Let me set the stage.
I'm working with 100+ other developers and probably 300+ total people on this one project. We're subdivided into 20 scrum teams of ~5 developers each.
So far so good.
The problem is that I'm working for a non-tech Fortune 25 company. The person who signs our checks and justified this project is literally six levels of management above me and reports directly to the CEO. To quantify progress and justify the cost (hundreds of millions of dollars) our project is treated as its own organization which has to satisfy a service-level agreement (SLA). This SLA contains completely arbitrary metrics like velocity, defect count, and myriad other things which are only tangentially related to software quality and execution.
This has resulted in the most ridiculous management strategy wherein everything is locally optimized rather than globally optimized. Here are some examples:
1. At the end of every three week sprint, there's a rush to complete stories because if anything is moved, management looks bad. Software engineering standards go out the window because there's literally not enough hours in the day, and nobody is willing to work overtime just to make some manager look better.
2. Speaking of poor software engineering standards, there's so much pressure to close defects immediately that the standard practice is to hotfix/patch everything rather than solve underlying issues. This has created enormous amounts of technical debt in our code base, making working on anything you haven't touched nearly impossible.
3. We're subjected to completely arbitrary pressure any time we're missing the metrics in our SLA. We're not in production or even close to it, and yet our management has had the gall to force six day work weeks across the entire project so the defect count could be brought down. Our strategy? More hotfixing/patching and pushing defects off to other teams, because when you're giving up your Saturday based on some completely imagined emergency, there isn't much incentive to do things the right way.
The end result? Extremely brittle software which satisfies acceptance criteria and has a pretty veneer, but will break the moment you sneeze on it. 200 million dollars well spent.