November 25, 2008

We live in the Information Age and, with regard to information, nothing can be more important than Truth. Yet one should always be ready to suspect information to be a lie. To overcome scepticism, we have stressed the use of recommendations and analysed the three sources of truth: science, authority and popularity. Does it mean there was nothing more to say on the subject?

Simson L. Garfinkel has given the lie to such complacency in a insightful study on "Wikipedia and the Meaning of Truth" (*). Wikipedia is not known to be a font of absolute truth even though legal opinions may quote it for greater readability. Garfinkel's original contribution is to show the internal consistency of the Wikipedia approach to truth.

Wikipedia aims at "received truth, the consensus view of a subject", what I call truth by popularity. So doing, it is in no position to appeal to expert authority. To stay true to its spirit, it must extend the same consideration to every contributor alike. It does not mean Wikipedia holds its material to be true, rather "objective truth isn't all that important".

The result could be pandemonium. Indeed some edits turn into shouting matches which its administrators interrupt by freezing the pages concerned. A misnomer. Administrators actually restrict access based on the privilege level rewarding contributors' past achievements over time. If it reminds you of Dungeons & Dragons (1), you are not far from the truth. Wikipedia is a truly successful massively multiplayer online role-playing game.

Administrators and, for less controversial subjects, editors apply the "verifiability" principle, one of Wikipedia "core content policies". As Garkinkel says, "verifiability is really an appeal to authority - not the authority of truth, but the authority of other publications." In other words, Wikipedia aims at popular truth filtered through the electoral college of published authors. If these electors disagree, Wikipedia will stay neutral but reserves the right to weigh electors according to the reputation of their publisher. Direct democracy it is not. Neither will it condone the tyranny of the majority (2).

The result does not have to be true. Indeed, pending a reliable source, it always lags factual, historical truth unless an original contributor attempts to flout editorial censorship. Wikipedia is a land of fleeting scoops and nothing guarantees truth's ultimate emergence. But we all use it, don't we?

Reacting to Garfinkel's analysis, Matt Mahoney recalls how H.B. Phillips reconciliates so called "objective truth" with truth by popularity (**). "When nearly all agree who claim to know, it is reasonable to assume the majority view is correct.[...] The probability of error [...] does not justify hesitation". But as Matt Mahoney himself realizes, this is tautological nonsense. Popular truth is nothing more, nothing less than what it is. In Galileo's time, half a century before Newton, heliocentrism was not yet scientifically proven and went against the popular truth. It did not make, on this specific topic, Galileo's authority any less right and the Church's any less wrong.

More eye-opening is Garfinkel's sympathetic telling of Jaron Lanier's travails. Although his Wikipedia entry (3) has since been corrected and Garfinkel's article duly included as a reference, Lanier could not for a time correct his being erroneously described as "a film director". Someone insists you are what you know you are not, but you cannot originate this fact yourself. One can sense a true tension here but is it "digital Maoism"?

Rather Lanier's frustration reveals a truth about truth. Absolute truth, objective truth, truth in short, cannot exist for society short of communication. Truth by popularity is a mirror held by others, whether we like our reflection in it or not. Truth by authority implies the faith of the believer. Even scientific truth requires the reader to be able to follow the proof or to replicate the experiment, pending which, truth stays in limbo. Until he found a proper channel through friendly receivers with enough clout, society held Lanier's factual self-knowledge to be not emitted, non existent, not true.

If for society truth needs communication, it follows truth is noisy. What to make then of Jackie Calmes' account of the Obama administration vetting process (***), which asks job seekers to recall "if [they] have ever sent an electronic communication [that could embarrass the President-Elect]"? First it takes as an authority someone who has no longer access to all the records concerned. A witness memoir does not History make. Second it ignores the ease with which such electronic records can be "enhanced" by malicious hands. In truth it throws a challenge to hackers of every nation.

Two years ago a rumor falsely implicated Senator Clinton in a smear attempt on then Senator Obama. Were she to become US Secretary of State officially, how difficult could it be for hostile foreign interests to plant an embarrassing, highly believable bait in an apparently well managed electronic archive and hook on the media for all its worth at the right time ? How could Hillary Clinton prove she did not send the incriminating message?

Instead of trying to wring truth out of every bit of information we send, the US President-Elect should reflect on our rights to eprivacy and on the need to nullify the legal standing of electronic records produced in violation of them. Truth to tell, I think it would prove to be a popular decision.

Philippe Coueignoux

• (*) ..... Wikipedia and the Meaning of Truth, by Simson L. Garfinkel (Technology Review) - November/December, 2008
• (**) ... The Privilege of Being Wrong, by Matt Mahoney (Technology Review) - November/December, 2008
• (***) . For a Washington Job, Be Prepared to Tell All, by Jackie Calmes (New-York Times) - November 13, 2008
• (1) see Dungeons & Dragons Online in the Wikipedia
• (2) see the tyranny of the majority in the Wikipedia
• (2) see Jaron Lanier in the Wikipedia