International Journal on Criminology Volume 4, Number 2, Winter 2016 | Page 76

Telling Tales with Inspector PredPol
akin to Behaviorism for Dummies.
Let us leave the books to one side, and focus on strategy. We discovered in
November 2013 that the US military in Afghanistan was trying out a “new predictive
model” called the Global Events Database, designed by a political science professor
from Pennsylvania State University. This software, we were told, “collects news on
the internet” and “catalogs all sorts of events, from local elections to genocides.” It
then extracts “short and long-term predictions” from the data that might be useful for
“managing crises” and “predicting conflict levels in Afghanistan.”
TV was quick to get in on the act: in the US, the series Person of Interest
(broadcast in France on TF1) had 14 million viewers on average on CBS TV in 2012.
The show tells the story of a computer genius who invents a machine that can thwart
terrorist attacks and predict heinous crimes. What a brilliant contraption! It is an
intelligent computer that even has feelings—artificial intelligence with a heart. Our
hero battles with the government, the mafia, and corrupt New York police officers,
and (of course) prevents crimes before they are committed. Person of Interest, say
the critics, has something of Minority Report about it in yet another reference to
the celebrated film, which is the immutable marker of the predictive rip-off. Even
US universities are now rushing to offer students diplomas or masters in predictive
analytics.
Predicting Crime: the Tales the Media are Quick to Tell
The articles in the press all claim that predictive policing “reduces the crime
rate by analyzing data on criminal offenses and where they are committed” or that
“the predictive method can be effective for people at risk.” In similar vein, they
report that: “Maryland is generating electronic predictions on criminals on bail to
see who will kill and who will be killed”; “Scientists and the police have designed
predictive systems that can tell which convicted killers will strike again”; “It is now
possible to use sophisticated computer analysis to predict where and when crimes will
be committed”; and finally: “Predictive policing programs based on algorithms and
historical data can guess the location and nature of future crimes.”
These articles have a strong whiff of marketing about them rather than critical
analysis—so much so that you would say they are more like editorial advertisements.
Take a look at the following examples: “Predictive software used by the police is twice
as effective as a human analyst working with the same data”; “A predictive policing
tool has reduced burglary by a third in 5 months in a Los Angeles neighborhood”;
and: “Throughout the US, dozens of police departments have already bought similar
systems.” For The Police Chief, the mouthpiece of the powerful International
Association of Chiefs of Police, “predictive policing marks the beginning of a new
era.”
75