Linux and Python are both projects where people can vote with their feet, though. The projects themselves are not a democracy, but they're certainly not the same sort of ballgame as employment. If I decide my management chain is doing the wrong thing, there's a huge personal cost to finding another job, and little guarantee it will be better. If I decide Linux is doing the wrong thing, I can just stop using it or contributing to it.
That is - the fact that Linux and Python have BDFLs is not the reason for their success. The fact that they have effective BDFLs who make sound decisions that actually work is the reason for their success. And the BDFLs only have power insofar as the community still believes in them - Guido stepped down and Linus has faced significant opposition.
BDFLs and traditional corporate governance are quite efficient, yes. But democracy doesn't aim to be efficient; it promises to be pointed in the right direction. There's no sense in going at 100x speed in the wrong direction. Democracy isn't the only way to get there - if you have tangible data about leadership and where they're pointed and how they make decisions, and the cost of switching is fairly low (so that you can keep pressure on leadership), that works too.
When was the last time you had any data about how effective your management chain would be when you started your job?
(I'm happy in my current job because I asked hard questions about management at each place I interviewed, and I found one where my own management chain seemed to have good people - but even so, only one person in my management chain is still the same as when I interviewed, so I still took a gamble, and I had far less data on any of them from a bit over a day of interviews than anyone has on Linus or Guido.)