Literally the only argument I've seen that is even close to reasonable
is that some people like text logs and journald is a binary log format,
and fixing that requires adding one line to a config file.
Please, someone explain this to me.
I'll give it a shot
See, traditionally, this isn't how things are done in the Linux
world. Generally, if you have a snazzy idea for an init system (for
example) what you'd do is offer it as an option and let users decide.
This just isn't true. Ubuntu didn't offer users a choice with upstart, and other distributions never seriously offered an alternative to sys v init. Distributions offer choices when it's convenient/possible. Most distros don't offer alternative kernel builds because it's a pain to do and they feel it's not worth the effort. Most (all?) distros don't bother supporting the BSD coreutils in place of GNU (and in fact many packages depend on them being the GNU versions). Once again, it's not impossible it's just not felt worth the effort.
It takes real effort to support multiple init systems, so the question becomes, is it worth it? The people actually doing the work in many distributions don't think it is. You can either work to change that by putting in your own time/money, try to convince them that it is worth the time, or just use systemd.
If systemd is as good as its supporters suggest, then it'll become widely adopted without all this ballyhoo. Conversely, if it's failings are severe enough that it can't gain widespread acceptance without politicising the entire debate, then I don't want it anywhere near me.
This is exactly what is happening. Distribution maintainers are choosing to use systemd because they find it the best of the options available. Nobody is strong arming anyone as far as I can see. Also, once again, the only people I see making this political are those who seem to find systemd emotionally repulsive. All of the arguments I've seen in favor of systemd are purely of the "it works better and has more features" variety.