As recorded in Saint Matthew's Gospel, experts had told King Herod of the birth of a future rival in Bethlehem. Lacking more precise information, Herod zeroed in on the most efficient solution and ordered all babies under two to be put to death within the vicinity (1). Some commentators have made some chilling remarks on the lack of corroborating sources:
- the death of what, after further reflection, must have been less than two dozen babies only mattered to the victims, the survivers especially.
- Herod had such an established reputation for cruelty, mentioning a few more anonymous victims would have achieved nothing
Since Herod's times, humankind has not made much progress in the art of governance.
Ms Dunn, the soon to be former chairwoman of Hewlett Packard, knew she had the perfect way to find the source of a leak from her board. She ordered all potential suspects, board members and journalists, to have their communications analysed (*). Better than Herod, she did nab the intended target. But while she cared no more than he about violating the privacy of no more than two dozen people, she foolishly overlooked the requirement for victims to be anonymous and powerless as well as few. When a renowned name in venture capital, Mr Perkins, became aware of having been spied upon, though innocent, he did not lack recourse and the media could hardly have been expected to hush the story.
Scott Richter, the self-appointed King of Spam, did Herod proud too. Amidst a huge crowd of email users, you can be sure to find your elusive target, the very few who will welcome your solicitation. It can thus be very lucrative to abuse millions of powerless victims by sending them unwanted emails. Know to quit while you are ahead, though. Too many innocent victims attract the attention.
I admit to bias. Governments now rely on developments in mathematics and economics unknown to Herod. Here is an example of such progress:
- economics: in the absence of an explicit monetary loss by the victim, the cost of a privacy violation is zero
- mathematics: zero multiplied by a finite number, however large, is still zero:
- politics: profiling entire populations is a cost efficient solution to fight insecurity
For the sake of example, assume:
- a visible minority makes up 10% of the total population
- statistically, 1% of its members are criminals
- the crime rate of the majority is 18 times lower
Assume further, you are Chief of Security in a town of 20,000. Your mandate is to capture as many criminals as you can within a limited budget, which allows you to check 10% of the population. You may:
- capture 20 criminals by focusing on the visible minority of 2,000, which you can check in its entirety
- capture 3 criminals by random checks, which will cover 200 people in the minority and 1,800 in the majority
While actual figures may vary, profiling is shown to work... as long as you follow in Herod's footsteps. For what happens when profiling is used in such an optimal way? In our academic example, every minority member with a clear conscience knows that:
- he or she will certainly be checked by the security forces
- no majority member will ever be checked
- all apprehended criminals will be minority members
i.e. in the name of security, the minority is singled out as a criminal class and 99% of its members become innocent victims of racial discrimination.
Nobody can blame the Chief of Security for adopting profiling. Mathematics are on his or her side. Yet one viscerally knows he would be in the wrong. How can it be? Let us not romanticize minorities. Violence is always more tempting to minority members as a means to advancement. The real flaw must be the starting assumption that privacy violations have zero costs. Create a justified and lasting resentment in a self-aware fraction of the people and costs will add up. Maybe these costs are difficult to measure until too late. They are no less real.
At least in the fight against terrorism, our Chief of security is not in such a quandary. Profiling, which relies on statistics as used in pattern recognition, does not work for a simple reason. Terrorism is not a matter of statistics. If anything, it would be an application of game theory. If it is known mothers with babies are exempt from in depth searches before boarding an airplane, the mastermind of terror will look for the one mother ready to sacrifice herself and her baby. Same with nice grannies. As novelists gave us Miss Marple and Mrs. Pollifax (2), no doubt there too exists a suicide bomber in this somewhat unsettling sisterhood.
It is all the more distressing to learn that profiling is commonly used against logic, as analysed by Bernard E. Harcourt (**) from an academic perspective and Claude Ajit Moraes (***) from personal experience (see 05/16 fillip).
But when tackling ordinary crime, where statistics are meaningful, what can a well intentioned Chief of security do? Knowingly squander his or her budget or knowingly undermine democracy? In such a bind, hypocrisy is sure to help in the short term and who would cast the first stone? In the long term, I see no better solution than to put a price on privacy violations.
Once it is admitted that each individual is entitled by the US constitution to data rights, one can construe both a practice and a theory:
- innocents should received a compensation for every security check, according to a schedule approved by some legislative authority
- the compensation for innocents targeted through profiling should be higher, in proportion to the discrimination ratio (3)
- the right to violate specific individuals' data rights should be granted only by the judiciary branch and only to assermented officers of the law
This notion of compensation is fairly common. It applies both to property, in the case of seizure by eminent domain, and time, witness jury duty fees. Assuming a publicly supported level is set:
- our Chief of security is now in a position to optimize performance. Too much targeted profiling will quickly prove uneconomical while too little will leave the number of arrests wanting
- non authorized persons who violate individuals' eprivacy would automatically commit the theft of the corresponding fee while impersonating an officer, even before one looks into the legality of their means
Such measures would not of course deter those moved by personal prejudice or human greed. For such individuals, pretexting, profiling, spamming, all of the above will do. As we are wont to say in French, this all too human behavior is "as old as Herod" (4).
Philippe Coueignoux
- (*).......H.P. Spied On Writers In Leaks, by Damon Darlin (New York Times) - September 7, 2006
- (**)....Search and Defend, by Bernard E. Harcourt (New York Times) - August 25, 2006
- (***)Intelligence and infiltration are the keys to detecting terrorists, letters to the editor (Financial Times) - August 17, 2006
- (1) see The massacre of the Innocents in Wikipedia
- (2) check Mrs. Pollifax and Miss Marple in Wikipedia
- (3) i.e. if a minority member has a 50% probability to be checked versus 10% for the population on average, his or her compensation is 5 times higher.
Compensation is computed and paid at the end of each period during which statistics are kept.
- (4) "vieux comme Hérode", to say "as old as the hills".
|