Can police predict crimes?

New technology makes prevention possible
May 24, 2013

Police are responding to a perplexing crime mystery. Since the mid-1990s, there have been sustained reductions in crime, and these have continued despite the onset of austerity. It is a trend running counter to Home Office modeling, based upon long-term historical data, that confidently asserted crime would increase in a recession.

A complex set of social, political, economic and technological factors have all contributed to this crime drop. But as the effects of austerity “bite” and police budgets are cut, there is a need to find more efficient ways of managing crime. Contrary to the austerity mantra of “doing more with less” often cited by senior police officers, this involves police “doing less with more”—intervening less often in social life, but with greater impact. At its best, this means anticipating problems and preventing them from happening in the first place.

Against this backdrop Tom Winsor, the new Chief Inspector of Constabulary, used a recent speech to urge police to do more preventing of crime and less reacting to it. The newly elected Police and Crime Commissioners are also investing significantly in crime prevention. In policing and criminal justice policy, prevention is clearly where the action is. Why is this happening now, and how should police respond to this “preventative turn”?

The answer to the first question has much to do with the shrinking police budget. Reacting to crime after it has happened has become the default mode of policing. Chasing bad guys, detecting crime and catching criminals—“reactive policing”—offers a seductive vision of police work. It provides police officers with a clarity of mission, and police managers with an activity that is relatively easy to measure. Preventing crime, either before it happens or at an early stage, is less dramatic and more difficult to measure in terms of success. But if you can get it right, it is considerably more cost effective.

Modern policing organisations are already doing more and more preventative work. Domestic violence and child protection are two areas where this has happened, with regular Multi-Agency Risk Assessment Conferences involving police and other agencies sharing data on troubled individuals and families. Likewise, counter-terrorism policing has been transformed by the Preventing Violent Extremism agenda. These all attempt to spot “signals” of vulnerability prior to occurrences of criminal behaviour. Once such “risk markers” are identified, police can implement a range of early interventions. This includes “situational crime prevention” measures that work through “target hardening,” such as improved car security.

The journey towards this more preventative approach originated with “problem-solving” and “hotspots policing.” The latter holds that crime clusters in certain locations, and that by targeting these clusters police can effectively control crime. Problem-solving policing, which has been heavily promoted by the Home Office, urges police to target underlying problems rather than just responding to individual incidents. Both of these models involve “downstream prevention,” requiring multiple incidents to occur to steer interventions.

In the immediate future, the combination of austerity and new information technologies will bring about a shift towards more “upstream prevention”—intervening prior to the occurrence of multiple crimes. New data mining and improved analytics have started to enable far earlier interventions, for example with Suspicious Activity Reports for criminal financing. To facilitate this move “upstream,” we need to connect all of these different predictive and preventative activities to re-think prevention for today’s world. For unlike crime investigation or patrol work, prevention has not been understood as a discrete policing discipline involving its own knowledge base and techniques.

The potential of this emerging approach is illustrated by recent work undertaken by Her Majesty’s Inspectorate of Constabulary and Cardiff University to improve the police response to anti-social behaviour. This involved using a combination of sophisticated data analytics, new call-handling technologies, and evidence-based policing practice to leverage increased police efficacy across much of the country.

This is not quite Minority Report style “pre-crime,” but shifting the accent from problem-solving to problem finding. For example, there is now strong evidence that untreated anti-social behaviour acts like a magnet for other problems. Likewise, it has been shown that a burglary significantly increases the likelihood that further future offences will take place locally. In both instances, understanding the markers of risk affords the possibility of early intervention to deter or disrupt potential offenders, or alternatively to protect possible victims.

There are real practical challenges for police in making this shift. These include how to handle the inevitable “false positives” (predicting crime that doesn't happen) and “false negatives” (failing to predict crimes that do). There is also the phenomenon of what the philosopher Ian Hacking labels “the looping effects of human kind”: human behaviour adapts and responds to interventions, so effective prevention measures may become progressively less so. Police will have to continually update their tactics.

New technology will play an important role. Academics and police are conducting promising experiments using sophisticated algorithms to predict burglary patterns. But at the same time, technology will increase the pressure on the police. With the advent of big data and the spread of sentiment-mining tools to analyse data from social media and elsewhere, the capacity of police to know about previously hidden crimes will expand massively.

As these technologies transform police work, it will pose profound challenges. What should police do if, for example, they discover lots more people with a sexual interest in children than they previously knew about? The prevention challenge for the police may soon become far more difficult.