Yes, I would definitely agree that good managers are much rarer than good engineers. Maybe it's because there's an order of magnitude less managers than engineers at a company, so you have less people to emulate and learn from? Maybe it's because engineers are able to use their off-time to improve their own skills, while for management you really only have books and no easy way for applied practice?

And then if we're talking about technical managers and not just CEOs, the literature there is even more limited, since most middle-management in the past has been of known repetitive tasks, rather than more variable more creative work. And it's not like there are schools that can spend four years teaching you how to be a good manager. MBA programs tend to care more about the details of running a business and let the people aspect figure itself out.

So maybe an analogy might be, if all the engineers you hired were self-taught, and if you only had one engineer assigned to each team. Then maybe your average quality of engineer would match your quality of manager.

One factor is that for engineers it is relatively simple to measure performance, while measuring manager performance is usually a very obscure and subjective process, allowing entry and promotions for weak managers.

For example, 30 minutes ago a senior director told her entire organization that the primary differentiation in the performance review for the managers will be "quality of communication". Not only this is subjective, but it is not a core performance indicator.

Yeah, interviewing is definitely harder. Again, we can draw the analogy to engineers. Is there a large pool of managers that have gone through interview training for managers and have dialed-in questions that give them good signal on how successful the managers will be? And then you run your candidate through multiple rounds of these interviewers? Are you running your candidates through different simulated encounters with different people and different setups? Or would you say it's closer to interviewing an engineer without asking them to code? So let's say you're trying to interview engineers without asking them to code, then you probably need to at least see all code they produced at their last job and ask their past coworkers about them. So then the analogous requirement for a manager would be to then interview a representative sample of their past reports?

Measuring how "good" your managers is always a hard topic. That's why I found Google's research on it so interesting, since they had both the means and the incentive to run a large-scale study using themselves as the dataset.

I'm not entirely convinced either that evaluating engineers works out that much better. What is the core performance indicator there? Lines of code? Tickets closed? Bugs created? Reputation amongst other engineers? All of those metrics are wrought with perverse incentives. I'd probably end up preferring evaluating managers on "quality of communication" compared to evaluating engineers on any of those.