Predictive policing is not 'Minority Report' ...at least not yet
Using computer algorithms, 'predictive policing' estimates where certain types of crime may occur, in a similar way that meteorologists forecast weather
Intelligence-led policing, an umbrella concept which includes “predictive policing,” could help move police departments beyond the increasingly outdated Community Policing model and toward a more-professional approach that envisions crime suppression and control as the most important result of good police work. The premise of “predictive policing” is that human behavior is predictable, which opens the door to computer-assisted techniques that can forecast subtle patterns of specified criminal activity.
Though at first blush it appears that intelligence-led policing and “predictive policing” are the same, they are, in fact, slightly different. According to David Sklansky of the University of California at Berkley, “[t]he main distinction between intelligence-led policing and predictive policing is that predictive policing claims to be more ambitious and more technologically sophisticated...” In other words, “predictive policing” may be more aggressive in its application of intelligence-led policing techniques, a distinction that may be more obvious in theory than practice.
Let’s also be clear at the outset. “Predictive policing” isn’t predictive. Using computer algorithms it estimates where certain types of crime may occur, in a similar way that meteorologists forecast weather or seismologists estimate earthquake aftershocks. No crystal balls, necromancy, witchcraft, prophecy, or freakish “precogs” — only inputs, outputs, and informed analysis of criminal patterns, not individual criminal behavior.
This approach is likely more useful in large departments with multiple but unlinked databases that supervisors and administrators now use to track and report past criminal behavior and request future resources. Smaller departments lacking technological resources may find themselves using less sophisticated, manual techniques.
Using “predictive policing” models, supervisors can allocate current police resources in anticipation of criminal activity. In essence, it takes a large amount of data and, with proper analysis, transforms information into actionable intelligence.
In 2009, spurred on by the tragedy of 9/11, and building on the experience of William Bratton’s popular and sometimes controversial CompStat program, the National Institute of Justice awarded grants to seven large departments to pioneer “predictive policing.” These departments were the LAPD, DC Metro, Chicago PD, Shreveport, LAPD, Boston PD, the Maryland State Police, and the NYPD.
Other, smaller departments, to include Santa Cruz (Calif.) and Richmond (Va.) worked on similar “predictive policing” projects, apparently with little support from Washington.
Though the results from the seven large departments are not readily available, some initial results have appeared promising in Santa Cruz and Richmond, at least according to proponents:
• As early as 2003, police in Richmond used a “predictive policing strategy” to respond to the expectation of gunfire in the city during New Year’s Eve. As a result of police deployment based on the analysis, complaints were reportedly down 47 percent and the number of recovered weapons was up 246 percent.
• In July 2011, the Santa Cruz PD reported a 27 percent drop in the number of reported burglaries as a result of their directed patrolling tactics.
Unnecessary Rhetorical Barriers
However, proponents — who are often also responsible for implementation — may overstate the effectiveness or obscure results behind carefully crafted language. A lot of tax dollars are at stake and to admit that the program produced less than decisive results may go against the grain. In fact, overselling the idea could lead to understandable objections from officers and civil libertarians alike.
Much of the talk of effectiveness concerns better using “limited” police resources, all of which may rightly resonate with managers and leaders but not necessarily patrol offices. Influencing this sort of change in a department is a difficult endeavor and takes time to implement if leaders want to limit turmoil.
Attendees and presenters at “predictive policing” symposia, working groups, and “advanced analytic” seminars appear to work hard to ensure the language is chock-full of jargon which may make many officers wary of the concept. When supporters use such terms as a “paradigm shift in policing,” or assert that “predictive policing” will “take policing to the next level,” or even refer to “stakeholders,” the proponents may be unintentionally creating unnecessary rhetorical barriers to officer acceptance.
Many experienced officers will rightly see that “predictive policing” may simply reflect the results of officer intuition. “It’s in the remaining 10 to 15 percent where police intuition may not be quite as accurate” where the program would be most effective, at least according to Jeffrey Brantingham, a UCLA anthropology professor who is assisting with LAPD’s program.
The Department of Justice, which has been responsible for funding Community Policing, law enforcement’s sacred cow for the last twenty years - maintains that “predictive policing” complements Community Policing. However, in University of California-Berkley Prof. David A. Sklansky’s informed view, intelligence-led and “predictive” techniques could well return policing to a more professional model without the myriad problems encountered in the 1960’s. Quoting an “influential backer” (Temple University’s Jerry H. Ratcliff http://jratcliffe.net/ of intelligence-led policing, Dr. Sklansky argues that no longer will a “community’s concerns... trump an objective assessment of the criminal environment.”
Leaders don’t want to lead an evolution and intelligence-led policing is not revolutionary, but evolutionary. It won’t change what officers do, but may change when and where they do it. So long as the crime suppression objectives are clear, and building on the partnership aspects of Community Policing, intelligence-led policing may well be an important and necessary step to getting police performance on track for the twenty-first century. Overselling the idea now could well backfire, creating hesitancy and skepticism from rank-and-file officers that could slow the idea’s adoption. That would be an unfortunate loss of an opportunity to leverage modern technology in a way that furthers effective law enforcement.