[JoshTriplett]

I participated in those discussions first-hand, and recall them rather well. At no point was "we don't have a choice" taken seriously as an argument; even most of the pro-systemd folks didn't consider that a valid argument for systemd, and many of the most judicious and respected voices in the discussion called out such arguments as being unhelpful. Debian is large enough and stubborn enough that they're more than willing (occasionally entirely too willing) to do their own thing if they think they have a better path.

As for the rest, I consider https://news.ycombinator.com/item?id=23062725 a good start. systemd was successful because it provided working code that people (including distribution maintainers) wanted to use, for a variety of reasons. This is the type of problem where no amount of architecture-in-a-vacuum discussions make up for actually doing the work and handling all the myriad cases that come up when people use it in practice. systemd did that; the next init system after systemd will need to do that as well.