In a case of art imitating life, police across the country are warming up to the concept of predictive policing.  This idea seems to parallel the Tom Cruise film Minority Report, in which individuals are arrested for crimes which they are expected to commit in the future. In the film, a fictional “pre-crime unit” receives authority from a three-man group of psychics called “pre-cogs” to arrest individuals before the crime is committed.  Not surprisingly, the idea of predictive policing has many civil libertarians very concerned.

However, today’s predictive policing has more to do with Walmart and Amazon retail strategies, based on statistical models of earthquake aftershock predictions, than it does with the film.  Predictive policing is simply an extension of predictive statistics, a field of study used to prognosticate everything from how many Pop Tarts should be stocked before a hurricane to where the hurricane is headed.

The actual mathematics come from a formula developed to predict aftershocks in California.  A recursive algorithm is a formula that takes historical and real time data to create a prediction of a future event.  Recursive means that the system keeps shaping its predictions as new data comes in.  George Mohler of UCLA modified the earthquake algorithm to predict burglaries and auto theft.  Mohler notes that the same seismologist model has been used to predict the spread of epidemics; the transition from physical phenomena to social activity was a good fit for the formula.  Burglaries and auto theft were chosen because these crimes have a greater tendency to display patterns than murder and assault.

The Mohler model was tested in a simulation run on the San Fernando Valley using crime statistics beginning in 2006.  The results were startling.  The Mohler model successfully predicted future criminal activity 20 – 95% more often than CompStat.  The reason?  CompStat analyzes historical data that indicates a pattern and relies upon management to extrapolate the pattern forward to a general area.

The Mohler model produces a daily list of predicted “hot-spots” of 500 ft. X 500 ft. and the expected times of the crime.  This method permits officers to be in a position to interdict burglaries and auto theft before or as they occur.  In the current era of extremely tight police budgets, such careful targeting of manpower can maximize resources.

However, Mohler was not the driving force behind all this.  LAPD Captain Sean Malinowski was asked several years ago to supply crime statistics to UCLA for a research project.  Capt. Malinowski has an academic mind and quickly became intrigued with the idea that predictive statistics could be used to place police resources where a crime was going to occur before it did.  Through the next several years, he immersed himself in the subject, attending lectures and reading everything he could find on the subject.  Capt. Malinowski soon collaborated with several academics at the University as the method took shape.  On January 23, 2012, Captain Sean Malinowski will present the technology to delegates of Defence Geospatial Intelligence conference in London, England.

As always, there is the dark side.  The use of this computer technology represents an excellent tool for law enforcement authorities.   The potential for abuse by equally computer-savvy terrorists or criminals also exists.  An understanding of the predictive pattern in which police deploy could leave a target less protected if police are concentrated at another location.

How this technology will play out in the courtroom remains to be seen.  If a police officer is directed to a specific location based on this system, will there be sufficient grounds for search and seizure to face a Fourth Amendment challenge?  Civil libertarians will no doubt test predictive policing.  Supreme Court challenges will follow as this newer procedure becomes known outside of the law enforcement community.

Learn more about this article here: