April 12, 2011

At first glance, the world is for ever swayed by chance and necessity. For a proof, one does not need to read Jerome Monod's account of molecular genetics (1). It is enough to listen to Tom Ashbrook's radio program about privacy (*).

Given digital advertising expenditures run in the billions of dollars, the economic and political pressure generated by market participants creates its own necessity. Michael Fertik, CEO of Reputation.com, spoke about "learned helplessness" to describe how callers felt about preserving privacy.

Yet, while millions of Americans use their mobile phones or access the Internet daily, comparatively few will suffer the type of harm to one's income or one's health against which Professor Latanya Sweeney was warning the audience. For those unfortunate ones, tough luck is a matter of chance.

As a result, the current lack of consumer data rights has all the appearances of a status quo firmly embedded in American society.

Prompting his guests to share their experience, their hopes, their proposals, Tom Ashbrook made them speak of "informed consent", "privacy market", "contracts" (2), "outsourcing of personal data processing" (3), "reputation and privacy" and much more. Listeners beware. The format constrains participants each time to show but the tip of the iceberg. Still this was an informative discussion. What should one make of it?

Chance is a convenient shield against potential liabilities. That should be our first lesson.

Take Raj Rajaratnam's planned defense as reported by Peter Latman (**). Galleon's CEO "practiced the so-called mosaic theory of stock trading", mixing multiple sources, "Wall Street research and public data, not inside information". If he received any of the latter from too talkative sources, it played no role in his investment decisions. In focusing on a few lucky stock picks and selecting evidence to imply a causal relationship when there is none, the public prosecutor is unfairly distorting the record in hindsight. Dismiss such bias and mere possibilities are no proof of responsibility.

A similar weakness appears each time one tries to single out causality among a multiplicity of potential factors. By enabling data to be copied at zero cost and multiplying the ways in which to "observe" consumers, our Information Age allows personal data holders to owe zero responsibility to personal data owners by diluting any causal link between any specific personal data copy and any instance of individual harm.

Still individual harm does happen and its global cost can be computed. Couldn't the latter be assessed collectively before the fact on all potential sources of leaks, as suggested by Professor Sweeney? Certainly. As privacy violations are but a form of pollution, compensating for eprivacy risks can be modeled on known control mechanisms to fight air and water pollution, distributing responsibility as one does a probability.

We must not however forget the essential limitations of statistical models. While delivering scientific results from the hypotheses on which they are built, their practical value depends on the validity of these hypotheses. Worse, however inconvenient, single events can always be subjectively dismissed as outliers too rare to consider. Statistical models run the risk of debasing science into fads. Gail Collins has fun mocking "medicine on the move" with "recommendations on when to get a mammogram, which seem to fluctuate between every five years and every five minutes" (***).

As Professor Sweeney further suggested, it would be preferable if the price due to an individual whenever related personal data is collected by another party was freely agreed upon between them. Part of this price would represent the risk of subsequent exposure born by the individual. This however requires that consumers own data about themselves, that it never be collected without an "informed consent", not bundled with some other desirable transaction but according to a valid data "contract" and that personal data holders be subject to fiduciary duties.

Fat chance. What aggregator would willingly pay genuine market price for data she can rob for free? A plunder economy generates its own necessity. So instead of being paid for personal data, why not be real and accept to pay for privacy. Nick Bilton does not hide "it can be difficult to delete" (****). Still, though ill winds can blow away any umbrella, umbrellas are useful. Privacy umbrellas are what Michael Fertik sells for a living.

This raised an issue. Even though privacy is at stake when one manages one's reputation, a personal fact may be kept secret for two very different reasons, fear of blackmail or desire to negotiate from strength. Michael Fertik is right to say the two overlap. Being a compulsive smoker may both raise one's value as a marketing target for tobacco addiction cures and harm one who seeks employment in an oil refinery. Still the reputation of bankers as third parties one can trust is so low nowadays, he should think twice before turning into a data banker and advertising his "data vaults".

Economic pressure may impose its necessity over isolated individuals. But as Michael Fretik said, "the status quo is an accident". Necessity itself is but a social construct, indeed a product of proanocracy, whose laws are a barely veiled quid pro quo of campaign financing.

There is no more vivid proof of this second lesson than to show an alternative. In order for service and goods to benefit from ever more precise consumer targeting, it is unnecessary to put consumer personal data at risk in the hand of intermediaries, trusted or not. One only needs to give each consumer his or her own confidential "personal store" to process "personal data" in response to external interactions away from any prying eye.

I claim to have developed such a confidential market mechanism (4). Will my claims ever be put to the test and, if so, proven right? Who knows?

Literature and real life are both replete with such independent inventors as myself. Am I a Casaubon (5), hopelessly lost in antiquated pursuits, a Lydgate (5) or a Séchard (6), swept by circumstances above their strength into a comfortable mediocrity, or, to take real examples, a John Harrison (7) or a Maurice Allais (8), who both lived long enough to see their work begrudgingly recognized? Only time will tell.

Necessity hardly dictates the outcome. Whether to attribute it to chance or to Divine Providence is not a matter of science, but by necessity of faith.

Philippe Coueignoux

• (*) ....... Paying for Privacy, by Tom Ashbrook (National Public Radio) - April 6, 2011 with the participation of Nick Bilton, Esther Dyson, Michael Fertik and Latanya Sweeney
• (**) ..... Prosecution Rests In the Galleon Trial, by Peter Latman (New York Times) - April 7, 2011
• (***) ... Medicine on the Move, by Gail Collins (New York Times) - April 7, 2011
• (****) . Erasing The Digital Past, by Nick Bilton (New York Times) - April 3, 2011
• (1) for more details, see Chance and Necessity in the wikipedia
• (2) see for instance our 2011 submission to the Federal Trade Commission in response to its consultation on privacy
• (3) see for instance the case illustrating our 2006 lecture on Handling of Medical Records
• (4) see my US patent portfolio, US Patent Number 6,092,197 and US Patent Applications 2006/0053279 and 2009/0076914.
• (5) for more details, see Middlemarch, by George Elliot in the wikipedia
• (6) for more details, see Les Illusions perdues, by Balzac in the wikipedia
• (7) for more details, see John Harrison in the wikipedia
• (8) for more details, see Maurice Allais in the wikipedia