> There's no end in either direction you look, but zooming in always reveal some simple interactions at work.
This is why I actually hate the phrase "from first principles". I can't tell you what the first principles are, but I can sure tell you that when someone says that they almost certainly are not starting from first principles and almost surely are (often unknowingly) making major assumptions. First principles are something you need to find, not start with.
You're right, and it's subjective because of priority. First principles should be tied to whatever thinking requires the most concern in the moment.
If you're in charge of budget, you probably put budget before some other minor concern. If you're in charge of delivering a software feature, bug, patch, etc, you probably put something else first. It all depends on when something is due, or if something could jeopardize long-term prospects.
Everyone is on the hook for something. First principles thinking might be: think of your customer, and what they expect. Will whatever it is I'm doing affect them over the long-term? What about short-term?
If there's no clear answer, there's a lot of latitude to do whatever you think is best.