Close ☰
Menu ☰

Google’s slow-mo train crash

Posted on: Tuesday 10th of April 2012

A fascinating letter to investors by Google CEO Larry Page. Here is a key passage:

“Understanding identity and relationships can also help us improve search. Today, most search results are generic, so two strangers sitting next to each other in a café will get very similar answers. Yet everyone’s life experiences are unique. We are all knowledgeable about different things; we have different interests and our preferences—for music, food, vacations, sports, movies, TV shows, and especially people—vary enormously.”

“Imagine how much better search would be if we added… you. Say you’ve been studying computer science for awhile like me, then the information you need won’t be that helpful to a relative novice and vice versa. If you’re searching for a particular person, you want the results for that person—not everyone else with the same name. These are hard problems to solve without knowing your identity, your interests, or the people you care about.”

For me, two things fall out of this. First, most value is defined by a specification, something that captures the individual’s current needs and preferences, different interests, levels of skill and knowledge and so on.

Second, the person who knows most about this specification is the person whose need/spec is being met. So, therefore, the best way to value is to elicit the spec from the person concerned – what we call VPI or volunteered personal information.

That’s how Google became great – by eliciting specifications from individuals about what they were interested in, when: the search term. Intriguingly however, Google’s original process only elicits two of the three ingredients of real relevance: it captures what I want, when but it misses out the ‘who I am’. And as Page points out, creating real relevance is a ‘hard problem’ without access to ‘who I am’ data.

Now. You would have thought that the obvious thing for Google to do was to emulate the model that made it great: empower people to input this information, in ways they can control, when they want and need it – by building on its original search model; a model driven by VPI.

But no. Instead, Google has adopted a behavioural targeting inspired stalking model. Under this model, instead of being completely open and up-front, it tries to get you to input information for one purpose (i.e. to write an email, watch a video, or manage your social network) and then uses this information to draw inferences about you to serve up ads it thinks might be relevant to you. This is the exact opposite of the approach that made it great.

I suspect Larry Page knows this. In an excruciatingly embarrassing passage with the heading ‘Love and trust’ he tells investors “We have always wanted Google to be a company that is deserving of great love” and that “We have always believed that it’s possible to make money without being evil.”

He then starts wriggling. Today, he says, “most of our revenue comes from advertising … and we work hard to make these advertisements relevant for users. Better ads are better for everyone—better information or offers for users, growth for businesses, and increased revenue for publishers to fund better content.”

He continues: “The recent changes we made to our privacy policies generated a lot of interest. But they will enable us to create a much better, more intuitive experience across Google.”

In fact, Page is wrong. Most of Google’s revenues don’t come from advertising. They come from VPI – the information individuals volunteer in their searches, which then drives the advertising. By forgetting the real source of his revenues – by abandoning his original model of user-controlled VPI for customer stalking – and by using ‘relevance’ as an excuse for an invasion of privacy, Page forgets what made Google great and what originally earned it ‘love and trust’.

For evidence of rapid loss of trust, just take a look at the comments on the BBC news item reporting his letter. Last time I looked, the first comment said “Maybe if I could choose whether to share or not, I might consider this a good idea.” This pretty much nails the whole thing in one go.

With loss of trust comes a loss of revenues. Not immediately, but in due course. Google is now a slow-mo train crash unfolding.

Alan Mitchell