Disappointing results for predictive policing

The Max Planck Institute in Freiburg does not see any proof of effectiveness for predictive policing in preventing home burglaries. Another study is expected next year from Hamburg.

So far, there has been no proof in Germany that so-called "predictive policing" leads to crime rates being lowered in a particular area. Two investigations aim to shed light on this: one "study of new technologies for predicting home burglaries and their consequences for policing practice" is currently underway at Hamburg University, however the project does not end until December 2018. In the meantime, evaluation of a predictive policing project in Baden-Württemberg by the Freiburg Max Planck Institute for Foreign and International Criminal Law has been completed.

Effects "moderate" at best

The investigation by the Max Planck Institute, officially presented at the end of August, followed a pilot project by the Baden-Württemberg police. Led by the Land Criminal Police Office in Stuttgart, PRECOBS software from the Institute for pattern-based prediction techniques (IfmPt) was used. The company has its headquarters in Oberhausen.

PRECOBS is based on "Near Repeat theory". Similar to the "Broken Windows theory" this assumes that criminal acts can follow earlier crimes within a certain period of time. Data on the crime location and time, stolen items and the modus operandi are thus weighted in a special scoring process and fed into a prediction calculation for further crimes. This aims to detect patterns and track down serial offenders.

The six-month evaluation phase in Baden-Württemberg ended on 30 April 2016. According to the final report that has now been made available, it remains "difficult to judge" whether the PRECOBS software is able to contribute towards a reduction in home burglaries and a turnaround in case development. The "criminality reducing effects" were only moderate and crime rates could not be clearly minimised by predictive policing on its own.

Also controversial among officers

In some areas involved in the pilot project, the number of burglaries fell, while in others an increase was recorded. In addition, most predictions involved urban areas, which have higher burglary rates and thus also more statistical data. The study is thus critical of the benefits for rural areas. In total, 183 alerts were processed in the six months, with the large cities of Karlsruhe, Pforzheim and Stuttgart predominantly affected.

The PRECOBS programme is described in the study as "user-friendly", however some "difficulties" were noted at the beginning. In the course of the pilot project, it was reported that the software was able to recognise domestic incidents as "false alarms".

PRECOBS nevertheless remains controversial among police officers. The Max Planck Institute carried out an online survey of 700 police offers in this respect. The results showed that the use of the software "had a strongly polarising effect" within the force. Only around half of those surveyed viewed it as a promising model, the other half rejected the process.

Andrej Hunko, MdB 2017