Predictive policing

from Wikipedia, the free encyclopedia
This article or paragraph presents the situation in Germany and Switzerland . Help us to describe the situation in other countries.

Predictive Policing [ prɪˈdɪktɪv pəˈliːsɪŋ ] or German predictive policing refers to the analysis of case data to calculate the probability of future crimes to control the use of police forces. Both location-related and person-related systems are used, sometimes in mixed forms.

Predictive policing is based on various aspects of statistics and / or social research and is a well-known example of the application of algorithms in everyday life.

Theoretical foundations

Repeat victimization

The theory is based on statistical surveys, e.g. B. 4% (2%) of citizens suffer 44% (41%) of crimes and questioning of criminals (e.g. two thirds of burglars break into a building that has already been attacked). It follows that previous victimization is a good predictor of further victimization:

  • The more often a victimization was found in the past, the higher the chance of further victimization in the future.
  • Re-victimizations usually happen very soon after the previous events.

Routine activity approach

According to this theory it is necessary

  • of a motivated offender
  • a suitable crime object and
  • lack of protective mechanisms,

so that a crime can occur. Eliminating any of these factors will prevent crime.

Rational choice theory (rational decision theory)

The rational choice theory is based on rationally thinking and acting perpetrators who weigh the advantages and disadvantages.

Boost hypothesis

The boost hypothesis is perpetrator-oriented. It is based on the assumption that the perpetrator wants to keep the effort in the search for the next crime as low as possible and thus familiar areas are preferred.

Flag hypothesis

The flag hypothesis is object-oriented. The reason for revictimization therefore lies in the object itself and its properties, e.g. B. the visibility of a house, the times of absence of the residents, a non-existent alarm system or even entry and escape options.

Near repeat victimization

  • If one type of crime occurs in an area, the probability of subsequent offenses in that area increases.
  • The hypothesis was mainly tested on the crime of breaking and entering into a home.
  • Buildings on the same side of the street as the one that was approached first are more at risk than those on the other side.
  • The risk of victimization is not permanent, but only increased for about a month.
  • It is highest 48 hours after the first act.

functionality

Per offense group, e.g. B. burglary , patterns are repeated in terms of their recurrence over time and in the type of objects attacked (e.g. villa districts). These data are automatically fed into a formula as parameters for a geographically determined area. This creates algorithms and a probability calculation is used to make a forecast as to whether a district will be affected by the offense area again and when: In its entirety, an operator forecast is created. From a predefined probability within a district for the occurrence of a so-called trigger offense (e.g. burglary), an alarm is triggered at the control center . Police measures such as uniformed patrols, prevention officers or civil patrols are then dispatched to the areas in order to uncover the acts. Commercial providers of forecasting software often speak of success rates of 85%, although these statements are scientifically untenable and do not allow a comparison of different forecasting solutions.

Real operation

Predictive policing is already used by the police authorities in some German federal states . Some of these are temporary pilot projects or test operations.

place was standing software status
Berlin Jan 19 KrimPro The project, which started in January 2016 in two police districts, was expanded to all of Berlin in October 2016. The project ended in June 2017 with the transfer to everyday organization.
Stuttgart / Karlsruhe Sep 15 PRECOBS 6-month test run from October 2015 (costs € 220,000)
Hamburg May 16 Assessment of the state of affairs Currently research project
Lower Saxony Jan 19 PreMAP After the end of the first project in 2014 (in cooperation with IBM), a new project was started in November 2016. The test districts were Salzgitter, Peine, Wolfenbüttel, Wolfsburg, Hanover and Osnabrück. PreMAP will continue to be used in Lower Saxony.
Hesse Oct 17 KLB-operative After a successful test in 5 selected police departments, KLB has been operationally deployed across the whole of Hesse since October 2017.
North Rhine-Westphalia Jan 18 SCALE After a successful 2-year test in Duisburg, Cologne, Essen, Gelsenkirchen, Düsseldorf and Bonn, SKALA is used nationwide in North Rhine-Westphalia.
Nuremberg Apr 17 PRECOBS operational since 2016
Munich Apr 17 Precobs operational since 2016
Zurich, Baselland and Aargau Jan 16 Precobs 80% correct predictions, reduction of the quota in the test by 15% (8.7% city average), statistically not meaningful, initially no expansion of the test area.

