Well, not all infrastructure work is confidential. I work on the Production Linux team at Google, and we're quite open about a lot of the work we do, and we tr

Well, not all infrastructure work is confidential. I work on the Production Linux team at Google, and we're quite open about a lot of the work we do, and we try to push as much as possible of the work we do upstream. I do have to second this thing about the "celebrity programmer". We have a group of people working on ext4, some of whom have been working on it at Google before I decided to start working there. (In fact, I learned about Google adopting ext4 from their contributions of no-journal mode for ext4 to the upstream.) I may be one of the more visible persons at Google, but I'm certainly not the only one working on ext4, and ext4 wouldn't be where it is today without some really talented folks who have been working on it with me for the last 2-3 years --- and I've only been at Google for 18 months! One of the reasons why I am at Google is precisely because it is a big company. It is only when have a very, very large number of machines in your data center does it make sense to put a whole team of people working on infrastructure code such as file systems. (I'm pretty confident, by the way, that Sun never recouped all of the money they poured into ZFS --- it's a project which is highly impressive from a business perspective, but from a business person's ROI perspective, not so much.) One of the things I can definitely say about Google is that while it may be a large company, compared to IBM, it has a minuscule amount of bureaucracy. I used to sit on the team that worked on the annual budget for the Linux Technology Center, and that would be weeks and weeks of conference calls where the ROI of work items at the granularity of 6 to 12 persons months worth of effort for the next fiscal year would be weighed, judged, and decided. Which is something I'm glad IBM does as a shareholder --- but as someone who had to sit on those interminable conference calls, not so much. I get to do far more real, honest coding work at Google than I did at MIT (where I spent the bulk of my time on Lotus Notes, PowerPoint via Crossover Office, and conference calls.) The one good thing about having gone through a very dispassionate review on the ROI of various different bits of technical work is I can say with fairly great authority that there is huge amounts of work which makes sense from an ROI perspective for me (and a whole team besides) to do at Google, but which wouldn't make sense at most other companies, just because of the scale at which it operates. If you are interested in working on very large scale systems problems, Google is definitely one of the best places in the world for you to do this kind of stuff. I know a tenured faculty member at Harvard who decided that it was far more fun to do systems work at Google, and who left his tenured faculty position for this reason.