A lot of people say that as a “knee jerk” reaction and with respect, I must absolutely disagree.

Systemd actually has much that is quite commendable, regardless of the portability argument.   Remember that the POSIX standard does not specify the init method. The problem that makes systemd questionable is “feature creep”.  An init system does not need to perform binary logging, for example.    I would be using it without rancor if the systemd team separated the logging function from the rest of the core, enforcing a fixed API that you can depend on for linking, and a better community process of governing its overall design.  The problem with systemd is not the goal, simply the implementation.    

Where I have a real problem with systemd - and apparently Linus does as well - is the lack of quality assurance.  Any userspace (non-kernel) booting tool that interferes with the normal operating of the kernel is a serious problem.  Linus refused to merge his code for this reason.  

From a common sense point of view, rather than that the popular: systemd is no worse than other tools used elsewhere, such as launchd or SMF.   SysV is hardly a “sacred cow”.  If you have been around the block a few times, you will at some point or another, see SysV go completely south on you.  GNU tools like sed and awk used in scripts have awful bugs from time to time just like anything else.
 The hype in the Linux community over the whole affair really shows how far Linux has gone away from FOSS toward a philosophy of FLOS and an overblown sense of their own importance.  

Eventually, all of this will straighten itself out, and if it doesn’t, I foresee a lot of new FreeBSD installs.   In the meantime, those of us concerned with mission critical systems have to eschew systemd, and hope Devuan does a better job.