I'm sorry, couldn't comment on your previous journal entries, but I think you're maybe going about this with a false set of assumptions.
For instance, regarding portability, you have apparently missed the part where Lennart explained that he's previously written portable software (e.g. PulseAudio) and that he knows about the tradeoffs. Maybe you think of the tradeoff differently, but it wasn't you who had to implement it so that's sort of a moot point.
Regarding GNOME's use of logind, you have to remember that ConsoleKit was dead. Instead systemd showed up with manpower and momentum. This is actually common theme among many of the small utilities bundled with systemd.
Regarding depending directly on DBUS instead of separating the communication mechanism from the API, I think your false assumption is that anyone in the desktop landscape would really prefer that to DBUS. The sole purpose of DBUS is to allow APIs to be developed between separate processes. And there's a real cost to adding abstraction layers of fat.
Regarding logind and systemd, you seem to start from the assumption that logind shouldn't depend on systemd. As far as I'm aware, apart from the small start up notification library (and perhaps udev), the systemd stuff isn't really intended to be run separately?
Personally I'd like to see competitors to systemd form, but I really think the value in collecting and maintaining a bunch of low-level utils necessary for getting a Linux system up and running to the point where you can run interesting stuff (i.e. applications) is under-appreciated.