I agree. I think 80% of what people wanted from systemd is in OpenRC with 0% of the risk. Had init changed some to add most of the functionality people want in systemd, systemd wouldn't have evolved. But it did and now we should be dealing with the world as it exists.
If someone wants something that is equally problematic today were individual solutions are evolving, we need a networking stack that has a better understanding of latency. Now is the time for the old school Unix guys to be fixing that, before the pot boils over.
The architecture doesn't matter too much. Systemd itself will be easily replaceable. For example FreeBSD is porting Apple's launchd over (openlaunchd). Given that systemd is based on launchd and that the FreeBSD people are going to want to run Gnome... I suspect that by 2020 or so openlaunchd would be an alternative, and likely one without the architectural issues.