The biggest "carrot" held out to idealistic people wanting to become a manager is exactly this: the (fake) opportunity to "fix things". Only after you've become a manager does it become apparent that long-lasting problems are long-lasting for a reason and that none of the choices low-level managers are allowed to make can meaningfully move the needle. Sometimes they fall for it a second time, but being a manager of managers is often even worse since now you don't even have direct reports to effect change.
The former is necessary to coordinate coarse-grained decision making, policy and vision that needs to be unified across thousands of people, most of whom will never meet each other, but who should ideally be rowing in the same direction.
The latter is necessary for the armies of individual contributors to get their respective jobs done. Trying to document and formalize this ad-hoc network holistically is impossible because it's too complex to be understood by any one person. Attempting to do so would require a non-trivial time commitment from all the workers, which would actually take away from them, you know, getting work done.
It's tempting to look at an org chart and assume this represents how things work on the ground, but Conway's Law is more ironclad than it may appear at first glance. Don't confuse legibility for operational capability. If lower level folks did not understand their goals and improvise, then large corporations would be even more rigid and brittle than they already are. They would be utterly incapable of responding to changes in the marketplace and smaller firms would dominate.
Organization / chain of command? Parts information? Charge Codes? Messages/Sentiment? Business Information Systems capture way more data than mahogany row might realize, and you can get some "split the atom" visualizations when you combine the right parameters, like RnD funding + messaging. "Huh. Looks like new tech needs a LOT of communications with the field technicians. Like, a LOT a lot - totally wiping out bandwidth in remote locations"
The data is maybe there but if you model everything at once it's going to be a minimally-significant graphviz blob.
Stupid punchline? Execs hold up the blob as scientific proof that their job is hard. "Don't change it! It's so complex and pretty! I can show this to the VP-Manager of Goofball Systems Inc to validate my existence!"
No, it's not hard or pretty, you just can't tell the difference, conceptually, between a hex driver and a lathe.
(As a former military officer, this is the same for generals btw. They may have a lot of "power" in the organization itself, but they're heavily constrained by outside factors. They have to make do with the budget they're given, and have very little control over hiring targets etc. Not to mention that during wartime the enemy will not be under control either)
