Sure, if you accept that there are engineers that are 5, 10, 20 times as valuable as their salary requirements, then that makes sense. You might as well scoop up those under-priced assets, even knowing that you can't necessarily measure them perfectly.

Since engineers love the product more than the marketplace, you keep them happy by paying them well and letting products flourish on their own technical merits. You do that and you don't have to worry too much about driven engineers leaving for startups.

And you have to provide them with a stimulating and rewarding environment, like, some kind of... super-college.

That's all pretty much conventional wisdom, but they're executing well on it. It worked for Bell Labs and Xerox in a simpler time. Of course, their underlying companies had too limited a mandate to take the technologies their "radical researchers"[1] created and bring them into the marketplace. But as we know, Google's mission statement is a less succinct but more exacting restatement of "do computer science," so they're good there.

I don't know, it still seems fundamentally sound to me. Maybe Google is bound for the "eternal September" effect, or maybe they're getting better at teasing out the engineers out there who are a better fit for the company. Mere hiring numbers don't really tell the whole story, I think.

[1] - http://equityprivate.typepad.com/ep/2006/04/what_spaceward_....