There’s an active debate going on now in my city about the role of the police, and whether we’ve gone too far as a city in terms of the “broken windows” approach to policing. This approach argues that tolerance of low-level offenses – turnstile jumping, public drunkenness and disorderly conduct, defacement of property, etc. – creates an environment in which more serious criminality thrives, both because criminals believe nobody is watching and because law-abiding citizens do not populate the streets. A vigorous police presence maintaining order both serves as a direct deterrent to criminal activity and incidentally may result in the apprehension for lesser charges of criminals already wanted on more serious charges.

The case against the “broken windows” approach holds on the one hand that an aggressive approach to pursuing minor offenders has had little to do with the historic drop in crime, arguing that it was driven primarily by demographic, economic and possibly even environmental factors; that to the extent that better policing was effective the most important element was simply increasing a visible police presence in high-crime areas, not targeting literal or metaphorical broken windows; and that in practice “broken windows” policing has been implemented in a way experienced as oppressive by minority communities, and has resulted in far too many tragedies like the death of Eric Garner.

William Bratton and George Kelling have mounted a vigorous defense of “broken windows” policing here. A reasonable place to start for a round-up of various theories behind the historic drop in American crime rates is here. The Dish also has a good run-down of the current discussion.

Whenever I read about this question, I think about how much of the discussion glosses over the ways in which demographic and economic variables are inevitably intertwined with police practices. What if, for example, “broken windows” policing is very popular with homeowners – much more popular than can be justified by any demonstrable change in crime. Well, policies that are really popular with homeowners tend to improve home values. Which, in turn, can produce economic and demographic change – which, in turn, can drive down crime rates.

The 1990s saw a massive drop in crime across the country, and a larger and faster drop in the big cities, among which New York was a leader. The 1990s also saw an urban economic and demographic renaissance – and, again, New York was a leader. Was “broken windows” policing pivotal in terms of reducing crime by making the environment less-favorable to criminality? Or did it only have a modest direct effect – but a more substantial indirect effect by facilitating demographic changes in the city? Or did causality run the other way, with more and more aggressive policing strategies the product of a changing demographic (and political) profile of the city that demanded them? Because both trends happened together, it’s undoubtedly hard to tease out the answer to the question.

Cross-city comparisons are also complicated by the fact that cities and regions compete with each other. Jersey City is competing with Hoboken and Newark for commuters looking for a cheaper alternative to both New York proper and the more expensive suburbs. If Jersey City implements “broken windows,” and thereby convinces commuters to move there, demographic change may drive crime down further. But this, in turn, will drive up housing values, creating “relative value” in Newark even if Newark didn’t implement “broken windows” policing. A rising tide in the region may lift all boats, at least to some degree – maybe to a considerable degree. But that doesn’t mean that there are no benefits to being a first mover rather than passively waiting for the tide.

Then there’s the problem of what we’re using as a baseline condition. For example: compare two jurisdictions, one that implemented “hot spot” policing (which involves concentrating police presence in high-crime areas) and one that did not. There’s evidence that “hot spot” policing can reduce crime – and not merely displace it. But my impression is that “hot spot” policing is not a resource-neutral strategy. At a minimum, you need to invest in the infrastructure and training needed to identify “hot spots” and deploy officers accordingly; at a maximum, you need to increase the size of the force so that “lukewarm” spots don’t wind up being virtually un-policed. So what you really want to know is what the bang is for the incremental dollar. That’s an especially important question when you start talking about much more expensive strategies, from “broken windows” policing (which definitely requires a larger force) to mass incarceration.

I’m sure the sociologists, criminologists and political scientists who investigate these questions are aware of these complicating factors, and I’m sure they try to control for them the best they can. But “best” may not be all that good. And when it isn’t, we fall back on a combination of common sense and personal bias.

So what’s my common sense, and my personal bias?

I know that, as a citizen of New York, I approve of low tolerance for offenses like graffiti and public disorder. I remember New York in the 1980s and I do not want to go back. (Neither does the Mayor, as it happens.) But I want to see order without oppression; I want every community in New York to feel like the police are there to protect them, and not to protect one part of the city from another.

“Hot spot” policing makes a lot of intuitive sense to me – it really amounts to no more than deploying your resources to maximize their direct impact. I suspect that “broken windows” policing has some direct effect on crime rates, but a small one; I suspect it has a bigger impact on gentrification, and that the arrow of causality runs both ways – that is to say, quality-of-life crime-fighting strategies make an urban area more attractive to gentrifiers, and a gentrifying neighborhood will increase political pressure to do quality-of-life policing. “Stop and frisk” was among the more intrusive strategies deployed, particularly when it was applied so widely, and had long ago reached the point of diminishing returns. I’m glad to see it cut back sharply under de Blasio.

I suspect that the average citizen of a high-crime area, likely to be someone in the left half of the income curve, approves of police efforts to improve quality of life and crack down on offenses that disrupt public order even if they are non-violent – and also approves of greater efforts by the police to integrate into and show sensitivity to the community. Neither approach is resource-neutral; the political question becomes whether the city – and the NYPD in particular – sees it as worthwhile to spend money and time on the latter even if it has no direct effect on crime rates (the statistic with the most bearing on the political fortunes of leadership of the NYPD).

Finally, I worry about the public choice consequences of a larger (unionized) police force, which are a big part of what has been revealed in the current fracas between Mayor de Blasio and the NYPD. The sheer weight of the police department means it has much more influence than it did before the 1990s. As well, this is an organization with a sense of mission and of accomplishment – crime, after all, went down an enormous amount in the past 20 years. It is obviously not taking kindly to suggestions that this accomplishment is only partly due to their efforts, or that those efforts are viewed by at a big chunk of the citizenry as self-serving.

But that’s why the head of the NYPD is a political appointee. Bratton does not need to turn against his own legacy – nor does he need to defend it aggressively. The consensus against allowing crime rates to go back up is overwhelming. What Bratton needs to demonstrate is that he has control over his department, and that he is committed both to keeping crime low and to reducing the perception that the police are an oppressive presence.

Which, however, genuinely represents a change of mission. It’s implicitly admitting that driving crime rates ever-lower is no longer the overwhelming priority – that the “change” goal is to lighten the police footprint. A change of that sort could very well be demoralizing – even threatening – to the NYPD. But Bratton surely remembers that CompStat itself was threatening when it was introduced – it meant telling beat officers that the computer knew better than they did how they should do their job.

So there’s a sale to be made: selling the department on the idea that improved community relations and more sensitive policing are really about a more effective and efficient use of police resources – a continuation of improvements in policing rather than a rebuke to the police. Ultimately, that sales job is what will deliver the results that critics of the police are rightfully demanding. I can’t think of anybody better to do it than Bratton. Now let’s see how much he makes that a priority.