In most police practice in Germany, only the crime of burglary is used for forecasts. In some cases, forecasts are already being made for other crimes, such as commercial burglaries or car crimes in North Rhine-Westphalia, whereby different aspects must be taken into account when preparing the forecast (forecast area, forecast time, type and occurrence of a crime, etc.)

criticism

Proof of the effectiveness of predictive policing is likely to be difficult due to the complexity of the influencing factors. A material-rich scientific evaluation of the Baden-Wuerttemberg pilot project on theft of burglary comes to the result for the forecasting software precobs: "To what extent predictive policing can contribute to a reduction in burglaries and a trend reversal in case development is difficult to assess even after the pilot project, despite some positive indications . "

Long-term use of the technology can also lead to problems with the underlying data sets. Because more police operations in a certain area usually mean that more criminal offenses are documented there. The use of the software therefore changes the number of cases - which in turn affects the future forecast. In the worst case, this creates self-fulfilling prophecies that have no added value in the fight against crime. This aspect must be considered when choosing the forecast offense and the appropriate method. In this context, the prognosis of controlled crime, which can be strongly influenced by police control activities, must be viewed critically. This phenomenon can not necessarily be observed in the case of burglary, for example.

The international secretary general of Amnesty International , Salil Shetty , sees the presumption of innocence threatened by predictive policing . He warns that discrimination against ethnic and religious minorities can be intensified through predictive policing.

reception

In Philip K. Dick's short story The Minority Report from 1956 as well as in the feature film Minority Report from 2002 based on this story , pre-crimes were discussed.

In 2017 Pre-Crime , a rather critical documentary by Monika Hielscher and Matthias Heeder, was released.

See also

Individual evidence

  1. ↑ ÖFIT Trend Show: Public information technology in the digital society. Predictive Policing (PDF). Public IT. March 2015. Retrieved February 25, 2016.
  2. ↑ Tobias Knobloch: Getting ahead of the situation: Predictive Policing in Germany . New Responsibility Foundation / Bertelsmann Foundation, Berlin / Gütersloh 2018 ( bertelsmann-stiftung.de [PDF; accessed on July 22, 2019]).
  3. ↑ Pollich, Daniela / Bode, Felix: Predictive Policing - On the importance of a (socially) scientifically guided approach . In: Police & Science . No. 3 . Verlag für Policewissenschaft, Wiesbaden 2017, p. 2-12 .
  4. ↑ Catrin Bialek et al .: Back to the Future . In: Handelsblatt . August 17, 2018, p. 42 .
  5. ↑ G. Farrell, K. Pease: Once bitten, twice bitten: repeat victimization and is implications for crime prevention. 1993. (PDF)
  6. ↑ I. Hearndon, C. Magill: Decision Making by House Burglars: Offenders' Perspective. 2004. (PDF)
  7. ↑ S. Cronje, JM Zietsman: Criminology. Pearson Education South Africa, Cape Town 2009, ISBN 978-1-77025-358-2 .
  8. ↑ criminally " particularly serious theft "
  9. ↑ Bode / Stoffel / Keim: Variability and validity of quality metrics in the area of ​​predictive policing. (PDF) In: DBVIS. University of Konstanz, 2017, accessed on January 1, 2018 .
  10. ↑ Where predictive policing is used. February 2016.
  11. ↑ Florian Stoffel, Felix Bode, Kai Seidensticker: Predictive Policing in Germany . 2018 ( uni-konstanz.de [accessed on May 1, 2019]).
  12. ↑ Police analysis in the Hamburg State Criminal Police Office. ( Memento from June 20, 2016 in the Internet Archive ) at: hamburg.de
  13. ↑ Florian Stoffel, Felix Bode, Kai Seidensticker: Predictive Policing in Germany . 2018 ( uni-konstanz.de [accessed on May 1, 2019]).
  14. ↑ a b LKA NRW: Project SKALA (Predictive Policing in NRW) - Results. In: Police North Rhine-Westphalia - State Criminal Police Office. LKA NRW, accessed on June 8, 2018 .
  15. ^ Kai Seidensticker: Predictive analyzes in space and time . In: Monthly for Criminology and Penal Reform / Journal of Criminology an Penal Reform . tape 100 , no. 4 , August 28, 2017, ISSN  2366-1968 , p. 291–306 , doi : 10.1515 / mkr-2017-1000405 .
  16. ↑ A. Gluba: Predictive Policing - an inventory. ( Memento of March 28, 2016 in the Internet Archive ) (PDF) 2014.
  17. ↑ Dominik Gerstner: Predictive Policing as an Instrument for Preventing Burglaries , Freiburg 2017, p. 85
  18. ↑ https://www.trendsderzukunft.de/predictive-policing-so-funktioniert-die-verbrecherjagd-mit-big-data/
  19. ^ Kai Seidensticker: Predictive Policing. Implementation and impact of crime forecasts. In: Police Info Report . No. 1/2019 , 2019.
  20. ^ Salil Shetty : Technology: force for progress, or tool of repression? Amnesty International's Secretary General Salil Shetty addresses Techfest in IIT Bombay on December 16, 2016. Retrieved April 1, 2017 .
  21. ^ Badische Zeitung: In the documentary film "Pre-Crime" computers predict crimes - Computer & Media - Badische Zeitung . ( badische-zeitung.de [accessed on November 3, 2017]).