The evaluation process is quite sophisticated, with thesis committees, talks, reviews, etc.
That's much more sophisticated than for software engineers. I'm not trying to imply SEs are evaluated by sprint planning points and velocity, but a lot of that comes down to sprint planning points and velocity. Beyond that, you have performance reviews, and I've never been in a company which did that well.
Code reviews are nice. But even a lot of that is pattern matching sanity checks, little better than a lint program, when a team is rushed.
I mean, do Ph.Ds know how to architect maintainable code? Perhaps not. Have they been encultured in a culture which rewards cowboy individualistic douchebags? Without a doubt. Are they arrogantly convinced of their own superiority? For the first few years after graduating, at least from the "better" schools, more likely than not, yes.
But I think your comments about lacking esoteric knowledge are completely off-base. Ph.Ds thrive on esoteric knowledge.