Duisburg.
On Tuesday starts with the police in Duisburg, a computer program that calculates the probability of burglary. Experts are skeptical.
This week, the Duisburg police makes first time by means of a computer software chasing burglars. Recently it has been reported a lot about those programs that the acts of professional burglars to predict now in NRW in the pilot authorities in Duisburg and Cologne. It is a nationally acclaimed trial, in accordance with high expectations. And that was a “problem,” says director Michael Jablonski Duisburg detective who even speaks of a “inflated expectations”. Because the program spits out any forecasts of individual deeds, no perpetrators, profiles or victims.
“That morning at the Düsseldorfer Straße 39 at 13.33 clock a burglary takes place, that’s not precisely”, explained the head the Kriminaldirektion now before the Working Group crime prevention. It is more about “the prognosis of a likelihood of offenses that may occur under certain conditions in certain places at certain times.”
Computer beats at three times increased probability alarm
What Commissioner Computer returns at the end, is a three-fold increased likelihood that there could be a break in a certain range. To define these areas, the authority has divided the city into 500 individual neighborhoods. Every Wednesday dubbed the Landeskriminalamt predictions for burglary vulnerable neighborhoods, so-called “hotspots”.
predict where burglars strike the next time: Duisburg is a model city for a new software to the police, which is used to the end of 2016.
On Thursday morning, the head of the Criminal Investigation Commissioner then decides due to its long experience, such as the police responding with strips that go around there, with undercover officers who lay in wait, or with officials, local residents to open french doors or point window. Predictive policing based on computer analysis So what now neudeutsch as “predictive policing” is making the rounds in the police
To:. More burglars red-handed catch
The goal: more burglars flagrante delicto to catch because directly a patrol car is nearby, or to repress the perpetrators at least by presence. “We will, however, not all hotspots that appear to us to edit. For us simply lack the people, “Jablonski sends anticipated. There are for this one-year pilot project, no additional staff, the Authority has to make do with what she
Even the division into districts is a balancing act. On the one hand they have to be so large that the number of cases at all still play a role. Secondly, it must be so small-scale that the police can ever respond to the forecast. The bare figures illustrate the problem: Each of the 500 districts still includes an average of 160 residential buildings with 500 households, mathematically there are only three burglaries in
The police not only helps geschehener fact – she wants to prevent crime and, before they happen. This software from Oberhausen ….
is lined helps the manufacturer IBM with “very comprehensive data sources,” the program says Jablonski. Above all, there are police data, all anonymous, but also external to household sizes, the demographic and social structure, to geographical location, allows the conclusions to be quick escape routes. In short, just everything can affect the burglary crime. The computer attempts then to calculate using algorithms from a pattern.
huge market for software publishers
Such programs are for software publishers a huge market. IBM, which was awarded the contract after a European tender, was one of the first developers in Zurich and US cities, the crime should be sunk by predictions. “You will find all statements that it is going very well, but you will not find a single published evaluation” Kriminaldirektor Jablonski keeps his skepticism is not behind the mountain. A corresponding study to demonstrate the efficacy to the NRW pilot supply, the result will submit the accompanying Association for Innovative Social Research in Bremen to the Interior Ministry, however, only one year
The key question here. How can we ever determine why the prediction has not arrived? Because the perpetrators saw the patrol car and are piled? ? Or because just the prognosis was poor
In addition, it would not even be unusual, the alleged success of the software should be without a flash read in the statistics: Just before the project start had the Duisburg police with mobile stations within a week advised more than 1,000 tenants and homeowners how they can protect their own home better against burglars. Whether these personal counseling ultimately was more effective than the forecast of the computer, but will hardly be ascertained at the end.
No comments:
Post a Comment