This was my experience mostly in my 10 years at Google at a certain level.

But I will say this: at a certain point in a large company once the revenue-machine is discovered and deployed, what you want to be building is systems that let you ship and build reliably on top of that foundation without destroying it.

Google in its best phase -- which was already in decline when I joined in 2011 -- did have a slow and cautious development cycle where multiple levels of review covered everything. OWNERS, "readability", very uptight code review. And in order to survive in this environment you had to have a pile of code reviews all running concurrently because making progress on any single one could take days and days to get through review.

But that was kind of the point because pushing the wrong thing and breaking the money printing machine is far worse than moving slow.

But IMHO this didn't scale past 30k, 40k engineers. And inside Google, the culture shifted from one that was SWE/SRE driven to one that was PM driven. And the perf/promo culture for them had really perverse incentives.

Also I have a theory about Google in particular -- its founders and all its initial strong hires all came from academia not industry. And so its internal culture became biased towards a "publish or perish" structure, and "perf" performance reviews honestly looked more like a thesis defense committee for someone's masters/PHD than anything I'd encountered in the software industry before.