Data-driven crime prediction: Is that a good idea?

Bike theft, vandalism, incivilities: all of these are considered as ‘petty’ crime. However, if they occur frequently in a neighbourhood, they can have a significant impact on your life. Oskar Gstrein at the UG/Campus Fryslân is supporting six police forces from all over Europe in reducing these crimes.

So-called petty crimes can considerably spoil your enjoyment of life, your feeling of safety, and level of comfort in your living environment. To tackle this, the European Union is funding the Cutting Crime Impact (CCI) project. Researchers and police forces from various countries including the Netherlands are working together on the CCI project to reshape police work by means of new insights and techniques. The final goal: to fight crime, increase safety, and improve trust in the police.

The researchers involved are focusing on four themes: Which role does the design of a neighbourhood have on crime; how can buildings and public squares make it more difficult to commit crime through their design? Additionally, how can the feeling of insecurity among citizens be reduced? This also requires strategies to improve the contact between police and neighbourhoods through, for example, a neighbourhood police officer (‘community policing’). The final theme that Gstrein and his colleagues Anno Bunnik and Andrej Zwitter are focusing on is the possibility of predicting crime: ‘predictive policing’.

Predictive policing is a new technique of connecting various data, through which it may be possible to predict when and where street crime or burglary will take place. The National Police Force is already working with the Crime Anticipation System (‘CAS’). Through a raster map of cities in the Netherlands, divided into 125 by 125-metre sections, it allows police to see where and when the chances of crime occurring are the greatest. The idea is that by using this technique, the police can prevent crimes and deploy their staff effectively.

Models that predict when something might happen on the basis of historic crime data... This idea sounds suspicious to many. Only recently Dutch courts found that the use of the system ‘Syri’ - a prediction model to anticipate fraud when receiving social benefits – was illegal. ‘I know that there are many concerns about the use of data and algorithms, due to privacy concerns, but also due to lacking transparency and accountability,’ admits Gstrein. ‘In the case of predictive policing, people associate this with the film Minority Report, which is set in a future in which the police can arrest criminals before they have even committed a crime. But this isn’t the direction that we want to go in. We shouldn’t use a system if people don’t know how it works or if people don’t trust it. We must be aware of the disadvantages and risks. That is where our task lies. We function as a critical friend.’

According to Gstrein, the question is whether predictive policing would indeed help to reduce crime levels. Still, he finds it useful to develop such a system in order to learn more about the environmental factors and context behind crime, as well as to enable useful information sharing and better management within police. Potentially, predictive policing might also become a tool to better explain decisions and practices. To this end, he is researching the existing literature, legislation, international treaties and UN resolutions on (human) rights and privacy. Armed with this knowledge, he is testing ideas and plans and publishing articles and advice about the subject.

As a lawyer and philosopher, Gstrein is uniquely aware of questions concerning human rights and ethics. ‘I think that we need to think about all of the information concerning us. Often we share it freely and much of it is stored without us being aware. We have yet to fully explore the possibilities of this new digital society, which will keep us busy for a while. However, just like we once invented curtains to stop people from peering inside, we now need to discover how to share digital data with the right person at the right time.’

Dr Oskar Gstrein is Assistant Professor at the UG/Campus Fryslân. As part of the Data Research Centre at Campus Fryslân, he is involved, among other things, with the Cutting Crime Impact project, which is searching for ways to reduce crime and reshape police work.