The MPs charged with guarding prisons in Iraq are indeed poorly
trained and woefully ill-equipped to perform their duties, yet
even had they received proper training there would still exist a
problem at a more rudimentary level. The system architecture of
the prisons of Iraq and, more generally, of the world, suffer
from design flaws that make them hopelessly prone to barbaric
abuse. While the US should be ashamed of its treatment of
prisoners in Iraq, similarly terrible things happen here on US
soil in domestic prisons all the time.
One of the marks of a good system is the implementation of
protocols that are resilient and fault tolerant. Small defects
should not result in catastrophic failure. No system will
manage to operate flawlessly, but breakdowns should be well
isolated, and provisions for automated repair in place.
The probability of a system containing defective components
rises in proportion with the number of components in the system.
Consequently, the larger the system, the more adaptive it must
be to component failure. With an organization the size of the
US military, consisting of a number of individuals on the order
of millions, one is virtually guaranteed to have some rotten
apples. As such, it is absolutely imperative that one assume a
model of distrust and craft a system under these auspices.
The problem with prisons in general is that they place an
enormous amount of trust in guards. In some ways this is
unavoidable, yet there are some interesting strategies available
for embedding safeguards against abuse.
Before we continue, we must first consider the asset we are
trying to protect, the threats to it, and the resultant risk...
The asset in question is the dignity of prisoners. This,
however, is really only a sub-goal. What we are really trying
to protect is the reputation of the US and its credibility in
the world at large. The threat to this is abuse of prisoners by
guards. The risk, it would seem, is enormous; not only is the
probability of abuse quite high, but the damage to the public's
perception of America in the case of such transgressions is
great, and the consequences tragic for both US military
personnel and Iraqi citizens.
How then might we go about mitigating the risk? The answer lies
in leveraging different guards against one another.
Consider the present situation... In various prisons complexes
guards are assigned details. These details can last for an
indeterminate period of time. During these assignments, guards
have substantial opportunity to probe one another for
personality traits. Nobody is going to come out and expose
his/her true nature in one fell swoop, but in a tit-for-tat
fashion, guards sharing details can gradually feel one another
out without putting themselves at substantial risk for censure
or punishment. When an entire detail consists of sadistic
guards, disaster is imminent.
The obvious solution is, of course, to somehow manage to make
certain that no shift fails to have at least one decent
individual that guarantees honesty. This solution is simple and
functional, except for one niggling detail: it's impossible to
actually ascertain the reputability of an individual until it is
too late. Ask any HR person and they will tell you that
successful personnel screening is an enormously difficult task.
Far better would be a system in which one could be reasonably
guaranteed of integrity despite the exclusive presence of
malicious individuals on a particular guard detail. In fact,
such a scenario is possible; one must simply exploit the
properties of the guards in a way that brings about
accountability regardless of the guards' desires to misbehave.
What if, instead of hoping that guards behave decently, one were
to foster an environment of mutual distrust that kept all
involved parties honest? To accomplish this, the general
strategy would be to minimize the opportunity of trust building,
and maximize the tendency to snitch on those who engage in
malicious behavior. This is counter to the model that the
military uses in most situations, but this represents a sensible
exception to standard protocol as the task of guarding prisoners
is fundamentally different from that of any other task in the
First and foremost, in lieu of assigning guards to specific
details for extended durations, they should instead be assigned
to details on the order of days. The actual length would be
contingent on psychological experimentation, but the general
idea would be to make guard rotations occur frequently enough
that the time between rotations spans less than the period
of time required for guards to come to trust one another.
As an added layer, one might also provide generous incentives to
those who proffer evidence of malign behavior. Not only should
abuse of prisoners be punished, but audit of abusers could be
rewarded. When a guard provides a credible report of another
guard's abuses, the former ought to receive some kind of
In analyzing such a system, a tendency toward resiliency becomes
evident. Whereas before one required at least one honest
individual for a detail to be kept honest, now one could
ostensibly have an entire detail composed of unscrupulous
individuals and still have it kept honest as the result of mutual
Imagine the thought process of a guard prone to abusing
prisoners. If he has not known the other guard(s) on his detail
for a time sufficient to feel them out, then he is apt to act
with great trepidation in his treatment of prisoners. He will
be wary of other guards perhaps hankering to make a profit from
turning him in for abuses. At the same time, he is (assuming a
proper incentive structure) more than happy to cash in on the
indiscretions of his fellow guards.
Instead of a gradual increase in comfort with one's fellow
guards, there is instead a constant distrust that is resultant
from the fact that one never manages to acclimate oneself to a
given set of co-workers. Whereas in the present system the
probability of abuse increases with the passage of time, in the
proposed system it stays at a more or less fixed low value,
something more precisely described as a saw-toothed function
with a very low maximum amplitude.
The beauty of such a system is that it could be composed
entirely of malevolent guards, and yet still run things in a
decent fashion. Any ill-intentioned guard is apt to look
askance at all of his compatriots, as well as view them as a
chance to augment his income.
Of course, There Ain't No Such Thing As A Free Lunch. Such a
system would impose both financial costs on the operators of
such a system, as well as psychological costs on the guards.
Man-power and transportation capacity would prove the main
logistics of concern. For optimal functionality, no two guards
would ever share a shift more than once. Not only might this
prove problematic in the case of a limited guard population, but
furthermore there is the issue of transporting guards. Instead
of providing constant housing to guards at a specific location,
both transportation and transient housing must be arranged.
There also exists a high human cost for the guards of the
system. Starting a new job is always stressful, and the
proposed system more or less simulates having a new job on an
ongoing basis. Guards would never grow at ease with their
co-workers, and would also be largely incapable of settling down
in one place. This stress could translate into reduced
effectiveness in their ability to perform in the work place. As
a corollary of these psychological costs would also be a
financial cost: less desirable work conditions would dictate
better pay to keep the labor pool adequate, and turn over could
be an issue as well, resulting in significant expenditure on
training. The only way to mitigate the psychological strain
would probably be to provide for long vacations to guards to
recover from tours of duty.
The question, then, is whether the benefits outweigh the costs.
Actually, more accurate is the question of whether the perceived
benefits outweigh the perceived costs, and if political factions
will be adequately aligned. In the case of prisoners of war, it
may very well be deemed a worthwhile trade-off; the costs of
political backlash are enormous, the military is fairly well
funded, and there is a willingness to impose severe
psychological strain on soldiers. In the case of domestic
prisoners, people are apt to be tragically disinterested in such
a proposal; nobody cares about these prisoners, the prison
systems are woefully under-funded, and there exists an extremely
lax code of discipline and a lack of oversight for the guards.
Fundamentally, people show interest in what the media covers and
little else. In the hustle and bustle of making a living, it's
easy to miss the things that are right under one's nose,
noticing only the things to which the media holds up a
magnifying glass. It took only a single photograph to inflame
the whole world about atrocities against Iraqis occurring on the
other side of the planet, and yet such barbarism is common place
right here in our own backyard.
In light of such facts, it is difficult to believe that calls
for Rumsfeld's resignation are something other than political
opportunism. If the dignified treatment of prisoners were truly
a sincere issue in our body politic, then there is plenty of
cause for concern on a daily basis. The mundane just doesn't
make for good news.