I'm not sure dot-com era DEC had much stagnation or institutional knowledge that wasn't continuously overrun, nor would it be comparable to most big orgs these days.
He started in a greenfield industry, the immediate pre-requisite for dot com era, then through the dot com era. There was no institutional knowledge when he started, and a good portion of it would be irrelevant when he quit. It was all new.
When he started at DEC in 1982, they had 67,000 employees and almost $4 billion in revenue. It seems like that type of success and size would imply some institutional knowledge. Their revenue, income, and employee count started stagnating in 1989. He worked there until 1998.
No need to keep going back and forth on this as you seem to have dug in your heels.
It's not back a fourth as much as you think I've stated some hard black and white rule, without exceptions. I think it's generally true. In this case, is the exception DEC or Jim Keller? Would he agree? I don't know. Some large orgs run like a collection of startups, internally.
But, I don't think DEC, a company working through the beginning of computer through peak dot com era, where every aspect was doubling or completely changing every year, is a context where holding onto ideas formed in an old context was viable or possible. You would, necessarily, have to temporarily suspend your trust in the institutional knowledge, with every new problem, since the whole compute world that the institutional knowledge was built on would have shifted under you.
Jim Keller’s own biography kind of dispels this notion. He worked at DEC for 16 years when he was a young engineer (24-40 years old).