Interdiction prediction

The line between fiction and reality, blurry as it is already these days, may soon be undergoing a complete obliteration for the purposes of law enforcement, courtesy of the Los Angeles Police Department. One of the most consistently understaffed police departments in the country — 10,000 officers, 4 million people, you do the frickin' math — has resorted in the past to any number of strategies to control crime. What the LAPD is predicted to do about it in the future could be a crime in itself.

The operative word is "predicted" or, more accurately, "predictive." The Los Angeles Times reported Saturday that the department is investigating predictive policing, which is, for the most part, exactly what it sounds like.

"Predictive policing is rooted in the notion that it is possible, through sophisticated computer analysis of information about previous crimes, to predict where and when crimes will occur," The Times' Joel Rubin reports. "At universities and technology companies in the U.S. and abroad, scientists are working to develop computer programs that, in the most optimistic scenarios, could enable police to anticipate, and possibly prevent, many types of crime."

Rubin reports that some of the cutting-edge work in the field is being done at the University of California-Los Angeles.

Another researcher, at Santa Clara University near San Jose, is investigating whether it's possible "to forecast the time and place of crimes using the same mathematical formulas that seismologists use to predict the distribution of aftershocks from an earthquake," Rubin reports.

"The naysayers want you to believe that humans are too complex and too random — that this sort of math can't be done," said Jeff Brantingham, a UCLA anthropologist who is helping to supervise the university's predictive policing project.

"But humans are not nearly as random as we think," he told The Times. "In a sense, crime is just a physical process, and if you can explain how offenders move and how they mix with their victims, you can understand an incredible amount."

◊ ◊ ◊

You're forgiven if you think you've seen this movie before. You probably have. The 2002 Steven Spielberg sci-fi film "Minority Report," starring Tom Cruise and Colin Farrell, proposed just such an approach to law enforcement in Washington, D.C., with officers of a shadowy Precrime Unit dispatched to arrest unsuspecting murderers before they become murderers. The result: six years without a homicide in the nation's capital.

Some in the real world are skeptical of the whole thing, including some in the LAPD. ""There is the science of policing, and there is the art of policing," Deputy Chief Michael Downing told The Times. "It is really important that we learn how to blend the two. If it becomes all about the science, I worry we'll lose the important nuances," said Downing, who directs the department's counterterrorism units.

As you might expect, one of those "important nuances" has to do with identification of suspects, and exactly who gets to build the models and set the definitional standards of suspects and suspect behavior.

In a city like Los Angeles — with a weave of people of different races and ethnicities, a history of explosive evidence of racial intolerance, and a parallel history of fractious police-community relations — predictive policing has the potential for abuse, just like any crime-fighting tool. The flashpoints that led to riots in Watts in 1965 and South-Central in 1992 had decidedly strong racial overtones; what would prevent a rogue element of the LAPD Predictive Unit from exacting pre-emptive retribution on former felons of another color or ethnicity — or innocent people who just happen to know former felons, or innocent people mistaken for felons?

◊ ◊ ◊

There's another little matter. Brantingham of UCLA said "humans are not nearly as random as we think," and given the fairly narrow motivational range of human behavior, that's probably true. But what happens to predictive-policing models when a tractor-trailer suddenly jacknifes on Interstate 10 — preventing a predictive suspect from even getting to where he's expected to commit a crime?

What happens when a predictive suspect is on his way to commit a crime the LAPD anticipates when, like all criminals, he decides to capitalize on a new opportunity — taking a left turn to seize on that unexpected chance, rather than heading for his previous (and predicted) destination?

Brantingham's confidence in the predictability of human behavior seems to have a lot invested in people staying put, not going anywhere, not traveling beyond a given range or outside a given neighborhood — hardly the case in a dynamic, constantly mobile city like Los Angeles.

Brantingham's calculus also appears to dismiss any way to factor in how the infrastructure could play a part in effective predictive policing, as well as the predictable unpredictability of variables that have nothing to do with human behavior: the weather; the traffic; the likelihood of accidents; the sudden chaos of earthquakes; a water main blows; an electrical substation shuts down — all the deus ex machina events that characterize life in any modern city, the kind of events that define life in L.A. The kind of events whose ripple effect would have an impact on everyone in the city — including the suspects who suddenly aren't where the LAPD expects them to be.

The next step in this tantalizing speculative forensic enterprise may come soon; Rubin reports that the LAPD may be first among equals as the Justice Department decides which big-city department will secure a $3 million grant to further study the idea.

So many big ideas that take the nation by storm often start in California. You have to hope that if the LAPD takes the lead in this provocative, Nostradamus-by-algorithm approach to law enforcement, they get the bugs worked out. First by understanding that, as a rule, bugs don't always conveniently stay where you think they are.

---

Image credits: 'Minority Report' still: © 2002 DreamWorks/Twentieth Century Fox. LA freeway map detail: iNetours.com