edit: In the contrived scenario you referenced, I agree that the internal analysis wouldn't begin and end at simply "put the users first!" :P But since the vas

edit: In the contrived scenario you referenced, I agree that the internal analysis wouldn't begin and end at simply "put the users first!" :P But since the vast majority of activity at Google isn't in that critical path, I'm not sure how useful the thought experiment is. >> if someone higher up the chain can ... then it is manager driven Again, it's not this simple. Things have gradations in the real world. I've seen the "business case" be ignored because a VP doesn't think they could get their engineering team excited about it because (a) it's uninteresting or uncomplicated to implement and their perf-focused workers will transfer to other, shinier teams or (b) it's counter to the project's culture or philosophy and they will similarly lose people. Google is "management driven" in some scenarios and "engineering driven" in others. It's part of the gallows humor in Google that some mucky muck or PM is unable get their team to do something we think makes business sense. edit #2: This is what is really meant by "bottom-up" in the case of Google. The specific incentives from their performance review system, the size of and breadth of projects at the company, and the degree of mobility of its engineers combine to give engineers concrete leverage over their management. I've worked at 4 very large companies and Google is an outlier in this regard. A number of success stories at Google can be traced to that structure, and at other times it's criticized as leading to inefficiency or redundancy, lack of focus or long term commitment, etc. I'm not even sure what we're arguing about. Of course ads dominate the company, it's what drives the most revenue. If you're trying to have an informed view of the place that would allow you to predict or interpret some specific thing they do with some level of certainty, you can't just look at the mean statistic.