Lets get this straight the basic ideas embodied in the original systemd are fine and in many respects are an definite improvement over sysvinit.
But in case anyone has not yet got it let me try to explain.
The primary problem with systemd as it stands is it's becoming (perhaps is) no longer just an init system. It's growing into a major operating dependency of the GNULinux system. Soon GNULinux will not be able to run without it.
One of the cornerstones of Unix-like systems is that there has never been any part of the operating system, not even the kernel, which depends on the necessary existence of another part. This is what makes Unix-like systems so flexible.
What happens if another, much better, init system is proposed in the future? It will be impossible to introduce it because systemd will have it's tentacles so deep into everything that virtually all the operating system will have to be rewritten.
I remember devil saying, when systemd was introduced into siduction, that it will do until something better comes along (and he quoted a few examples). I'm sorry to say we are almost at the stage when devil's hope will be but a dream. It will never, ever happen. We'll be stuck with systemd for ever. No GNULinux software, even an init system, should control so much of the future developement of the operating system.
This isn't the flexibility which unix-like systems offer; it's a strait jacket.
If systemd is to remain part of GNOLinux (and I think it should) then it should restrict itself to being an init system and cease being a dependency for the rest of the system.
Aside:
Why are most of you so afraid of losing systemd? Lighten up! I can understand the distribution devs wanting it because it makes life a darn sight easier for them. But the ordinary user who hardly cares what an init system does?
And who in the Debian emails I quoted in my original post has expressed a wish to go back to sysvinit? Nobody. Ensuring there are choices is not necessarily a retrograde attitude.