It has had a large impact on policy; for instance, New York City has recently pursued a policy of cracking down hard on "quality of life" crimes such as urinating in the street. (ZeroTolerance?)
However, the theory is speculative, and there is evidence against it. "I still to this day do not know if improving order will or will not reduce crime," Dr. Wilson, now a professor emeritus at the University of California, Los Angeles, said in a telephone interview (New York Times, 2004/01/06). "People have not understood that this was a speculation."
From the same nytimes article: "From June to October 1995, trained observers drove a sport utility vehicle at 5 miles per hour down every street in 196 carefully selected Chicago neighborhoods.
As they drove, a pair of video recorders, one on each side of the S.U.V., recorded social activities and physical features: litter, graffiti, drug deals, public drinking, everything within the camera's view. When the researchers were done, 11,408 blocks had been observed and videotaped. Then the police records on homicide, robbery and burglary were pulled for each of these 196 neighborhoods, along with in-person surveys of 8,782 residents.
In a landmark 1997 paper that he wrote with colleagues in the journal Science, and in a subsequent study in The American Journal of Sociology, Dr. Felton Earls reported that most major crimes were linked not to "broken windows" but to two other neighborhood variables: concentrated poverty and what he calls collective efficacy (another name for community-mindedness or community-activeness).
For a short time, you can view [a New York Times article on the subject].
Of course, this evidence has to do with the impact of broken windows on crime; its relevance to wiki, where we are trying more to preserve order or clarity or style rather than to prevent crime, is unknown.
For the application to software development see Wiki:FixBrokenWindows
I'm not sure how to apply this page to online communities. It's unclear what a "crime" is in this place (is it a fight? is it bad behaviour? is it violating a role?). It's unclear what "broken window" might be (a shallow page? a typo? an invalid argument?).
It's also unclear what the social scientist meant. Typically one would talk about "correlations" if two observations seem to be somehow connected. But this needn't mean a cause-effect relationship. To assume it without proof seems unscientific. So what does it mean that $50 millions are spent that way?
I doubt that this is relevant to Meatball or online communities.
I think it's very relevant. Check out some of the backlinks.
"Broken windows" are anything that is in a broken way that is unpleasant or annoying to look at: personal attacks, flame wars, genuinely off-topic content, flooding, spam, etc. A "crime" is anything that goes against the community's internal law (c.f. LegalSolution) or its expectations (c.f. CommunityExpectation). The most common "crime" in most communities is to break windows in some way, but there are others. -- MartinHarper
The point is to remember monkey do as monkey see. You want to remove GuidePosts that teach behaviour that is inappropriate, and add GuidePosts that teach behaviour that is desired. By fixing broken windows, you accomplish these things:
A punitive approach to breaking windows, as the New York "quality of life" crimes suggest is not what FixBrokenWindows is about. It's about healing. -- SunirShah
I feel that the data from Earls' study may not be relevant for online communities, since it is talking about crime, and we are talking about something else (we might think of breaking a community's expectations are analogous to stealing a car, and it is, but sociologically it is pretty different). But I thought people here would be interested anyway, as a tangential side-note to all of the talk about Wiki:FixBrokenWindows, and also because of the relevance of "collective efficacy" in Earls' study. -- BayleShanks
I believe in FixBrokenWindows as a core value needing no justification or rationale. We have a tidy front lawn because---because we want to have a tidy front lawn. Whether the study in question is valid or not, people do indeed infer that a behavior is acceptable if they see evidence of it being conducted openly by others. Graffiti tends to multiply. --Steve