One strong argument (maybe the only one) for the concept of controlling superstar players is the reality that they
are going to leave, eventually; maybe for a better job, maybe retirement, maybe they get hit by a bus. If your output is entirely dependent on a superstar, it inevitably won't be durable against their departure; one way to look at corporations is through the lens of efficiency maximizers, but another equally valid way of thinking about them is through the lens of durability maximizers. The day-to-day actions of most people inside a corporation is far better pattern-matched toward Survivability, not Efficiency.
Take any company, and grade it (personally!) on both "how ideologically driven is their mission" and "how productive are they"; eight times out of ten it'll correlate. New companies like OpenAI ("we're inventing machine intelligence"; strong ideology/mission); extremely productive. Musk's companies, Tesla and SpaceX, extremely strong, important mission, extremely productive. Some smaller companies like 37Signals, very ideologically/personality driven, very productive. There's even a few big tech companies left, like Cloudflare, that have strong, coherent mission drive ("Building A Better Internet") and have productivity to match it.
One counter-example (and why its, like, eight out of ten): Microsoft. No one has any clue what Microsoft's mission is. That product factory still churns out more new stuff than anyone on the west coast.
Those two paragraphs are 100% connected, because roughly the only way to mitigate (not eliminate) the risk of superstars leaving is high, coherent mission drive. "High" means "its clear, its communicated, its meaningful" and "coherent" means "its practiced, its demonstrated, its genuine". Compensation is another way, but the problems with compensation are (1) you will lose the compensation fight with competitors eventually, (2) it selects for actors that aren't interested in anything except compensation, and (3) it breeds complacency/retirement without simultaneous high+coherent mission drive (Nvidia's problem right now).
Oxide [1] is an interesting case-study in mitigating some negative impacts of high compensation; "Everyone at Oxide makes $201,227 USD, regardless of location."
It just comes down to: Poor Leadership. If the product is boring, if leadership can't coherently communicate mission, direction, value, etc; superstars aren't going to join or stay for very long. Everything else is window dressing. In the absence of that, the most effective way to maintain continuity of operations is normalization and "the bell curve".
Also: If you find yourself believing "well, some companies just don't operate in an area where high mission drive is even possible, not every company is on the bleeding edge of AI CryptoTech", just stop. I don't care what you're doing; building a high, coherent mission is always possible. Amazon's mission is to be the most customer-centric company on the planet. That's a freakin awesome mission that would resonate with a ton of superstars, and its a mission that every damn company could have, if their leadership chooses it and is high quality enough to live it.
[1] https://oxide.computer/careers