Language mirrors society. As rocks witness the earth's past, written and oral records capture the flow of history. When I was young, indisputable knowledge was called gospel truth. Over the years "as seen on television" has become a litmus test for veracity. At present I see the emergence of "it must be true, I saw it on the Internet".

If Internet is to become the source of all truth, it befits us to understand what this means. Indeed the first time Harry Eyres mentions the Internet in "Slow Lane", his weekly column as a life taster (*), it quickly turns to thoughts about truth.

A few months ago (8/22 fillip), I outlined both the issue, all Internet productions are potential lies, and the solution, good recommendation mechanisms. Pragmatic as it is, this approach skipped around the harder task of defining the nature of truth. Skeptics of course have it easy, but cowardice taints those too quick to wash their hands of sticky tasks.

Truth comes to us from three sources: science, authority and popularity. I believe a statement because I can prove it, because I have faith in the person from whom I received it or because it is the general opinion of the people. Such a simple description belies a complex reality. Outside of mathematics, science refutes rather than proves (1). Time constraints lead one to take most science on faith, leaving it to expert authority. And while claims of scientific or religious authority are not so popular today, popularity itself lends moral authority to well meaning entertainers (2). Infighting between sources of truth does not invalidate my model though. When feisty triplets trip one another, it helps parents to know who's who.

At its core, Internet is a new means of communication. As such it gives science a tremendous boost by making the publication of proofs as easy and universal as one can wish (3). Pierre de Fermat would not be able today to argue about the lack of marginal space to cheat us from the "marvellous proof" of his last theorem (4). For the same reason, Internet is also of immense benefit to sources of authority to whom it provides a quasi free forum to disseminate their messages or recommendations. If I dare to attack the power of advertisers, lawyers and governments (5/30 fillip), it is because the Internet gives me a convenient soapbox to claim authority on matters of eprivacy.

The genius of Internet however is to provide an ideal platform for truth by popularity. Not the top-down sort achieved by time-tested fame or money-backed marketing, but the bottom-up kind where users vote with their clicks to elevate the visibility of what they like. In previous reviews of recommendation mechanisms, we have often found a conscious attempt to harness popularity. Such is the case at Google (9/12 fillip) and Netflix (10/10 fillip). In other mechanisms, the link between popular votes and recommendations is even more direct.

At Digg (5) for example, registered users submit stories they like to the vote of the membership. Outsiders can search the site as one would use Google. Beyond the obvious differences, that Google actively seeks out all Internet content and retains control of its ranking algorithm, the real key behind Digg is its registration process. It may restrict voting to the insiders but it also enables Diggs to better protect process integrity. According to our criteria (8/22 fillip), Digg's recommendation mechanism is totally decentralized within the membership, and each elementary recommendation can be traced back to its author. It is true Digg users may hide behind a pseudonym, thereby limiting their responsiblity, but Digg provides an abuse hotline, the first explicit protest process we have found so far.

The problem is, a clean recommendation mechanism based on popularity is not necessarily good when information is so easily distorted and propagated both. Crowds are prompt to embrace the most outlandish rumor as fact when it reinforces their fears and prejudices. Which coach class frequent flyer would question the standing room only Airbus "seating" arrangement for Asian airlines? While the story graced the first page of the New-York Times business section last April (6) and led to a quick and contrite correction, I for one would have voted for it on Digg. Last week (11/28 fillip), I suggested Beaumarchais' Bazile would have found Internet ideal for calumny. Optimists counter that knowledgeable people denounce falsehood on the spot as fast as it is sown. I persist. On Google today the Airbus rumor still ranks higher than independent corrections (7). To push the latter off the first page of results would only require a bit of Google bombing (11/07 fillip). And what if the rumor claims the victim before the correction. Ian Austen gives us a report as chilling as it is concise on how to YouTube teachers out of work (**), simply provoke, record and post. In capable and malicious hands the Internet is a Weapon of Massive Disinformation.

This indictment of popularity as a source of truth should not be confused for a disdain of democratic rule. True, western voters regularly elect the candidate who lies the most convincingly. Notice however the alacrity with which the same electorate boots the party in power out of office. Through periodic balloting, the correction mechanism optimists see in popular wisdom does work, even if it makes for a jerky ride. If the Internet makes it easy to cast votes, it does not help to conduct democratic ballots. Without the accountability this would create, truth and falsehood accumulate votes forever and nothing guarantees truth always comes up on top.

To counter popular rumors and calumnies, I see only one solution, the ability to trace them back to their origins. Force the author to take responsibility under his or her own identity. This stands counter to most modern modes of thinking (8). But what if falsehood were radioactive? Experts the world over would scour the Internet as British sleuths today frantically track polonium traces from the Atlantic to the Ural (9). While impossible on Internet as the ultimate agora, it can be done on Internet as the ultimate pneumatic network where complete respect for privacy is balanced by a chain of responsibility.

No solution is ideal. Without democratic Justice, a system of individual responsibility is as dangerous than anonymous anarchy. Governments are prone to take a concept like "the death of the author" a little too literally. They would love nothing more than a good chain of responsibility to track our every thought (10)(11) under the guise of total surveillance (6/06, 7/04, 8/29 fillips) and universal censorship (9/05, 10/31 fillips).

If to use Harry Eyres'words, you want the Internet to be more than "a great big adult playground", I challenge you then to relay my calls for a revolution in the name of Privacy, Identity, Responsibility.

Philippe Coueignoux

• (*) One man's truth is another's world wide web of lies, by Harry Eyres (Financial Times) - November 25, 2006
• (**) Telling Tales Out of School, on YouTube, by Ian Austen (New York Times) - November 27, 2006
• (1) according to Karl Popper's definition of falsification
• (2) see for example Bono's activities
• (3) see for example the proof of Poincaré's conjecture
• (4) contrast Pierre de Fermat's little theorem with his last theorem: some proofs are easier to write and to follow.
• (5) check here to dig deeper on Digg
• (6) One Day, That Economy Ticket May Buy You a Place to Stand, by Christopher Elliott (New York Times) - April, 2006
  Its Internet version today includes the correction. This only shows how Internet helps a source of authority to restore its reputation.
• (7) check if this is still the case by searching Google for standing only passenger airbus
• (8) see how in The Death of the Author, Roland Barthes attempts to free texts from their authors.
• (9) Radioactive Trail Is Found In Case of Poisoned Ex-Spy, by Sarah Lyall and Steven Lee Myers (New York Times) - December, 2006
• (10) Justice Official Opens Spying Inquiry, by Eric Lichtblau (New York Times) - November, 2006
• (11) Subpoenas And The Press, by David Carr (New York Times) - November, 2006