What you're saying sounds very nice and correct, except it isn’t necessarily true.
It's extremely easy to draw wrong lessons in retrospect. There are so many variables, including personalities, market conditions, timing, constraints, and accidents of history. You can't recall or even really understand these things with any level of accuracy.
What ends up being most useful is the way experience fundamentally changes you as a person, not your regretful shower thoughts posted on Twitter.
So it may seem counterintuitive, but if John Carmack wants to create another breakthrough technology, he might be better off re-creating id Software’s in 1995, including the chaos, rather than trying to avoid it by applying all his "lessons learned".