What does Sarbanes-Oxley have to do with locking down parts of the source tree? I worked there a couple of years after you did, and there were a few parts of the source tree that were not public, but that was usually for crypto reasons. Config files, not algorithms.

I'm not sure what to tell you about Google today versus then. Yes, you will be a cog, but in one of the shiniest and most well-maintained machines ever. There's a non-negligible chance of doing a significant 20% project.

The one thing you'll notice is how much stricter the standards are for testing and code quality generally, while the codebase has expanded exponentially. Sometimes this results in code that's so great it practically makes you weep. Other times, especially for the really old projects, it becomes a morass of incomprehensibility but whose quality is carefully husbanded. I knew of a guy who spent weeks getting one change into the basic webserving code, because running all the tests took an entire day, and by the time he was done, other people had committed new changes that broke his change. It was Xeno's Changelist.

Maybe you can answer me one thing: in the early days, did people think they were a moral force for good in the world? As Dave Winer correctly notes, one of the amazing/insufferable things about Google is that many engineers there really think this -- especially the pre-IPO crowd. It's the sort of attitude that enables them to open for business in totalitarian China, because how could you deny the Chinese the wonderfulness of Google?

Also note the proliferation of teams which make promotional videos starring themselves. This started with a team that happens to have a very charismatic dude working for them, but it seems that everyone's doing it now.