It was a crazy idea, but Richmond, California, wouldn't have signed off on DeVone Boggan's plan if it had been suffering from an abundance of sanity. For years, the Bay Area city had been battling one of the nation's worst homicide rates and spending millions of dollars on anti-crime programs to no avail. Boggan, who'd been working to keep teen offenders out of prison ... wondered: What if we identified the most likely perpetrators and paid them to stay out of trouble? In late 2007, Boggan launched the Office of Neighborhood Safety, an experimental public-private partnership that's introduced the Richmond model" for rolling back street violence. It has done it with a mix of data mining and mentoring, and by crossing lines that other anti-crime initiatives have only tiptoed around. The program's street team sifts through police records and its own intelligence to determine, with actuarial detachment, the 50 people in Richmond most likely to shoot someone and to be shot themselves. ONS tracks them and approaches the most lethal (and vulnerable) on the list, offering them a spot in a program that includes a stipend to turn their lives around. So far, the results have been promising. In 2007, when Boggan's program began, Richmond was America's ninth most dangerous city, with 47 killings among its 106,000 residents. In 2013, it saw its lowest number of homicides in 33 years, and its homicide rate fell to 15. In exchange for shunning dangerous behavior, ONS fellows receive anywhere from $300 to $1,000 per month.