Comment by dasht

I think that Mr. Schmidt is trying to do two things: (1) Preparing for Google to eventually be "outed" about just how much data it actually hands over to the government on the basis of the Patriot Act; (2) Distracting the general public from the broader range of privacy concerns that Google raises (specifically: private surveillance as contrasted with government surveillance).

He mentions the Patriot Act and I would paraphrase him as throwing up his arms and saying "Hey, we're subject to those laws just like you! We have no choice!" Alas, Google does have the choice and unique opportunity to challenge those laws but it appears that they simply decline. That is why I think he's preparing for Google to be outed vis a vis the extent of its cooperation with government.

Next, asked about privacy concerns, he immediately reframes the question as if it were a question about government surveillance - neatly ducking any issues about private surveillance. Bartalomo fails to follow up.

The issue about private surveillance is significant in mundane (but serious) ways and in more speculative, cultural crisis kind of ways.

A mundane concern is the question of whether or not Google "leaks like a sieve". For example, if someone at Google dislikes me, how hard is it for them to scratch around for dirt in my (nominally) private information? If someone at Google is a friend or (black market) business partner of a recruiter at some other firm to which I apply for employment, is an unauthorized background check possible? Do such things happen? It's hard to imagine that they are technically hard to prevent....

A speculative but serious cultural crisis concern regards Google's deliberate and accidental implementation of what we could dub behavioral tracking and targeted manipulation. For example, let's suppose that one day I get a crazy suspicion from something I read on a blog. I think that there could possibly be a vast conspiracy and that my best chances for survival might be to liquify all of my assets and invest everything I have in personal caches of peppermint candy. I don't fully believe it but I start researching via Google. Google's AIs quickly figure out that I linger on ad-carrying content from backers of this conspiracy theory and, pretty soon, that stuff is at the top of all my searches and predicts 40% of the ads I see. From my perspective: that stuff is "all over the Internet" and I begin to wonder if my neighbors aren't in on the conspiracy since none of them talk about it. Pretty soon, Google has built me a personal, private channel of conspiracy news and peppermint candy ads. I sell my house, leave my family, cash out my retirement plan, by a van to live down by the river, and fill up a rent-a-store spot with boxes of candy.

Now, that same personal bug - my vulnerability to a conspiracy theory - may have been there all along. The change here is that back in, say, 1975 - the biggest mistake I might make is to subscribe to a few newsletters and magazines, perhaps hedge by buying a dozen boxes of candy rather than a gross. Meanwhile, the TV, the radio, all these other things are feeding me counter-evidence to the conspiracy and I'd eventually have a much better chance of coming to my senses.

But Google is unprecedented in its comprehensiveness as a source of monitoring and manipulating my attention. Like the Vegas slot machine that convinces me (wrongly) that I have a system to beat the slots because it pays out 49% of the time, so long as I pay it with my recognizable frequent-gambler credit card -- Google's AI can potentially really mess a person up.

That is a privacy concern. That's a privacy concern directly raised by Google's stated intentions of dabbling in behavioral tracking. And it's a concern that Schmidt deflects from attention by turning the question into a question about government surveillance.

Now, things can get even worse. Suppose that Google's AI doesn't cause large numbers of people to drop out of society and load up on peppermint candy but that, shucks, little discrete tweaks here and there can use the same behavioral science effects to, say, influence the next election for the local dog catcher. And one of the candidates for dog catcher has a slush fund, and a friend at Google.... Or, forget the friend at Google... suppose the dog catcher candidate has a friend who, this year at least, has found some underhanded tricks that currently work really well for SEO....

The ethics of running search and ad placement are unprecedented and difficult and hinge very much on how we update our understanding of "privacy". They hinge very much on the question of privacy vs. behavioral tracking. Schmidt appears uninterested in talking about these issues and very much interested in deflecting attention away from them.

At most only slightly exaggerating: we're all analogous to lab rats, now, in Google's massive-scale human-subject psychological research project with unprecedented amounts of automated surveillance and stimulus production. With money and power hinging on the outcome.