IEEE Technology and Society Magazine - June 2019 - 45

crimes,8 it more often focuses on motives, and as
such employs psychological theories of choice and
sociological theories of behavior, and generally focuses on maximizing the likelihood and cost of penalties
for wrongdoing by stricter enforcement and harsher
penalties.9 The temporality also becomes deeply problematic here. There is an obvious utility in preventing
crimes before they occur, but our notions of individual
responsibility, guilt, and punishment rest on the commission of acts - of actually doing certain things that
constitute crimes - rather than imagining, desiring,
or simply being psychologically pre-disposed or circumstantially inclined toward doing things which
would be criminal. In some instances, planning or discussing criminal acts with others are acts that can
themselves constitute a lesser crime, such as conspiracy or solicitation to commit a crime, and a failed
attempt, e.g., to kill someone, can still constitute the
crime of attempted murder even if nobody is actually
hurt. But there are, and should be, different standards
for citizens who have committed no crime, those in
the act of committing a crime, those suspected of a
crime, those convicted of a crime, and those who
have served their sentences for a crime. How should
law enforcement treat "those 'likely' to commit a
crime"? And does the epistemic basis for that likelihood determination matter?
The classification of individuals also becomes critical
here. When we say that an individual is "likely to commit
a crime" is that based on their individual behavior and
actions, or because of membership in a certain demographic group? "Profiling" becomes problematic in the
latter case, when individuals are classified according to
population-level statistics and biases. Statistics are notorious for not distinguishing correlations in data from
causal reasons, and it would be unjust to treat people
with suspicion for coincidental correlations when the
underlying causal mechanisms for criminal behavior are
absent. This kind of profiling becomes deeply problematic when it becomes prejudicial, and the correlation is
taken as itself constitutive of guilt, or warranting a presumption of guilt, rather than a presumption of
innocence.10
8

For instance, adding better locks to protect property, such as ignition
immobilizers on cars, or making it more difficult to resell stolen goods [31].
In some cases, increasing the policing of crimes may actually have counterintuitive effects of increasing crime, according to an economic analysis of
the theft of art works [32].
9
Rarely do these approaches take into account the outright irrationality
or the failure of individuals to actually think about committing crimes in
rational terms. This is because cognition in the wild follows other lines of
reason and risk assessment, from inflamed passions, to rational biases, to
human necessity.
10
For example, if one is worried about a copycat bombing like the Boston
Marathon bombing, it might make sense to flag individuals who shop for
pressure cookers and backpacks. However, one should still presume there
is a reasonable explanation for this rather than presuming they must be terrorists for doing so [32].

JUNE 2019

∕

According to the U.S. legal system, criminal liability
and guilt depend upon a combination of actus reus (the
"guilty act") and mens rea ("the guilty mind"). That is,
one must actually commit the act for which one is held
responsible, and one must have had in mind the intention, or at least the awareness, that one was doing
something wrong, or should have known (as mere ignorance of the law is not a suitable defense). From this
perspective, one cannot be guilty of a crime before
actually committing the act, and should not be held liable for a crime not committed. And this is where precrime clashes with fundamental concepts of justice. If
society, and police, act upon precrimes, and those suspected of them, in the same way as already committed
crimes, then they are treating as guilty, or at the very
least as suspect, those who have not yet, and not actually, committed a crime. This is a profound form of prejudice, in which judgments are made not only before
relevant evidence of a criminal act can be obtained and
analyzed, but before such evidence can even exist.
Rather, judgement is passed on information derived
from statistical inference, patterns, trends and probabilities. But a statistical likelihood of an event is neither an
event nor an act.11 And it is fundamentally unjust to
treat someone as guilty of a crime they did not commit.
Moreover, it is powerfully felt as an injustice when individuals and communities are treated "as if" they are
guilty of doing something they have not yet, or not individually, done, based simply on their being members of
a category or demographic group. Indeed, the imposition of social categories can even give rise to new social
identities [35] - and thus machine-generated categories
are likely to create new types of people. This makes the
creation and designation of a "criminal type" deeply
problematic.
Still, there is a practical concern that law enforcement cannot ignore information about likely crimes
without sacrificing their duty to prevent crime. While the
scope and nature of that duty are themselves contested,
this is a powerful intuition. Indeed, it is the same intuition that motivates much data-driven management.
That is, if we can use historical data to predict future
trends and events, and thus better allocate valuable
resources towards fulfilling a mission or goal, then we
should do so. While not incorrect - certainly better use
of information can improve policing in many ways - if
pursued without careful consideration, caution, and
11
Just consider gambling on horse races, which historically gave rise to
modern statistics [33]. Oddsmakers go to great lengths to provide accurate
statistical predictions of the chances for each horse in a race. Yet, whichever horse is the favorite to win does not necessarily win - the actual
outcome of the race matters. The favorite only wins about 1/3 of the time
[34]. Gambling would not make sense if this were not the case - though in
many games of chance it can be argued that it is mathematically irrational
to place bets at all.

IEEE TECHNOLOGY AND SOCIETY MAGAZINE

IEEE Technology and Society Magazine - June 2019

Table of Contents for the Digital Edition of IEEE Technology and Society Magazine - June 2019