Behind the curtain

Adverlab today discusses New Scientist's compilation of predictions for the future, quoting Google's Peter Norvig:

"In 50 years the [search] scene will be transformed. Instead of typing a few words into a search engine, people will discuss their needs with a digital intermediary, which will offer suggestions and refinements."

Adverlab goes on: "some advertising will be targeted not at humans but at their robotic assistants powered by artificial intelligence to make the most optimal purchase decisions" - echoes here of Vin Crosbie's penultimate Clickz article, the second part of his Long View manifesto.

If this is where the finest minds at Google thinks this is where the company should be heading, then I would tentatively opine that we have finally found the thing that will bring down the unstoppable behemoth of the first digital revolution.

For my money Chris Anderson's biggest contribution to the understanding of the digital age is not his Long Tail thesis (which is really just a way of articulating what Jeff Bezos and others had already expressed as corporate strategy) but his thinking on human responses to probabilistic systems. (I'm not going to try to summarise his insights, you should go and read his article before continuing, and if you have time read Nick Carr's eloquent but irrelevant rebuttal too). Why is Google so successful - no, why is Google's success as a search engine clearly and counter-intuitively unrelated to the quality of its search results? Because by an accident of positioning it has a taken a probabilistic algorithm, that all things being equal the human mind should recoil from trusting, and anthropomorphised it to make us think of it as a source of definitive authority - "Just Google It". Google is all about the spin, and Chris Anderson saw it first.

What, therefore, could be wrong with Norvig's "digital intermediary", with Adverlab's advertising aimed "not at humans but at their robotic assistants"?

It's a suspension of disbelief too far.

While not necessarily easy to game Google it's certainly possible - the renowned technique of the Googlebomb (Wired) works. These are the sort of quirks that currently we can rationalise away as eccentricities, as further proof that Google is not merely ultimately authoritative but engagingly human in its fallibility. When we're just searching for pages on the web, it's no big deal.

But a digital intermediary, that makes or validates purchasing decisions on a probabilistic basis and can be gamed by targeted advertising? At that point the curtain falls away and, rather crucially, there is no-one behind it at all. If that's the direction that Google thinks it is going, it doesn't understand the absurd psychological serendipity that is the foundation of its success.