One more go round on this false issue of ‘Racial Profiling’…
There is no such thing. It would be pointless and idiotic to ‘profile’ race. Race is assumed one of the most directly observable characteristics of a human – and generally, there would be no reason to construct a profile to predict or classify race. While it is true that some misclassification is present in the everyday typing of race by visible characteristics – for the most part no one needs to use a scorecard or profile to make a pretty good guess at race.
A profile is a tool, a template, against which to compare individuals and assess the probability that they are involved in something you want to scrutinize and interdict (drug smuggling, terrorism, etc.). It is not a perfect predictor – just an efficient screening device. To be minimally efficient it needs to be only slightly better than random chance assignment. A good system needs to be much better than random chance – not only because of the extraordinary levels of damage that a terrorist can accomplish – leveraged by nuclear, radiological or biological technologies – but also because 99.9+% of those screened are not terrorists and to burden them unduly lays a very heavy cost in terms of inefficiency on the vast majority of the traveling public.
Before 9/11/2001 there was a need to screen air travelers and border crossers for drug courier activity. Hence, there were ‘drug courier profiles’ – models that aimed to sort out the mass of air travelers or border crosses and rank them on the probability that they were drug couriers. These models were based on actual experience of persons stopped in airports or at border crossing stations. You build a model with multiple variables (origin-destination pair, elapsed trip time, time at destination, method of payment, country of origin, number of days in advance of flight payment is made, etc.) which, taken together, form a predictive score than can be used to segment high probability (…of being a drug courier) from low probability.
The entire purpose of profiling is to focus limited police, border patrol, DEA or FBI (and now DHS and TSA) resources on those who are most likely to be drug couriers (or terrorists, or illegal aliens, etc.)
The use of the term ‘racial profiling’ has been an attempt by Jesse Jackson, Al Sharpton, et. al. to make an issue out of something that does not exist. It is really a very silly thing to assert that such a thing exists or is a serious issue. To allow the continual trafficking of such a false label in the major media without challenge is shameless.
If police were arbitrarily stopping individuals simply because they are black – they are not using a profile. And if they are using a properly constructed profile – they may be stopping black people – but not simply because they are black. Rather it would be because other characteristics of their travel, or their situation at the time of a border crossing, or their manner of travel were highly correlated with drug courier or terrorist behavior.
Building a screening model for air terrorism is difficult. The 19 who did commit terrorist acts on 9/11/2001 were cast against a background of perhaps millions who were to fly that day who did not. The sparseness of the data makes building a statistical model difficult. Other techniques – such as neural networking can be used to mine data – but there again, the number of instances of actual terrorists observed to contribute data for model development is very small. Terrorists of the al Qaeda variety have developed organizational cells and methods (cash exchange by hand) which minimize the transactional trail which can be monitored.
In lieu of statistically or neural network derived models – judgmental models will have to do. A rational development of a judgmental model is accomplished by taking the common characteristics of known terrorists and generalizing them as characteristics in a model or profile to screen for likely terrorists.
Other efforts to screen terrorists will come from analysis of patterns in credit card usage and perhaps even cell phone usage or even electric power consumption. Patterns in large datasets can be examined to find the particular elements that isolate a high concentration of likely terrorists. Then conventional law enforcement and investigational techniques can be applied to this highly suspect sub-population.
The ignorant reflex against profiling is just that - but if we are stupid enough to listen to those who reject profiling as a concept (e.g., the ACLU)- we are in danger. It's that simple.
Similarly - if we were to listen to those who complain about the hypothetical 80 year old white grandmother (in a wheelchair, no doubt) being scrutinized do not realize that al Qaeda and Hezbollah and the other terrorist organizations are directed by some folks who, despite being derange, are quite intelligent. If they knew that there was a rule excluding 80 year old white grandmothers in wheelchairs from scrutiny - they would immediately work on recruiting one (wittingly or unwittingly).
Any screening system has to be a series of screens. Algorithms which target risk, combined with random assignment (there goes grandma), as well as behavioral and judgmental screens can e combined to create a system which minimizes risk.
FWIW, I have no problem with profiling.
bttt