He also was a serious cocaine addict who injected it every day. It's of course an oversimplified historical narrative, but at least part of the reason medicine does things this way is the traditions were set up by someone out of his mind on coke.
One study compares the ACGME limits (max 24 hour shifts) against a "flexible" system with 28 hour ones. Unsurprisingly, there's no effect--since both groups are totally zonked at the end. Meanwhile, data from lab experiments and virtually every other field (trucking, aviation, etc) shows that performance craters well before that.
The rationale for long shifts (other than money) is "handoff": when a new doctor takes over, details about the patient's condition and treatment plan get lost in the shuffle. In some cases, they might not ever have been recorded (e.g., a subliminal impression of a patient's color or breathing). Many people are weirdly nihilistic about improving this, but it seems like it almost has to be the lower-hanging fruit since solving it doesn't require butting up against fundamental limits of human physiology.
Another thing is hospitals adopting air industry style checklists, which doctors fought tooth and nail as 'insulting'. Despite every ounce of evidence massively pointing toward this saving lives.
There's, I suppose, a weird societal level "getting your money's worth" from their compensation - since their actual hourly rate would plummet on the basis of working 30 hour shifts non-stop and the like.
Not to mention that the institution as a whole has enormous difficulty actually properly handing off care between different providers, yet we would only see benefits from being effective at doing so - not to mention minimizing the impact of bad doctors.