The Milken Institute had a forum on public safety last weak that included Bill Bratton and the two people to whom he gives primary credit for defining the “broken windows” style of policing he practices: George Kelling and James Q. Wilson. Unusually for such a public event, some actual intellectual progress was made.
There have always been two versions of the “broken windows” idea, but the two are often confused. Both are arguments for cracking down on relatively minor offenses as ways of reducing major crime, but the proposed mechanisms are entirely distinct. (Both might be at work, but neither implies the other.)
Broken Windows (A), the source of the original metaphor, is a fundamentally a “signalling” idea. It holds that minor crimes, and even conditions that are not themselves criminal, convey to people who might commit major offenses the implicit message that the territory in question is “unowned” and out of control, and therefore a safe place to commit a crime.
Wilson said frankly that the evidence for Broken Windows (A) as a crime control strategy was suggestive but less than convincing. (It’s probably much more effective in some circumstances than others.) He also said that the justification of Broken Windows (A) policies didn’t rest on its efficacy in controlling major crime, but in the fact that if you talk to people in disordered neighborhoods, order maintenance is what they want. Broken Windows (A), he said, contributes to “public happiness,” whic is a legitimate purpose of democratic government. If it also controls major crime, that’s a bonus.
Broken Windows (B), while it supports the same policy, is a completely different argument. It rests on a demonstrable fact about offenders rather than a plausible, but still speculative, idea about situations. Ever since Jan and Marcia Chaiken’s Varieties of Criminal Behavior, we’ve known that people who commit a lot of serious crime are usually (not always) people who commit a lot of crime, period. Most people who — to use Kelling’s examples — jump subway turnstiles or park in handicapped zones — aren’t armed robbers, but arresting a random sample of fare-beaters or handicapped-zone-parkers will likely yield a disproportionate number of armed robbers.
Wilson said something else that badly needed saying. “Broken Windows” has nothing to do with “zero tolerance,” which taken literally is an impossible policy for an urban police force. Of course officers are going to exercise discretion about what law violations to tolerate. Broken Windows policing, though means defining in advance and publicly what sorts of behavior will not be tolerated, something that will properly vary from area to area. By defining a particular offense in a particular neighborhood for (temporary and local) zero tolerance, the police can change offender behavior much more quickly, making many fewer arrests, than would be the case if offenders are left guessing about what will, and what will not, get them arrested. Think of it as targeted zero tolerance.
Postscript My friend David Kennedy has targeted zero tolerance idea about how to use this to greatly reduce the cost of shutting down a street drug market; I bet it would work. If you have any influence with your local police department, send me an email and I’ll tell you all about it.
Update: Kevin Drum finds the above interesting, but notes that the two versions of BW lead to “the same style of policing.” That’s true in the abstract, but not always in detail.
One set of offenses — or even non-offenses, if police are skilled at making pretextual traffic stops to search people who look as if they might be bad actors for guns or drugs, or just to check on whether they have outstanding warrants — might produce the highest yields of arrests of very bad actors, as provided for in BW (B), while another set yields the maximum in order preservation and neighborhood protection, the goal of BW (A). It might still be that a smart police department would recognize that a strategy of making arrests that can be justified under BW (A) is likely to be more sustainable politically, but otherwise the two theories lead to somewhat different operational approaches, if all you want to do is reduce the Part I crime numbers. That’s why Wilson’s point is crucial: neighborhood protection through reduction of disorder is valuable in its own right, in addition to whatever contribution it makes to reducing serious crime.