International Journal on Criminology Volume 4, Number 2, Winter 2016 | Page 79

International Journal on Criminology
- First, and most importantly: today, as for the foreseeable future, uncertainty,
entropy, randomness, and chaos (in the scientific sense) belong to the realm
of the unforeseeable and unpredictable. In other words, anything that at any
given time may be possible, will not necessarily take place. Or, as Donald
Rumsfeld, US defense minister during the Iraq war, once famously reported:
“There are known knowns; there are things we know we know. We also know
there are known unknowns; that is to say we know there are some things we
do not know. But there are also unknown unknowns; the ones we do not know
we do not know.”
To claim that algorithmic observation can be used to model or standardize
unknown unknowns would clearly be intellectual fraud because what is
retrospective can never be predictive. Otherwise, having a list of all the
previous lottery draws would mean you would win the next one. If all the data
entered into an algorithmic cruncher was derived from past events, the result
would be nothing more than a probability, an often clumsy probability, a rough
and ready extrapolation: criminals were in such-and-such a spot yesterday at a
certain time, so they will be in the same place tomorrow.
- Since the first computers were invented, scientists have been obsessed by the
idea of prediction. At the outbreak of World War II, Norbert Wiener, the father
of cybernetics, tried to design a model that would predict the movements of
German fighter planes so it would be easier to shoot them down. He failed in
his attempts, it is said, because there was insufficient computing capacity.
- Over the last 30 years and more, various military research institutes—
including DARPA in the US—have spent millions of dollars on discovering
how to aggregate and combine masses of seemingly disconnected and disparate
data. Their aim has been to uncover correlations so they can carry out analyses
or make predictions on, for instance, future riots or attacks.
DARPA launched the Data to Decisions program in 2010 with a budget of
$92 million. The project was designed to develop an algorithm to connect, exploit,
interpret, and anticipate events using the mass of information stored, sold, and
exchanged on the internet—and always motived by the same old dream: to predict
social unrest, terrorist attacks and strategically significant events. But, judging by the
turmoil US foreign policy is experiencing from Afghanistan to Iraq, DARPA does not
yet seem to have found the predictive equivalent of the philosopher's stone.
78