Whether or not “broken windows” policing tactics actually work is one of those debates that will never really end, mainly because there are so many different understandings of what broken windows means.

“Whenever somebody mentions broken windows, the question should be which version?,” says Princeton University political scientist Jonathan Mummolo, who is dubious of broken windows-linked claims about the efficacy of stop-and-frisk tactics and high-volume misdemeanour arrests.

You know which version of broken windows really does appear to work? Fixing broken windows. The term “broken windows” comes from a 1982 Atlantic magazine article by criminologist George L. Kelling and political scientist James Q. Wilson. “Social psychologists and police officers tend to agree,” they wrote, “that if a window in a building is broken and is left unrepaired, all the rest of the windows will soon be broken”.

As empirical evidence for this assertion, Kelling and Wilson offered a clever if not exactly dispositive late-1960s experiment by Stanford University psychologist Philip Zimbardo, who arranged to park one car with the hood up and no licence plate on a street in the impoverished New York City borough of the Bronx, and another on a street in affluent Palo Alto, California.

Thieves and vandals attacked the Bronx car and stripped it of everything of value in 24 hours, while the one in Palo Alto sat untouched for more than a week. Then Zimbardo smashed it in a few places with a sledgehammer. Within a few hours, locals had torn the car apart and turned it upside down.

Kelling and Wilson’s article was mainly about police tactics, though, and what they saw as the need for police to return to their traditional role of maintaining order rather than just trying to solve crimes.

Kelling had studied an experiment with police foot patrols in Newark, New Jersey, which had not succeeded in reducing crime, but had left residents of foot-patrolled areas feeling safer and more favourably disposed to the police.

He and Wilson argued that such efforts to ensure public order were of value even if they didn’t bring immediate crime declines.

The rest is history. In 1989, Kelling was asked by the chairman of the Metropolitan Transit Authority in New York City for advice on combating subway crime.

Kelling and the new chief of the transit police, William Bratton, hired in 1990, targeted disorder on the trains and in stations, cracking down on panhandlers and turnstile-jumpers, among other things.

The experiment was seen as a great success, newly elected mayor Rudy Giuliani put Bratton in charge of the entire New York Police Department in 1994, and crime plummeted citywide.

At least, that’s how Bratton and Kelling tell it; there’s a compelling two-part episode of the podcast Reply All from last year that gives most of the credit to transit police officer turned deputy police commissioner Jack Maple, whose CompStat crime-tracking system enabled the police to better identify trends and crime hotspots that they needed to target.

Bratton comes off fine in this account, but Giuliani does not. According to Reply All’s P.J. Vogt, the mayor kept pushing Maple for more arrests of minor wrongdoers even as crime rates fell.

Maple resisted, and in 1996 both he and Bratton left the department. Maple died in 2001; Bratton returned to run the NYPD from 2014 to 2016, and has continued to espouse his version of broken windows.

Meanwhile, crime kept falling. Not just in New York, but pretty much everywhere in the US.

Violent crime rates are still much higher than they were before the 1970s, but it’s possible this is an artefact of improved crime reporting and certain crimes, domestic violence, for example, being taken more seriously by police. The murder rate, which is less likely to be affected by such changes, is about where it was in 1960.

The urge to attribute this spectacular fall in crime to broken windows policing was, of course, irresistible to Giuliani and his fans, but it didn’t sit well with many researchers.

The decline had been too big and too widespread to be accounted for by a change in policing tactics in a few cities.

Scholars offered up alternative explanations ranging from the waning of the crack epidemic and the ageing of the baby-boomers to regulations that cut down on childhood lead exposure and the liberalisation of abortion laws.

Many others tried to suss out the direct effects of police tactics. Some studies backed up the theory that targeting disorderly behaviour curtailed crime, but others did not.

In 2004, a National Academies of Sciences, Engineering and Medicine report on Fairness and Effectiveness in Policing endorsed Kelling and Wilson’s original argument that “disorder should be an important focus of community crime control”, but concluded that there was no convincing evidence that arresting lots of people for minor offences led to reductions in serious crime.

In fact, even Kelling contended that “I’ve never been long on arrests as an outcome”. (Kelling died this May; Wilson in 2012.) The one policing method for which the 2004 National Academies panel found “strong evidence of effectiveness” was “problem solving in hotspots” — that is, flexible, not necessarily arrest-focused approaches to addressing localised crime problems.

