August 28, 2007

Food, energy, data. My last fillip focused on the downside of overabundance in each of these fundamental goods. Lest we draw the parallel too far, we better remember a fundamental difference separates information from food and energy. The latter two cannot be used without being consumed, which raises the spectrum of future famine in the midst of actual abundance. No such limit applies to information. The catch, for there is always a catch, is that this very fact makes it almost impossible to reliably dispose of data (1). Remember that data can be bad as well as good and draw the conclusion yourself: truth is doomed to drown in a flood of lies.

Why then be surprised to spot sudden surges on Our Sea of information? Observers have lately witnessed the discovery of a shocking fact, i.e. Wikipedia is not a font of absolute truth. More than a year ago, Yochai Benkler has clearly explained to us in the Wealth of Networks that truth on Wikipedia comes, not from the unerring knowledge of its contributors, but from the collective capacity of all its readers to issue corrections. What is not guaranteed is the speed at which truth will emerge. In the mean time, Wikipedia welcomes bad and good contributors alike. Wikipedia can help to spread defamatory comments, as Noam Cohen reminded us in a recent report (*), or insert favorable spin, as analyzed by Katie Hafner (**).

I see no need to make more waves. But, faithful reader, forgive me if I add my grain of salt. The Wikipedia controversy is a lesson in democracy.

The participation of the people may be at the root of democracy but this notion is easily misunderstood. As regard to elections, secret ballots make vote rigging more problematic. As regard to justice, the right to be judged by a jury of one's peers protects individuals against tyranny. One should be careful though to think through why these mechanisms do work.

Secret ballots work well when genuine universal suffrage delivers an exhaustive snapshot of the body politic. Faked democracies pay homage to this model and no longer use rotten boroughs, poll taxes and the like. Rather they preselect the candidates, pronaocracies through campaign financing, theocraties and communist aristocracies through co-optation. Jury trials work well when genuine random sampling delivers an unbiased representation of the community. Democratic parodies carefully preselect the jury members to achieve the desired level of sample bias.

The issue with Wikipedia is to confuse the means for the substance. Its editorial anonymity mimicks secret balloting but fails to deliver exhaustivity. Most of us do not contribute. Worse, its very self-correcting principle is incompatible with a snapshot picture. More motivated members spend as much time as possible propagating their opinions, a behavior akin to ballot box stuffing. On the other hand, Wikipedia contributors' self appointment ensures maximum sample bias. Wikipedia aims to achieve truth by democratic popularity. It delivers an abundance of potentially biased data.

Wikipedia can be improved though and the current controversy shows the way. In a recent unconscious masterpiece of irony, a Boston Globe editorial (***) lists an intriguing sample of organizations caught harboring anonymous contributors, from the National Rifle Association to the Central Intelligence Agency. Strangely enough Renée Loth does not include the corporate owner, the New-York Times, duly mentioned in Katie Hafner's story. Could it reflect the editorial page editor's favorite paranoias? No harm done. Both authors assume their identities and therefore their responsibilities. Let the reader decide.

Why don't Wikipedia contributors and editors accept to shoulder responsibility and disclose their identities? In switching to a model of truth based on authority, Wikipedia could still differ from its commercial competitors by remaining open to all contributors.What it takes is to build a meta recommendation mechanism, not to recommend facts but fact purveyors, with an appropriate redress process. Easy? Not by far. Democratic? Absolutely.

Philippe Coueignoux

• (*) ..... Defending Wikipedia's Impolite Side, by Noam Cohen (New-York Times) - August 20, 2007
• (**) ... Lifting Corporate Fingerprints From the Editing of Wikipedia, by Katie Hafner (New-York Times) - August 19, 2007
• (***) . Behind the e-curtain, editorial (Boston Globe) - August 26, 2007
• (1) see Disposal of Digital Information in my lectures on Liabilities and Vulnerabilies in the Information Age.