Sufficiently good managers
I see engineers complaining about how good managers are rare1, and I would like to suggest that perhaps they haven't thought it through. If the managers were really bad, they would be moved elsewhere. On the contrary, there are plenty of sufficiently good managers out there, it's just that the "goodness" metric applied is not what you'd expect.
Given that engineers like to self-manage, they mostly benefit from manager's greater experience and feedback (if up-levelled from an engineer), social skills ("have you talked to them?"), and interfacing with the rest of the company (career management, help with finding adequately challenging opportunities, inter-team cooperation.)
However, from the perspective of company owners, most of these functions are not very important. Unless the startup is growing explosively, there is not enough money for significant raises and no space for promotions, resulting in high employee turnover and low return on investment per person. Managers are mostly promoted senior engineers with no extra training (career growth, self-management, and money saving!) so their social skills are on-par with ambient levels in engineering.
On the other hand, a startup lives or dies by its product, so owners need managers to deliver features, preferably often (sprints ahoy!) Secondarily, it's also nice if they keep engineers too busy to think about moving away by flailing about what to build, communication chaos, and endless re-orgs.
Between company owners and engineers, who decides promotions and pays salaries? The owners. So their metrics are used to determine whether a manager is good. You know how products have usability problems because decision-makers don't actually use the product themselves? Same thing applies to engineering managers. They are sufficiently good, as long as they deliver.
Disclaimer, we're talking about startups and startuppy tech companies here. Giant corpos apparently do things differently, and have a proliferation of different kinds of managers, some of whom are even trained in DISC assessment. I wouldn't know, I refuse to work for Accenture.
↩