All crimes are lined up with crimes

Police use algorithms to predict crimes - critics see "superstition"

In this gallery: 2 images

A software that predicts where and when a criminal will strike: What after a science fiction scenario in the style of Minority Report sounds, is already a reality in Austria. "Predictive policing" is now part of everyday life for the police. The Federal Criminal Police Office in Vienna hires its own programmers who deal with "crime mapping", the geographical allocation of crime. Algorithms are currently being used more and more to combat so-called twilight breaks.

Several factors

Specifically, the software defines zones in which there is an increased risk of break-ins. To do this, the system is fed the relevant statistical data. Burglaries are often committed by serial perpetrators. They often have a certain pattern of crime and try it again soon. The police save the place and time of the offenses as well as any additional information recorded on the perpetrators' approach.

The proximity to a motorway entrance is also relevant, for example, as this would allow perpetrators to flee more quickly. It is also important how many houses there are in a settlement, as burglars hope for more loot in larger settlements. "As soon as hotspots are identified, more and more police officers are sent there," explains Vincenz Kriegs-Au, press spokesman for the Federal Criminal Police Office, to STANDARD. This has significantly reduced the number of successful break-ins.

Decline in burglaries

According to crime statistics, around 44 percent of all break-in attempts were foiled in 2017. Overall, the number of burglaries reported fell from 6,680 to 5,808. The Federal Criminal Police Office justifies this development, among other things, with the successful investigative work. The algorithmically supported crime forecast has been in use by the Austrian security authorities for some time, says Kriegs-Au. The first tests were already carried out in 2015. Real-time warnings are also currently being tested via Facebook. Via the social network, residents of certain areas are to be alerted "promptly and in a targeted manner" in order to be able to take preventive action.

Report suspects

The police are hoping, for example, that suspicious observations will be reported. Unlike in the USA, for example, there is no holistic system that comprehensively monitors all crime and makes predictions, but some independent programs are in use. "We have our own programmers who create and adapt them," says Kriegs-Au.

Criticism too

The use of algorithms also provokes criticism. Angelika Adensamer, for example, a lawyer at the fundamental rights NGO Epicenter Works, says about the STANDARD that the effect of predictive policing cannot be verified.

"If the police drive to a suspected future crime scene where nothing happens, it is impossible to determine whether the prediction was wrong or whether the presence of the police prevented the crime," said Adensamer. Studies from Germany showed that the desired effects did not occur in the event of a break-in.


Crime is a social phenomenon and does not follow clear patterns. Therefore, even when technology is used, predictive policing is a "superstition". "The best criminal policy is still a good and fair social policy that starts at the roots of crime and not on calculated symptoms," says Adensamer.

In Germany and the USA, predictive policing is used in numerous cities - and there, too, is sharply criticized. The German activist Matthias Monroy criticizes, for example, that crime prediction reinforces phenomena such as "racial profiling" - that is, a targeted police approach based on certain criteria such as skin color. It is mainly black people who stop the police near potential crime scenes.

The Federal Criminal Police Office itself deliberately does not refer to the work as "predictive policing", but rather as "crime mapping", since the focus is on geographical factors. (Muzayen Al-Youssef, Markus Sulzbacher, November 3rd, 2018)