It's Alive

A fair amount of buzz in the blogosphere today about yesterday's Yahoo! announcement that the portal's ad revenues were expected to drop because of a "slowing US economy." This is a charming idea, but the likely reason is that Yahoo!'s ad service sucks.

Search, I'm coming to realize, is a classic example of an affordance error. Humans seem predisposed at some level to make these. In essence, an affordance error has to do with misjudging the purpose of some device based on how it looks like it's supposed to work. The classic example is pushing a door when you're supposed to pull it. (Remember that Far Side cartoon--"The School for the Gifted, " I think? The Far Side Fat Kid is pushing with all his might at a door marked PULL. Sorry, Gary Larson.) But there's a pair of subtler, more insidious affordance errors. The first is assuming that something that looks simple actually is simple. The second type, potentially very dangerous, occurs when you assume that you can easily control and even predict the behavior of a simple (looking) device or situation. --Some examples? Water of unknown depth. Grenades. Search engines.

It's simple, right? You type in the word, you get a relevant site. (How quickly we've forgotten premillennial search experiences...but I digress.) How hard can that be? And wouldn't it be a good idea to start one of those yourself? Might be some money in it.

Well, the problem is that Google's simple interface hides an engine of fiendish power and complexity--in short, the Google engine is an artificial neural network that uses the sites it indexes to determine how to index and emphasize its own sites. Yahoo! is the equivalent of a really big phone directory. It's a good directory, but it doesn't learn, and it certainly doesn't learn faster the longer you run it. In fact, you could make a good argument that Google's engine is actually intelligent...probably not conscious, but a non-human intelligence nonetheless.

Of course, to anyone who's not reading about cognitive mapping and modeling, this is so much irrelevance. If you aren't a pilot and you compare a ramjet to a rotary airplane engine, you might well assume that the ramjet is inferior...it just looks too simple to work as well as the rotary. But that clean interface hides a product with a market advantage so superior that it's hard to describe, let alone explain. If you don't get this, you have no business starting a search engine company. If you do and you still start one, you're either stupid, deluded, or actively dishonest.

So this clarifies what's going on with Yahoo!. They're not stupid. They're hearing complaints from their customers and have been for a while. They know there's a problem. The answer must be that they just can't do anything about it. Even if they managed to build a neural net--and Google has done its best to hire everyone who knows anything about them--it would never catch up to a neural net that's been learning and studying for years now. So what is to be done?

1) Distract people with Panama. I wonder if this much-delayed product exists. Meanwhile, move quietly and calmly toward the lifeboats.
2) Go social. This plays to Yahoo! strengths--they've long been known as a pretty social-focused culture for a Net company, and they're trying to acquire social sites.
3) Admit you're beat when it comes to search. Which was probably a mistake--poor "message control"--but wasn't news to anyone geeking out on the topic.

The problem is that making the switch is going to take a long time and be expensive for Yahoo!. Even figuring out a new revenue model (and a culture to go with it) is going to take a long time and be expensive. Meanwhile, Google's revenue model is an extension of its neural net (see bottom of the first page in particular).

The result is clear. So's the bet.