February 2, 2010

Forever fast and furious, their flow fascinates us. But, like a conjurer's hands, the news hide more than they reveal. If last week I implicitly boasted to being able to see through such smoke and mirrors, I need today to be more explicit about the nature of my claims.

I try to contribute a theory of information with which to understand our Information Age. Such a claim to scientific truth is modest by nature as I open myself to being proven wrong. Take for instance the diabolo effect which purports to explain both Microsoft's past success and Google's present strategy. As far as Microsoft goes, I may not have ventured much but the future will tell how accurately I have foretold Google's fate.

Unlike dragons, diabolos do not live for ever. Recall I stated Oracle has also benefited from such an effect, although its relational database diabolo has never been as total as Microsoft's operating system. These days are over. Read Ashley Vance on its acquisition of Sun Microsystems. "[It] wants to be able to claim to prospective customers that it [...] has more of the parts to be an end-to-end service provider" (*).

The vertically integrated IT supplier's business model is perfectly valid. But by its adoption, Oracle acknowledges it is no longer in a position to compel hardware architects and application developers to integrate their products with its platform at their own costs and to its own benefit. Of course perhaps Oracle paid $7.4 billions just for getting its hands on Java in the hope of building a new diabolo for the cloud. It would be a stiff price, even when accounting for MySql, and Neelie Kroes does not think so (1). But then she also let Google swallow DoubleClick, didn't she?

My view of information theory draws as much from sociology as it does from economics and engineering. The closer it gets to a philosophy, the more value comes from the questions it asks rather than the solutions it offers. So allow me to question the news and draw your own conclusions.

Stephanie Clifford makes clear how much effort went into picking a little 'i' icon to flag "online ads that use demographics and behavioral data to tell consumers what is happening" (**). This is a welcome development but, may I ask, will the content match the packaging for quality? Will consumers click the 'i' and find what personal profile data has truly been used for targeting them or just another bland, generic so-called "privacy policy"?

Beyond Google's belated stand against Chinese censorship, turn now to Ian Austen's account (***) of how "a respected magazine about Canadian history" decided to change its name because "Web filters at schools and junk mail filters in e-mail programs [...] block access to material containing the magazine's name". To avoid this fate, I will only say MIT chose the animal concerned as its mascot "because of its remarkable engineering and mechanical skill and its habits of industry" (2). Here then are some questions for Jason Pontin, editor in chief and publisher of Technology Review.

Is the Chinese government behaving differently than our schools? Isn't such censorship similar in its effects to a ranking recommendation system like Google's, whose highly guarded proprietary algorithm, out of a million items, picks the few you will ever see? What matters most, whether a recommender erases or selects or whether its influence over its users is monopolistic? If the latter, what insures our Information Age can sustain a diversity of recommenders? How to enable each family the freedom to follow its own preferred recommenders (3)?

It is not only because we share an incidental interest in some unmentionable rodent that I call on Jason Pontin. As a successful, yet non dominant, recommender of technical news, he knows the subject better than most and, having once advised his colleagues to live within their means, he is bound to be sensitive to the influence of scale. Isn't scale also the most crucial and mistaken factor behind the ongoing fight for copyright protection?

Courtesy of Samuel Laurent (****), we learn the French HADOPI approach (4) is being debated among the 39 countries discussing a potential "Anti-Counterfeiting Trade Agreement" (5). What would Jaron Lanier think? According to John Tierney's discussion of his latest book (*****), he should applaud as he is now hostile to copyright piracy, yet, warning us about software lock-in, he may well wonder the wisdom of entrenching what amounts to another layer of unescapable censorship to be administered by Internet access providers for the sake of corporate media interests.

Rather "Mr Lanier [...] proposes introducing innovations like a universal system of micropayments". I agree. But can copyrights protection studiously avoid looking at how disfunctional those rights have become? Why data rights should be limited to professional authors and corporate entities? Why should extravagant data rights encourage a system whose dominant stars impose their own "digital Maoism" by undermining cultural diversity?

Read Jonathan Mahler on "how a genre writer has transformed book publishing" (******). Although crafty as a businessman, James Patterson comes out refreshingly unpretentious as a craftman. The villains of the story are the publishers, those lazy recommenders who "became less and less willing to gamble on undiscovered talent and more inclined to hoard their resources for their most bankable authors. The effect was self-fulfilling".

My previous questions focused on the relationship between scale, censorship and recommendations. Underneath, Internet is but a universal communication system. Unfortunately it has become the battle field of our Age. James Blitz and Joseph Menn report how "the US and its Nato allies were urged [...] to fend off the threat of cyberattacks in the aftermath of the alleged Chinese assault on Google" (*******).

As a global agora fed by digital highways without an international police, Internet is a natural host for assorted prostitutes, highwaymen, vigilantes, soldiers and spies. Travel there being both easy and difficult to trace, no wonder any property open to it can be taken by sudden storm. But isn't this misleading? Can one truly loiter on the Internet? Rather one never gets on it without declaring precisely where one is going.

Of course one's destination may not be final. One may first want to pay a visit to Google to find out exactly where to go. And one may use a name (the URL) rather than an address (the IP), knowing the Internet will do the translation. Nevertheless one has always a single target at a time. Wouldn't it be more efficient then to ask the avowed destination whether it agrees to let the user access the network in the first place?

This of course flies in the face of current practice. Plus, peer to peer negotiations would consume significant resources and remote access control methods can be defeated as well as any other defensive measures. Also confidence games will always allow a subtle attacker to get at her target.

If I am wrong, the mistake will be entirely mine. But given the scale of the stakes, can one brush off such questions without giving solid reasons?

Philippe Coueignoux

• (*) ............. With Sun, Oracle Says It Has All the Pieces, by Ashley Vance (New York Times) - January 27, 2010
• (**) ............ A Little 'i' to Teach About Online Privacy, by Stephanie Clifford (New York Times) - January 27, 2010
• (***) ......... Web Filters Cause Name Change for a Magazine, by Ian Austen (New York Times) - January 25, 2010
• (****) ....... Acta, le traité «secret» qui inquiète le web, by Samuel Laurent (Le Figaro) - January 26, 2010
• (*****) ..... The Madness of Crowds and an Internet Delusion, by John Tierney (New York Times) - January 12, 2010
• (******) ... James Patterson Inc., by Jonathan Mahler (the New York Times Magazine) - January 24, 2010
• (*******) . US urges shared cybercrime defence, by James Blitz and Joseph Menn (Financial Times) - January 27, 2010
• (1) see the EU Competition Commissioner's press release
• (2) see the MIT mascot history
• (3) if internet access is too expensive for poor families, a local school has more control over them than the Chinese goverment will ever have over its citizens.
• (4) for the texts of the HADOPI laws, see the references for copying, in Liabilities and Vulnerabilities in the Information Age
• (5) see the entry on ACTA in the Wikepedia