Why the Sanders Campaign Is Right to Eschew "Broken Windows" Policing

After I read the detailed criminal-justice section recently published as part of Bernie Sanders’s 2020 campaign platform, I went and found the original 1982 Atlantic Magazine essay on “broken windows,” to figure out whether it was worth saying anything about the Sanders campaign’s full-throated rejection of the idea. (To quote: “And we must end harmful policing practices like racial profiling, stop and frisk, oppressive ‘broken windows’ policing, and the militarization of police forces — all of which actively undermine public safety and community trust in law enforcement.”)

My prior thinking, which I wanted to re-evaluate, is that broken windows as originally defined is not necessarily or automatically racist or “oppressive.” And it’s certainly not supposed to go hand-in-hand with militarization of police forces. Having re-read it, I still think that’s true, but I also think the Sanders campaign is right to say they want no part of it.

So first we have to talk about what George Kelling and James Wilson were originally talking about when they talked about a broken-windows approach to policing. I suspect a lot of people assume the phrase is synonymous with “stop-and-frisk,” a policy which gives officers wide leeway in stopping suspects and then itself became shorthand for the phenomenon of racist cops harassing people of color. (Here’s a 2016 NPR piece that explains how broken windows was linked to stop-and-frisk, mostly because the Bratton-led, Giuliani-backed NYPD talked about embracing the former and then embraced the latter. The article gives the impression that Kelling and Wilson were initially publicly okay with the association, though Kelling later regretted it.) Here’s the basic idea: say you have a building, or car, with a broken window. Not only does the broken window make looting the building or car logistically easier, it also sends a signal that whoever’s charged with protecting the building or car doesn’t have the energy, or the means, to defend the building or car; that the looting is more likely to go unpunished. Increase the number of broken windows on the street or neighborhood, and then the street or neighborhood risks becoming known as a place where the costs of crime are lower; and, whether crime actually goes up or not, that newfound reputation affects the residents of the street or neighborhood, making them more fearful and lowering their quality of life.

Broken-windows in its original form is thus place-based, which is why I’m talking about it in a planning blog. You go after marks of “disorder”, to use Wilson and Kelling’s word, not because by arresting the window-smasher (or graffiti artist, or homeless person passed out in the street) you increase the chances of catching a criminal who has also done violent crime. That’s what the NYPD claimed to have done after they began cracking down on subway violations, and that was the justification for stop-and-frisk. But it doesn’t necessarily follow from a place-based approach to broken-windows. You patch the window, or chase off or fine the graffiti artist, to send the signal that someone cares about the neighborhood, and that those who threaten its order will be punished.

Thus you don’t need arrests en masse to follow a broken-windows strategy. You don’t need to punish rule-breakers severely, because that’s not how you raise the cost of attempted crime. It’s not about throwing the graffiti artists in jail for long periods of time; it’s about cleaning up the subway cars, over and over. Kelling said as much, writing approvingly in 1991 of David Gunn’s “Clean Car” campaign for the NYCTA:

The Clean Car Program started by pulling graffiti-covered trains out of service, cleaning the cars, and sending them back out on the road. Police were assigned to ride fulltime on the first clean trains, and clean trains went into special protection yards. But the program went further; it guaranteed that the first “broken window” would not lay untended and lead to the next. Once a train was entered into the program and cleaned, it would never again be used while graffiti was on it. If a train was tagged by a graffitist, either it would be cleaned within two hours, or it would be removed from service. As a result, graffitists would never see their tags on clean trains again. They might be able to paint their tags over other graffiti on cars that were not yet entered into the program, but not on clean cars….

Early efforts to deal with graffiti assumed it was a law enforcement problem, so it was natural that the police deal with it. They did. They arrested graffitists—again and again. More trains got tagged, even as arrests increased. Meanwhile, other departments conducted “business as usual.” Gunn changed this. He understood that the graffiti problem was a complex mix of vandalism, poor maintenance, inadequate leadership, and lack of resolve. Through his close oversight of an interdepartmental task force charged with yearly goals—goals they exceeded every year—graffiti on trains was wiped out.

Here’s where broken-windows gets more complicated than just arresting people (or just arresting people of color). If we think of the subways as one place, it becomes clear that for the broken-windows approach to work, the enforcers need intimate knowledge of the place. Cars needed to be tracked, classified, cleaned immediately if graffiti’ed upon, and repeatedly inspected. The same for a neighborhood: why is that window broken? Was it broken yesterday? Who’s responsible for fixing it? Can they fix it? If not, do they need help, or motivation in the form of a threat of a code-violation? And so on.

