New research is shedding deeper light on the underlying social processes which do work, however, and that's why this is a "Today I Learned" article instead of a "Politics, Society, and the Law" article. This team did a large-scale data analysis of Boston between 2011 and 2012, and found that the events (from arrest, 911, and 311 records) fell into a few natural categories: private neglect, like rats in buildings or parking on lawns; public denigration, like graffiti and broken windows; private conflict, like domestic and landlord-tenant disputes; public disorder, like reports of panhandlers and drunks; and public violence. They broke public violence down further into "basic" violence, violence involving guns, and homicides.
They compared how these different kinds of issue cropped up over space and time. While it wasn't possible to test if one thing caused another, it was possible to do what's called "cross-time correlation:" does having a lot of public denigration in a place, for example, correlate with having more public disorder or violence there later?
The answers were quite interesting. Unsurprisingly, the strongest correlations are between private conflict and public disorder and violence. Those, in turn, tend to feed back on themselves, sometimes escalating to guns, which are (by far) the main predictor of homicides. Perhaps more surprisingly, public denigration – the classic "broken windows" – showed no predictive power at all.
If we think about how conflicts tend to escalate, this makes a certain sense; if nobody had ever told you about "broken windows" theories, you would say that most fights (and murders) are between people who know each other, most fights start small and grow larger, fights between people can last a long time and spread to include other people, happen in private and in public, and so on, and probably more fights have their first origins in private than in public, but not by much.
The statements above probably seem pretty obvious, which is what made the broken windows idea seem so radical: it was upending all of this, suggesting that maybe the reason people thought it was OK to get into ever-escalating fights was the sense of decay around them, and if we just made everything look nicer, people would stop doing that.
It was a radical, but not crazy, idea; people do react to their surroundings and take cues from it. But the data increasingly seems to suggest that it's interesting, but wrong.
If this particular study has captured the real mechanisms – and as it's a study of just one city over one time window, it's far too small to give us real certainty of that – then it suggests that a more effective role for police would be to act as moderators of disputes, helping resolve and stop fights before they escalate. That's obviously a much harder job than ticketing panhandlers.
Of course, that answer may itself suffer from the blinders of asking "what can the police do?," when it's not obvious that the police are even the right mechanism. If there's one reliable pattern in sociological studies, it's that people don't become drug dealers, armed robbers, or junkies because they're stupid, inherently evil, or have some kind of cross-generational proclivity to do it; they do these things as fairly rational choices given an extremely limited set of options. 
That means that even murder is a symptom, rather than a cause, and actually fixing these problems will require answering deeper questions, like "why are people resolving their disputes by murder, rather than (say) talking it out, suing each other, or just moving away from each other?" In general, what we discover is that those alternatives aren't useful options to the people involved for various reasons which aren't always obvious to outsiders – and it's understanding that sort of thing which is the key to actually fixing things.
 And perhaps more importantly, it provides neat political narratives, as well as a good rationalization for policies that the public may want but not wish to admit to, such as forcibly removing the homeless or policing racial groups. The sad fact is that the politics of criminal law almost invariably boil down to something sordid.
 Even, perhaps especially, taking drugs. The key result is the famous "Rat Park" experiment (https://en.wikipedia.org/wiki/Rat_Park), which found the flaw in all those experiments that showed that rats will instantly become addicted to cocaine or heroin and take it until they die: the cages were confusing the experiment. When rats had an option of doing normal rat things or taking drugs, they had very little interest in drugs; they became addicted when it was a choice of that or being locked in a featureless white cage without drugs for months on end. This result has since been generalized beyond rats, but the key idea is there.
- Sr. Software Engineer, 2011 - presentI work on infrastructure for the data that powers Google Maps.
- Gilt GroupeSoftware Engineer, 2011 - 2011
- ChartbeatSr. Software Engineer, 2010 - 2011
- VMwareSoftware Engineer, 2005 - 2009
- Sun MicrosystemsSoftware Development Intern, 2003 - 2003
- IBM CanadaSoftware Development Intern, 2002 - 2003
- WaterlooComputer Science, 1999 - 2005
Genomics, Medicine, and Pseudoscience: Birth control pills have dramatic...
This is Steven Salzberg's blog on genomics, pseudoscience, medical breakthroughs, higher education, and other topics, including skepticism a
Yes, The Appeals Court Got Basically Everything Wrong In Deciding API's ...
Copyright expert and professor Pam Samuelson, one of the most respected scholars of copyright law, has published a short paper explaining wh
Trip report: a tale of mobile/WiFi in two developing countries, Haiti &a...
I've been away for the past 3 weeks in two very different, but very close countries: Haiti & Cuba, separated by less than 100km. While it wa
Stories from a Social Entrepreneur: Coding Bootcamps - Quick Recap
We had a great conversation about Coding Bootcamps yesterday at Cornell Tech's To Code and Beyond Conference (see yesterday's post). I think
Smoking Gun: MPAA Emails Reveal Plan To Run Anti-Google Smear Campaign V...
If you talk to the reporters who work for various big media companies, they insist that they have true editorial independence from the busin
Bluesmart: Worlds First Smart, Connected Carry-On | Indiegogo
Location tracking, Bluetooth lock, USB battery charge, Smart scale and trip data app! | Crowdfunding is a democratic way to support the fund
Oil Well - every cooking oil compared - Information Is Beautiful
Which is the ultimate cooking oil? Every single cooking oil graded for health, smoke point, flavour and massage suitability.
Mobile Addicts Multiply Across the Globe | Flurry Insights Blog
By Simon Khalaf, SVP of Publisher Products On June 29th Bank of America released the findings of its second annual report on Consumer Mobili