November 10, 2009

I may be wrong but of the Aeneid and the Georgics I am willing to bet the former has proven the more popular work by far. I once confessed a preference for epics and History is on my side. Heroic looking looters have targeted rich farmland because that was where the money was long before evolution turned them into bankers. Now that data has replaced food as the font of value, we simply witness a new kind of plunder economy.

Yet too many laments over insider trading and privacy violations may be counterproductive. What if we were to focus on the lowly dataculturists?

Take Gerrit Wiesmann's article on a German entrepreneur, Philipp Goedeking (*). His company sells air traffic data to airports and airlines. "The total market [...] could be worth 100 millions of euros in revenues a year". Isn't it amazing? One would imagine his clients to be better informed than any outsider. Until, that is, one realizes "while individual airlines know their own passenger numbers, getting an overall picture is next to impossible".

Gerrit Wiesmann explains how this dedicated dataculturist takes the data available from public but fragmentary sources and compute what the actual air traffic had to be in order to be consistent with these inputs. As any entrepreneur, he may well boast his proprietary algorithm has created real barriers to entry. But has he realized the best barrier is his own modest market estimate?

Were the market worth $10 billion, Google would only have to offer all airlines a free universal reservation system and aggregate the data they would confide to its capacious vaults. The magic is that, in possession of all this information, Google would not need any fancy algorithm at all. Even better, Google would offer the data to airlines for free. Why not indeed when it could make a killing selling highly targeted advertising to Boeing and Airbus? "Dear Lufthansa, want to steal traffic from the competition between Frankfurt and Singapore? Click and look up our Boeing 777-200LR."

This is not so hypothetical a case given the current push for creating a US registry of medical devices, reported by Duff Wilson and Reed Abelson (**). However, as long as society continues to ignore data rights, issues are sure to ensue, putting dataculture sustainability in doubt. Especially if ideology erects additional obstacles. Why would data collectivization by the state be any worse than under Google?

To study sustainability, the first step is to determine who originates the data and the interest of such originators. A solution which would ignore the results of this analysis is certain to end in failure even if the timing of this event is unknown. Why dataculturists would react differently than farmers?

In the case at hand, the seed data is to be found in bookings. Its value to airlines is twofold. One part lies in the market research they derive from data available to them, the other in the competitive disadvantage which burdens those whose data become public. There is a twist however, redolent of the Prisoner Dilemna (1). Were all airlines to agree to share their proprietary data among themselves, they would very likely benefit overall.

ePrivacy ought not to be the privilege of the brand. But why solely focus on lowly individuals if large corporations help illustrate the real nature of dataculture? Airlines teach us two clear lessons. One we already know. When data influences decisions, it gains value from remaining confidential. Privacy is about money. The other lesson is that sharing should never be rushed into. If costs go to some and benefits to others, it creates victims bent on sabotage and revenge. If its benefits accrue to an established oligopoly, it will be suspect to antitrust enforcement agencies.

The European competition commissioner would view traffic data pooling as an incipient market share agreement among airlines. Short of pooling though, each airline better protects its own data, thus leaving an unsatisfied market niche for independent dataculture, witness Philipp Goedeking.

That competition and innovation can be both enhanced in a privacy friendly legal framework is encouraging. The US healthcare industry presents a different picture though. Justin Baer reports "IMS Health, a provider of prescription-drug data, agreed to sell itself [...] for $5,2bn" (***). Unfortunately that valuation is based in large part on stealing data from prescribing physicians and on the benign neglect of most US states.

When major economic forces are vitally threatened, they tend to stick to the status quo. As Ian Brown writes in his excellent review of the "creative industries", "much of the effort of the media industries to move into the information age has gone into lobbying, blocking new technologies and suing fans" (****). As it happened, data aggregators cum looters are naturally close to candidates running for high office. Expect some epic battles.

IMS Health is an incumbent. Will new patient health record services such as the one launched by Google be any better? Allow me to be sceptical. Google is no Twitter. Rather than adopting a Micawber business model, its health division can afford to live off its corporate patron's fortune. But subsidized economies are not a model for sustainability, are they? Sooner or later patients will pay for Google's financial patience and how can they pay for a free service if not with their personal data? By then it will be too late to reconsider as de-identification will prove to be a myth.

What about the implementation of electronic medical record systems by healthcare providers? In his field report (*****), David Talbot shows all the promises of this approach. Notwithstanding my previous examples, data sharing can and should indeed be great for dataculture. My point is that the potential for healthcare benefits should never excuse one from making an impact study on the data originators, the patients and their physicians.

Take David Talbot's first application, enabling doctors to access a patient's file "without borders". For the patient, this type of sharing entails a privacy risk but also direct, at times dramatic benefits. But if no value is attached to the data exchanged, it is not so for physicians. Since the cost of adding to a medical record is born by the writer, the physicians who take time to write more informative notes are taken advantage of by the free riders who limit themselves to the minimum they can get by. Without further measures, to encourage sharing is to encourage a race to the bottom.

Another application is to determine what medical treatments are the most appropriate, especially the most cost effective. For a patient, the privacy risks remain entire while the benefits become uncertain. Unless all healthcare actors were to forgo perks and profits, why impose a social burden on patients alone without their explicit consent? Neither utopia nor dystopia, sustainability asks data rights to be valued so as to be better exploited.

Unsustainable situations are not unknown to agriculture. Fortunati datacolae, beware! Dataculture is just as prone to such slow motion catastrophes.

Philippe Coueignoux

• (*) ......... Supply and demand in the sky, by Gerrit Wiesmann (Financial Times) - November 4, 2009
• (**) ....... The Medical Industry Grumbles, but It Stands to Gain, by Duff Wilson and Reed Abelson (New York Times) - November 9, 2009
• (***) ..... IMS Health in $5.2bn leveraged buy-out, by Justin Baer (Financial Times) - November 7, 2009
• (****) ... Can creative industries survive digital onslaught?, by Ian Brown (Financial Times) - November 3, 2009
• (*****) . Prescription: Networking, by David Talbot (Technology Review) - November/December, 2009
• (1) see the Prisoner Dilemna in the wikipedia