Google used to have a severe problem where code refactoring & maintenance was not rewarded in performance reviews while launches were highly regarded, which led to the effect of everybody trying to launch things as fast as possible and nobody cleaning up the messes left behind. Eventually launches started getting slowed down, Larry started asking "Why can't we have nice things?", and everybody responded "Because you've been paying us to rack up technical debt." As a result, teams were formed with the express purpose of code health & maintenance, those teams that were already working on those goals got more visibility, and refactoring contributions started counting for something in perf. Moreover, many ex-Googlers who were fed up with the situation went to Facebook and, I've heard, instituted a culture there where grungy engineering maintenance is valued by your peers.
None of this would've happened if people had just heroically fallen on their own sword and burnt out doing work nobody cared about. Sometimes it takes highly visible consequences before people with decision-making power realize there's a problem and start correcting it. If those consequences never happen, they'll keep believing it's not a problem and won't pay much attention to it.
That's a really, really good line, and also a good guiding principle.
None of this would've happened if people had just heroically fallen on their own sword and burnt out doing work nobody cared about.
Slight disagree here: it was entirely possible that many people could have sacrificed their health, happiness, and quite possibly lives to the altar of Google's code quality... and changed absolutely nothing about Google's trajectory because Google would neither have noticed their efforts nor, had it noticed them, cared. That's not one of my periodic elbow swipes at Big Daddy G, by the way, that's just Life At A Megacorp. It was substantially the same way at my old day job -- I managed to convince myself that with just a few more hours and a few more tech talks and a little more quiet politicking I was going to get us on the Modern Web Applications (TM) bandwagon, and yet, seen from the clarifying distance of a few years, I bet it is highly likely that you'd have to be really adept with the subversion command line to find traces of my lasting impact on that company as employee #4,256.
This is one reason I really prefer (personally) working for smaller companies. I like knowing that the stuff I work on matters. (And I totally get that there exist compensating differentials for working at Google, both for you personally and the other Googlers. You've previously mentioned "It feels great to have millions of people use something I made with my own hands", and I agree, that does sound awesome, and it's a flavor of awesome which will likely not be accessible to me for quite a while if ever due to the don't-work-for-e.g.-Google decision.)
There's actually another major reason why I'm working for Google: I feel that I learn more when I'm exposed to lots of other smart people who are all working on cutting-edge stuff. There are skills I can learn and have learned here - scalability, internationalization, accessibility, security, management, maintaining good relationships with partners in the face competing goals and stressful projects, etc. - that I simply could not learn if I were on my own or in a small company. There are also skills you've learned that I'm somewhat jealous of, like marketing, but I figure that I can pick many of them up on my own time or soon after quitting. And the combination of those skills will (hopefully) make me much more effective if I do decide to strike out on my own.
I thought it was interesting that your article focused a lot on your reputation within other people's minds, and less so on what's going on in your mind. Both of them are important, but I would rather learn how to do good work first and then prove to other people that I'm doing good work than prove to other people that I'm doing good work but not learn any more. I think I came dangerously close to the latter back in 2008 when I was #2 on the HN leaderboard, but fortunately realized there was a shitload of stuff I didn't know. And actually, I'm hoping there's a time in my future where I get to realize how little I know again.