Meanwhile, broken windows research that didn’t focus on police behaviour was delivering some interesting results.

There had been earlier studies of the impact of physical disorder, ie, actual broken windows, on crime, with mixed findings.

But the topic seems to have taken on new life with a 2008 article in the journal Science that reported the results of an experiment into whether disorderly conditions in the Dutch city of Groningen, graffiti on a wall behind a no-graffiti sign, bikes locked to a fence with a no-bike-parking sign, shopping carts strewn around a parking lot, etc, led people to violate other norms and rules.

Kees Keizer, Siegwart Lindenberg and Linda Steg of the University of Groningen concluded that, yes, “Signs of inappropriate behaviour like graffiti or broken windows lead to other inappropriate behaviour, eg, litter, stealing, which in turns results in the inhibition of other norms”.

Inspired in part by that study, University of Pennsylvania epidemiologist Charles Branas, who has since moved to Columbia, began looking for more ways to test the notion that disorder spreads.

Together with Penn criminologist John MacDonald and other scholars he designed a series of quasi-experimental studies that compared crime around vacant buildings in Philadelphia that had been equipped with new doors and windows by their owners in compliance with a new city law and similar ones that had not, and around vacant lots that had been cleaned up by the Pennsylvania Horticultural Society as part of a major urban-greening campaign and those that had not.

The results: a 39 per cent decline in gun violence around the vacant buildings that had been fixed up, and a 5 per cent decline around the vacant lots. More studies have since followed in what New York University sociologist Eric Klinenberg calls “one of the most exciting research experiments in contemporary social science”.

MacDonald, Branas and DePaul University’s Robert Stokes have a book coming out next week, Changing Places: The Science and Art of New Urban Planning, on the role urban landscapes can play in discouraging crime and promoting all-around wellbeing.

Meanwhile, other researchers have, over the past decade, attributed crime declines to community efforts to maintain vacant lots in Flint, Michigan, Business Improvement Districts that cleaned up shopping districts and provided security in Los Angeles, better lighting in public housing projects in New York City and high pollen counts around the US — which isn’t really on point, but seemed too interesting not to share.

One study found links between indications of physical disorder contained in calls for non-emergency services in Boston and subsequent crimes, another found that city blocks in Detroit with lower population densities and higher numbers of liquor licences were more crime-prone, yet others found that home foreclosures in Chicago and the Phoenix suburbs of Chandler and Glendale also brought increases in neighbourhood crime, albeit short-lived ones.

A group of Massachusetts Institute of Technology researchers put together a Google-Street-View-based tool to “predict the perceived safety of one million streetscapes”, although so far it has mainly been used to predict other things.

The physical disorder-related crime effects that have been identified aren’t always huge, and some studies have found none at all.

But in a 2018 assessment of policing strategies, a National Academies panel cited the Keizer and Branas research and concluded that evidence for what it called “Broken windows Policing II” — cleaning up vacant lots and other “place-based, problem-solving practices to reduce social disorder” — was both “strong” (that is, the studies were well designed) and “positive”.

The only other strategies to garner both those ratings were hotspots policing and “stop-question-frisk” tactics that were limited to such hotspots. More generalised stop-question-frisk, as was practised in New York City before being rolled back after 2013 in the face of court rulings and political opposition, rated only a “medium” and a “mixed”, as did “Broken windows Policing I”, which the panel described as “high-volume arrests for certain misdemeanours”.

The message here is not that replacing broken windows reduces crime while police efforts to maintain order do not. The evidence on the latter is more mixed in part because the research is more voluminous and in part because police approaches have varied so widely.

In his 2018 summing-up of the causes and consequences of the great crime decline, Uneasy Peace, Princeton sociologist Patrick Sharkey wrote of an emerging consensus that “more police on the street translates into less crime”.

Still, it is remarkable how little of the broken windows policy revolution has focused on the physical manifestations of disorder that generated the idea in the first place.

“They implemented broken windows, but instead of targeting windows they targeted people,” says Princeton’s Mummolo, whom I consulted because he studies policing but hasn’t been involved in the physical disorder-focused, Broken windows II research. “It was too metaphorical, maybe.”

This is one case where it may pay to take things literally.

• Justin Fox is a Bloomberg Opinion columnist covering business. He was the editorial director of Harvard Business Review and wrote for Time, Fortune and American Banker. He is the author of The Myth of the Rational Market