wrote this, there hadn't been much criticism. This was kind of the first—I think this was the first book.

Harcourt 

There was practically no criticism, and there's been very little criticism. This was the first criticism of the theory, and for many years, the only. 

Robinson 

There’s been a lot of understanding now that mass incarceration is a horror. But as you point out, there's this temptation to see this, whether rebranded a little bit, as community policing for the good liberals who don't like the idea of being associated with disorder. But as you say, we really need to avoid the temptation to revive this theory. Because you go through the evidence and there's just nothing for it. And I assume nothing has come out since that has changed your mind.

Harcourt 

No. Since then, I've done a number of other empirical studies testing it—none of them panned out. I'd say that the social science consensus is pretty much that it's not a disorder-causes-crime phenomenon. That just isn't right. Now, having a lot of police, which is the other aspect of it, can have some deterrent effects and costs, as we talked about. But in terms of, is it broken windows policing? No, that didn't pass it.

Transcript edited by Patrick Farnsworth.