This image was lost some time after publication.

Google's top executives desperately want to convince Wall Street that it's on the verge of cracking the $70 billion television-advertising business — automating it, rationalizing it, and ruling it, as it has done with the considerably smaller search-advertising market. They've even hired an NBC executive, Michael Steib, to sell broadcasters on the idea. The only problem: It will never work, as Google's own documentation shows. Google's triumph in search is a product of its skillful use of data. By analyzing what Web searchers click on and what advertisers say they'll pay, it's able to continuously refine the ads it displays to yield the most clicks for advertisers and the most profits for itself.

No such virtuous circle exists in television. As a substitute for clicks, Google has proposed tracking when TV viewers change channels during ads. But as the above example provided by Google suggests, no matter what the ad is or when it airs, 96 to 97 percent of the audience stays tuned in. It's not clear if those numbers are made up, but actual numbers may not prove that different. Studies from TiVo, supposedly the bane of all TV advertising, showed 40 percent of users don't bother to fast-forward through the ads. Outside of efficiency-obsessed Silicon Valley, commercial breaks just aren't that horrifying a notion.

Google's hoping to extract information from its television data, deriving some insights from channel-switching habits. But the information — do viewers care about the ads or not? — likely isn't there. Absent that, Google's systems have no way to refine themselves over time. All Google can promise, then, are cheaper rates for undesirable time slots — or, possibly, implementing technology that forces viewers to watch ads. No wonder TV executives are turning up their noses. Like the typical engineer, Google thinks it can do their job for them.