As I listened to Udi Manber, the head of core search team at Google, I was impressed by something that most of us understand in a different sense: engineering matters at Google. Most of us think about this in terms of the other things we know about Google; like the one day a week people get to work on their own project, or the fact that they build their own custom servers.
Manber talked about making search queries meaningful--understanding intent rather than just doing text matching. He outlines a number of upgrades to Google search that I've noticed over the years but never thought much about. These all require a commitment to engineering and constant product development.
A simple example of this that I thought of while he was talking is internationalization, or i18n. Google has deployed search engines in myriad languages and countries. Anyone who's done i18n knows it's not glamorous, it's not easy, and there's little you can do in the way of innovation to leverage technology. It just comes down to a commitment to good, solid engineering and working it out.
Of course Google isn't alone in doing i18n. Lots of companies do it. But plenty of companies put it off because they can't afford or don't want to afford the commitment that it takes. In general, no one's giving Google press over the increased utility of their search. Lots of stuff gets the media attention before that. But it's core to Google's business and it's clear it gets the engineering attention it deserves.