Hence a lot of Kelling and Wilson’s article talks about neighborhood patrolling. They advocate that police officers patrol on foot, rather than by car, because people on the street will be understandably less eager to approach cars, and the car-bound officers are thus going to be warier of those who do approach cars. (Triple all this if the officers are already approaching the situation as if they’re soldiers in hostile territory.) Furthermore, they define the community as law-abiding, and the police officer as an aide, not the sole enforcer: “The essence of the police role in maintaining order is to reinforce the informal control mechanisms of the community itself.” The setup ought to be police-and-community versus disrupters, not police versus community, with disrupters included in the community.

You may be snorting and rolling your eyes at this point, believing the history of American policing is so drenched with racism that any setup that involves cops bonding with majority of-color communities is naïve at best. And your distrust is relevant. But put it aside for just a second: even if a hypothetical law-enforcement agency were to adopt a more place-based, community-trusting approach, that still wouldn’t solve the problem of mutual distrust and perceptions of racism.

Here’s what I mean. Suppose you have a majority-black neighborhood that suffers from the spatial signals of neglect, and decreased quality of life as a result. Suppose the local police force decides to apply a broken-windows approach to this particular neighborhood. Car broken down by the side of the road? Let’s get it towed quickly. Guy hanging out looking angry on the street corner? Talk to him sooner rather than later. Broken window? Knock on the door, and if there’s no answer, look the landlord up and give them a call. And so on.

If the strategy works the way Kelling and Wilson say it should work, then that police force will have improved the lives of people of color in that neighborhood. But: the people who got targeted are also more likely to be people of color, and vulnerable. Someone who might not have the money to fix their car now has to deal with it getting towed. Someone who might not have the wherewithal to do home repairs now has notice to hurry it up with the home repairs. The angry-looking fellow, who has a perfect right to stand on a street corner, may be further upset after the conversation—because who wants an armed cop to come up and start chatting, however benign the cop’s intentions?

Of course, “a cop” is different from “Officer Bennett, who sat with my sister for two hours and talked her through a panic attack.” A broken-windows approach could, in theory, work if there’s a level of mutual trust between individual community members and their public servants. But that kind of trust is hard to establish. The Sanders campaign platform includes expanding the availability of unarmed first responders who can work with, and address situations before, the police, which sounds like a good idea to me; but those first responders, too, would need to build trust over time. And that trust, even if it were established, wouldn’t necessarily be generalizable or communicable outside the neighborhood. If Officer Bennett and Angry Dude get into an argument, the neighbors might be all, “Thank God, someone needed to tell Angry Dude he was being a jerk,” but someone outside the neighborhood might watch Officer Bennett’s bodycam footage and, lacking the neighborhood knowledge, see the argument as another example of racist police harassment.

This isn’t even limited to policing. Pete Buttigeg was following a place-based approach with his housing policy focusing on dilapidated houses in South Bend. But he was criticized by black advocates, who argued that the dilapidated homes weren’t the property of “absentee” landlords, but people of color who bought or inherited the houses and lacked the necessary resources to fix the properties. Compare that to the approach in Mobile, Alabama, where addressing “blight” seems to have been a careful step-by-step process, including tracking down every potential claimant to a property the city wanted fixed, even if there were 120 claimants and Alabama law required talking to all of them. It could be that the mayor of Mobile got a more sympathetic reporter than Buttigeg did, but I suspect that if Buttigeg were trying to work quickly (“1,000 Houses in 1,000 Days”) then he probably couldn’t have given his team time to investigate every absentee-landlord issue even if they’d requested it and he’d admired them for requesting.

Long story short, a non-oppressive, locally sensitive broken-windows approach doesn’t scale. It certainly doesn’t scale to the federal level. My initial reaction to a lot of the Sanders campaign points was to wonder how the goals would be achieved at the national level, when so many of them seemed dependent on state policies. (I will also admit to some cynicism about the promise to fund public defenders more broadly, when 40-year-old existing mandates, like those for special education via the Individuals with Disabilities Education Act, remain underfunded. The Sanders campaign promises to get federal IDEA funding up to 50% of the total; my understanding is it’s rarely been funded at even 20%.) If you’re running for mayor of a smaller municipality, “We’re going to invest resources in policies that are locally sensitive and careful, and devote time and thought to balancing the due-process rights of individuals with the needs of the larger communities,” you’re still biting off quite a chunk, but at least it’s within the realm of theoretically possible. A presidential campaign can’t, and therefore shouldn’t, go that route. No one can fix broken windows from the White House.