April 16, 2013
Having sold Manhattan to the Dutch for a pittance given its subsequent increase in value, the Lenape are believed to have made a really bad real estate deal. This is the wrong lesson to remember as, at the time, a few acres in such a sparsely populated region could not have been worth much, especially to a people to whom private ownership of land was unknown and whose economy was largely based on hunting and gathering.
We should be very careful indeed when judging the business acumen of the Lenape. Today's consumers have no right either to their private data, whether observed or inferred, and routinely "volunteer" what they know about themselves for even less than the Dutch felt obliged to offer. Yet, unless the poor Lenape, we already know how vital all this data is to our economy. Big Data keeps proclaiming it and right it is on this score.
Like real estate however, eprivacy is not as simple as it seems. For too many advocacy groups, it is mainly a matter of bringing lawsuits, for too many entrepreneurs, it is but a market to milk, for too many of a scientific bend, it is nothing but a gun versus armor fight in which gun research receives most of the money. Each perspective is needed but can but fail in isolation.
So in making its newly created "Personal Data Values and Policies" chair into, shall we say, a large sofa, to seat four professors, one in computer science, one in economy, one in philosophy, under the coordination of one in law, the Institut Mines-TÚlÚcom shows a good grasp of the issue (1).
In a time of budgetary constraints and in line with the American model, this chair is financed by organizations with a strong interest in the topic. Conflicts of interest are sure to follow. As a big bank, BNP Paribas is hardly neutral when it comes to eprivacy. A few years ago, the head of online operations for a French insurance company cynically told me that consumers should entrust their personal data to his company for processing it on his servers but that it would never entrust any proprietary algorithm to them for executing it on their computers. As always, Might makes Right.
However the presence among these financial partners of the CNIL, the French Data Protection Authority, is a good sign. The first conflict of interest being between BNP Paribas and the CNIL, one promotes peace by bringing the two opposite sides at the same table. It also proves the CNIL has abandoned the purely defensive view of its mission which so dismayed me ten years ago, when I thought it might prefer solutions to repression.
I therefore wish this initiative well. May the program outlined by the leading quartet (2) avoid three pitfalls which could threaten its future.
Claire Levallois-Barth sums up its overall goal as proving the benefits of privacy for both companies and citizens. Yet, as a recent report cogently shows, the state will go broke if it cannot find a way to tap the flows of the personal data economy. So one danger would be to ignore the direct interests of the state and Claire Levallois-Barth should not forget to enlist a tax law specialist to supplement her own legal expertise.
Another problem would be to put too little emphasis on the notion of recommendation, the real basis of an economy based on data, since all data must be first and foremost filtered against lies. Only Patrick Waelbroeck mentions the term and only in the context of digital id's. While there can be no real recommendation if id's cannot be trusted, the final value resides in the recommendation itself, not in any specific tool which makes it possible. As an economist, Patrick Waelbroeck speaks of "personal data markets". He should encompass recommendation based markets in his analysis.
That an individual recommendation could be monetized by its very recommender is of course anathema to those who hold money taints markets as something intrinsically evil. The only issue is to insure rewarding recommenders is free in practice of the conflict of interest which exists in principle. The challenge combines technology with ethics and sociology and should be an explicit task for Pierre-Antoine Chardel and Maryline-Laurent.
Suppose nothing is done in these regards and this promising mission may well miss its target in two singular ways.
First it will copy the past posture of the CNIL. Deprived of a direct and dynamic economic role even their advocates dare not examine, consumers will be relegated to an assisted status. The more their dignity will be proclaimed and protected, the more they will look like minors, and act the part.
On the one hand, denied ownership over their own performance, the so-called "personal traces", consumers will be the object of privacy laws. On the other hand, other intangible assets, whether cultural, like copyrights, or purely economic, like patents and trade secrets, will continue their autonomous and tumultuous life as legal property. Alone entitled to hold proper property rights, corporations and stars will be our new noble class.
Yet piracy violates copyright and privacy laws in the same spirit. And what difference it makes if piracy comes from corporations rather than from individuals, besides the fact the former is executed more systematically, the latter prosecuted more heavily and more automatically?
Niall Ferguson finds "too many property rights and too much law [in the US]", according to John Kay, (*). Actually we would benefit from new rights and new laws if only we could defeat the two obstacles the latter skillfully identifies. "The rule of the law operates in favour of bullies and the rich and privileged", i.e. pronaocracy. And, for Mansur Olson, "sclerosis aris[es] from the conflicting demand of too many established interests".
Segregating privacy from property and denying consumers the monopoly of their own data makes property law even more of a shambles. For it biases economic exchanges, entrenches pronaocracy and distracts from solving the central conundrum of our Information Age, how to equitably reward all economic actors, whether major or minor, when the value comes from sharing data and the power from monopolizing its access.
Second the mission may succumb to a subtle lie, at work whenever Big Data is concerned. Sales can be decentralized yet personalized online without anyone having to share anything about oneself (3). Only marketing requires one to centralize shared personal data, to develop its models.
The latter is the real privacy concern but how big is the demand? Read Michal Kosinski's study of how to "infer" highly confidential data such as gender, race, religion and sexual orientation, from innocuous looking "volunteered" data such as Facebook "likes" (4). For his model to reach over 75% accuracy, he needed less than 60,000 volunteers, each averaging 170 likes over some 55,000 possibilities.
Why insist in accessing 1 billion users' entire profile when 60,000 users' partial data is pretty good? Because it costs nothing. If Nielsen did not have to pay its television panels, wouldn't it record what everyone views? Biased property law leads to slippery science and rampant discrimination.
Forget Manhattan. Their property law no match to Western civilization, the Lenape are now parked into Indian Reservations. Will we be too?
- (*) ........... Property rights alone will not bring about prosperity, John Kay (Financial Times) - April 10, 2013
- (1) Chaire Pluridisciplinaire sur les Informations Personnelles - April 2013 - in French (an English version should be forthcoming)
- (2) Point de vue des Chercheurs, by Claire Levallois-Barth, Maryline Laurent, Patrick Waelbroeck, Pierre-Antoine Chardel - April 2013 - in French
- (3) for more details, see US Patents Number 6,092,197 and 7,945,954 and US Patent Application 2009/0076914
- (4) Private traits and attributes are predictable from digital records of human behavior, by Michal Kosinski
.... (Proceedings of the National Academy of Science of the United States of America) - early edition approved February 2013