When you work in security intelligence or law enforcement showing up late is usually a very bad thing. I am not talking about a neighbourhood party or a movie but rather getting to the scene of a crime or identifying a terrorist cell AFTER the act has been carried out. When that happens, people have already died. We are paid to stop these events from occurring, not clean up afterwards. When that happens we have failed.
So, to decrease the number of such events it stands to reason that being able to predict them BEFORE they occur is a good thing, right? Successfully predicting such incidents means people do not have to die. Who could possibly object to efforts to interdict criminals and terrorists before they act?
Well, there is some opposition and that opposition is tied to the fear that prediction is nothing more than profiling. And, surprise surprise, profiling is often seen as targeting certain people or certain groups of people based on characteristics like skin colour or religion or country of origin. I think most would agree (would they, though?) that overgeneralising on a whole section of society based on those characteristics is not something we want to pursue.
But efforts to bring prediction into security intelligence and law enforcement continue. The Los Angeles Police Department (LAPD) is using a commercial company named Palantir to develop a tool – already in use – called ‘PredPol’ (short for ‘predictive policing’) to help concentrate limited resources where they are most needed and where the possibility of criminal activity is highest. The system is essentially a sophisticated data driven analytic algorithm which its proponents maintain is not really that different from what cops have always done: collect information, analyse it and act, hopefully in time to prevent crime. Critics call it profiling, racist and ‘pseudoscience’.
What then of such an approach here in Canada? Apparently Canadian police forces are watching what their American counterparts are doing very closely but wary of ‘jumping in with both feet’ because of civil liberty concerns. Remember the ‘carding’ controversy in Toronto a few years back? That explains the caution.
As a former intelligence analyst and a counter terrorism specialist I am all for more data. The greater your data hoard the greater your knowledge – albeit not necessarily your understanding: the two are not synonymous. Having said that, there are a few things to bear in mind. One, having a program help wade through what you collect is not a bad idea. We in the business were swimming – perhaps drowning would be a more apt word – in data and if a system can help organise it for you then I am all for it. Two, however, I am not so sure how truly ‘predictive’ this is. Yes, it can assist in telling you what you have and maybe point you in a certain direction (geographic, gang or group affiliation, congregation of bad actors, etc.) but it does not really ‘predict’ anything. Just because area A has a historically high crime rate does not mean that a crime will occur there on a given day. There is far too much variability in human behaviour and decision making to ever (ever?) allow us to figure out in advance with 100% accuracy where the next event will take place. I have said it before and I will repeat it here: Minority Report was a novel/film, not a documentary.
In the end I am in favour of any tool that helps cops and spies do their jobs. These tools must, however, be as transparent as possible in order to gain public trust. If the public does not trust you, you are done: nothing after all beats a well-placed human source, whether it is inside a gang or a terrorist cell. The LAPD and other services would be wise to bear this in mind and to get the public onside. If not, their tasks will become harder, not easier, no matter how sexy the algorithm.
