August 25, 2009

Per Brad Stone (*), a US hacker with Russian backers is accused of stealing 130 million credit and debit card numbers, including from Heartland Payment Systems, a card processor. Meanwhile, per Haig Simonian and Joanna Chung (**), UBS has finally agreed to unmask 4,450 of its customers suspected of being US tax cheats. What can better illustrate the two extremes of mining privacy, whether for volume or value?

From another perspective, compare Heartland Payment Systems with UBS. When consumer privacy is a prized feature rather than a costly favor, security seems better, doesn't it? If Google, Facebook and the like insist on hoarding everyone's confidential profile, could they at least be made liable for providing Swiss bank strength protection against hackers, if not lawyers? Rules may be broken. They at least provide a useful warning.

This is why clarifying the rules of the game before the patient visits the doctor would help to cut healthcare costs. This advice extends to many other fields, from chads hanging on Florida ballots to female athletes bent on breaking records. Indeed read Alice Dreger's expert insight on sex (***).

Since "the I.A.A.F. has not sorted out the rules for sex typing", it "has called upon a geneticist, an endocrinologist, a gynecologist, a psychologist and so forth" to examine Caster Semenya. Next expect lawyers to meddle with the medal. Arbitrary by nature, often unjust when applied retroactively, rule making is seldom easy. Count on the specialist to reveal how rife with ambiguity his or her own chosen field can be.

Alice Dreger's title "Sex Verification: More Complicated Than X's and Y's" is a case in point. For optical character recognition experts, a sloppily hand-written capitalized X can be read as a Y or a V depending on the way its two bars intersect. At least there are only two sexes, aren't they?

Were Nature the sole source of uncertainties! Human frailty also casts a long shadow of doubt. Was a famous snapshot of the Spanish Civil War staged propaganda, asks Larry Rohter (****)? A recent visit by a French minister to a supermarket surely was, reveals Steven Erlanger (*****).

So to learn the truth, we perforce turn to trusted recommenders. Unfortunately, recommendations themselves can be faked. "Congressional investigators have uncovered five more letters sent to members of Congress that falsely claimed to be from charities", writes Stephanie Strom (******). And what about physicians who put their signatures on "scientific papers drafted by ghostwriters working for drug companies", as reports Natasha Singer (*******)?

There is nothing wrong in principle with self-interested recommendations. Monumental advertising has been around since the time of the Pyramids. Writing for Heathrow Airport may not make a new Horace of Alain de Botton but the latter is quite right to tell Andrew Adam Newman "[his] project recall[s] an era when patrons underwrote artists and writers" (********). Author independence though, commands a much higher value, commensurate to the fraud committed when signs of dependence are covered up or, in the case of credit rating agencies, waved away as immaterial.

The best model for independent recommendations is to need no income from such an activity. Nestor (1) did not dispense advice to the Achaeans for money nor do today's university professors who recommend their students. But despite Yochai BenKler's caution against exchanging money in such circumstances, there is nothing wrong either when the recommender is paid by the person who seeks advice. Greeks paid the Pythia for her oracles (2) and had only themselves to blame when they fell foul of her unconspicuous ambiguities.

In fact time dynamics and the prevalence of fraud are two excellent reasons why paid-for advice models should thrive on the Internet. In such a corrosive environment, those who will not pay to receive information may well want to pay in order to verify it per the rule of three. If he were inspired, Rupert Murdoch would notice newspapers are a natural source of recommendations.

For innovative business models, better turn to Oprah (3). No one to my knowledge has claimed her book club recommendations can be bought nor does she sells her advice to the public. Yet, by casting her very recommendations into a show, she earns good money from making them.

Still Nestor, Oprah and the Pythia do not exhaust sound recommendation models. What if it were possible to make those who are recommended pay their recommenders without compromising their independence? What if social networks were well positioned to implement such a scheme? Since corporations spend so much on marketing, it would go a long way towards retiring the Micawber business model, so dangerous to privacy.

Let a social network registers any user C who would be a paid recommender and any company A which values word of mouth recommendation. Then any time user B makes a purchase from company A based on C's recommendation, let A pays C an amount fixed and made public by the registry based on the purchase category. Can this new fangled manner of managing referral fees be efficient?

To avoid being discriminating against, companies only need to register their offer and since the recommender receives the same amount, no matter what he recommends, he or she should not be biased. Company A should not worry about being bilked since each payment requires a real transaction. And if user B arranges for friend C to recommend a purchase for the sake of getting the fee, chalk it up as word of mouth going from B to C instead of C to B.

Perhaps recommender C may fear that, the purchase made, user B will forget to tell C the transaction code company A collects from C as proof of recommendation. The registry needs only to allow C to give B part of the fee received from A. This is but another form of the rule of three whereby the information flow enables verifications and prevents participants from gaming the system.

Expect every user to jump at the opportunity. What's to lose? Like on eBay, also expect some recommenders to make it a day job. Even with low unit fees, one can earn a living on high volume since recommendations cost so little to distribute. As for the registry, a small cut assessed on all the fees can quickly add up to a genuine business model. Count on version 2.0 to allow multiple recommenders to split the fee from a single transaction.

Recommenders not so commendable would rather be paid twice, once, in secret, to endorse a product, then again to recommend it to one's friends. Hunting such double dippers and the companies which corrupt them would be the role of the FTC. Being a registered recommender while receiving endorsement money would be proof enough. Prosecution would not need to break the privacy of individual recommendations, a welcome relief.

Not that I expect social networks to listen to my advice. First I am no Nestor. Second Oprah has not invited me, yet. What a pity!

Philippe Coueignoux

• (*) ............... 3 Indicted in Theft of 130 Million Card Numbers, by Brad Stone (New York Times) - August 18, 2009
• (**) ............. US hails victory in UBS tax battle, by Haig Simonian and Joanna Chung (Financial Times) - August 20, 2009
• (***) ........... Sex Verification: More Complicated Than X's and Y's, by Alice Dreger (New York Times) - August 22, 2009
• (****) ......... New Doubts Raised Over Famous War Photo, by Larry Rohter (New York Times) - August 18, 2009
• (*****) ....... A Good-News Photo Op Embarrasses 2 French Ministers, by Steven Erlanger (New York Times) - August 20, 2009
• (******) ..... More Fake Letters to Congress on Energy Bill, by Stephanie Strom (New York Times) - August 19, 2009
• (*******) ... Ghosts in the Journals, by Natasha Singer (New York Times) - August 19, 2009
• (********) . Waiting at Heathrow, The Literary Experience, by Andrew Adam Newman (New York Times) - August 19, 2009
• (1) see Nestor in the Wikipedia
• (2) see the Pythia in the Wikipedia
• (3) see Oprah's Book Club in the Wikipedia