Arizona lawmakers are looking to invest $2 million in state funds in predictive policing technologies. Up until now, only individual police departments throughout the United States have used funding for the emerging technology; if the bill passes, Arizona would become the first state in the nation to throw money toward the technology.
The new bill would direct funds to one-year pilot projects to develop software that can predict all sorts of offenses—property crime, drug activity, traffic accidents, and gang and gun violence—in both urban and rural settings. The projects would likely kick off in Phoenix, Mesa, and Lake Havasu City.
While the technologies may be new, the ideas behind predictive policing are not. Police departments have been trying to get ahead of crime for decades. In 1994, the New York City Police Department introduced CompStat—a precursor to predictive policing that uses crime data and computer-mapping technology to visualize crime patterns and inform police responses. The program has been credited with dramatically decreasing New York City crime rates, which did in fact drop in the years after it was implemented. After that, the program rapidly spread to police departments across the U.S.
"After 40 years of uncontrolled crime increases, fear and disorder, in the 1990s we finally got it right,” former New York and Los Angeles Police Department police chief William Bratton, who pioneered the use of CompStat, wrote in the journal Policing in 2008. No formal studies were conducted, however, to determine if CompStat was really responsible for the plummeting crime rates.
Predictive policing would use data to figure out where crimes are going to happen next (just like Minority Report, but not nearly as cool).
The downside to CompStat is that it’s reactive. It crunches statistics to figure out where crimes have already happened or are in progress, and allows police to respond faster. Predictive policing, on the other hand, would use the same data—and more—to figure out where crimes are going to happen next (just like Minority Report, but not nearly as cool). Predictive software technologies integrate data on crimes, demographics, land use, the weather, parolee populations, emergency calls, businesses, foreclosures, and traffic accidents. PredPol, one such software start-up that many believe is a frontrunner for the Arizona contract, uses all that information to produce a list of crime hot spots—500-by-500-foot squares identifying where the risk of crimes occurring is highest during a given time window.
Police departments in Virginia, Tennessee, Georgia, California—and abroad, in the United Kingdom and the Netherlands—have all tried predictive analyses to fight crime, but research on the effectiveness of predictive policing is still relatively hard to come by. What results are out there, however, are decidedly mixed.
A 2014 report from the Rand Corporation looked at a predictive policing effort in Louisiana and found that the practice had no statistically significant effect on property crime. But there was some good news: Police officers themselves felt the program was beneficial, and it didn’t cost the department any more than their standard policing practices would have.
The police department in Richmond Virginia credited predictive policing strategies with its ability to reduce random gunfire on New Year's Eve (a problem that had plagued the city for years) by 47 percent and increase weapons seizures by 246 percent in 2003.
More recently, the city of Santa Cruz, California reported a 27 percent decrease in burglaries after implementing PredPol, and the Los Angeles Police Department reported that the system reduced property crimes by 13 percent. Predicting multiple types of crimes—thefts in addition to violent crimes—is still a problem for PredPol and other software technologies, but one that $2 million in funding from Arizona’s Department of Public Safety could potentially go a long way toward solving.
A tougher problem to solve may be convincing people that—given that predictive policing is based on information gathering technologies—the use of that information won't infringe on their privacy or civil liberties (Seriously, have you seen Minority Report?). "Community trust is huge as we move down this path," LAPD police chief Charlie Beck said at a symposium on the future of predictive policing. "We need to be extremely transparent.”