Posted on 18.12.2014 by Jan Andrš
Predictive policing sounds like a crystal clear idea, but there is always a catch. In this case, two.
In our last article on Predictive Policing, we discussed the relative simplicity of the system and questioned its usefulness for more serious crimes like murder or rape. And two big concerns remain: injustice and reasonable suspicion.
Biased data creates biased systems
The usual PredPol (cool abbrev., huh?) tactic is to identify danger zones and send policemen there when they are not needed elsewhere. And those programs reportedly have positive results. Although, as activists point out, police statistics are usually a little biased against certain minorities, and higher crime zones are usually in areas populated mostly by those minorities. For example:
But if historical arrest data shows that the majority of arrests for marijuana crimes in a city are made in a predominately black area, instead of in a predominately white area, predictive policing algorithms working off of this problematic data will recommend that officers deploy resources to the predominately black area — even if there is other information to show that people in the white area violate marijuana laws at about the same rate as their black counterparts.
We are not actually stating that police statistics are biased, but since the data for predictive policing programs flows only from police statistics, the system depends on them (almost) entirely. And if there is anything wrong with the statistics, the predictive policing program will actually make them a lot worse. When the data is focused against any group of people, the predictive policing system will, again, target the very same group of people. Police will be more present in “heated” areas and will also make more arrests there. So inhabitants of those areas might be caught in a “vicious spiral” of sorts.
Predictive policing and “reasonable suspicion”
Law-enforcement agencies claim that predictive policing consists mostly of crime deterrent activities – the very presence of police officers should prevent crimes from happening (and stats mentioned above suggest it does). But crime deterrent activities include not only patrolling, but also searches based on “reasonable suspicion” (a legal term used in many legal systems all around the world, in the U. S. covered by the 4th Amendment of the Constitution and subsequent ). Andrew Guthrie Ferguson from UDC argues that simply being in a zone is not enough: “Because predictive policing does not provide personal knowledge about an ongoing crime, or particularized identification of the suspect involved, it cannot support the weight of reasonable suspicion.” So any search based only on being in the zone is not enough. Ferguson also suggests that predictive policing forecasts alone, will not constitute sufficient evidence to justify reasonable suspicion or probable cause,” but instead will be seen by courts as a “plus factor” in making such determinations.
Even though any verdicts from courts, suggesting which way the law is going to lean, are yet to come, we know for sure that those two issues are going to decide the future of predictive policing.
Sources and further reading on the topic: