The only good thing I can see about systemd is the exposure of some Linux system APIs that were not exposed via the POSIX subsystem. Nice - but not required by most of us - and could be added to existing standards based initialization daemons without totally rewriting the rules.
Otherwise it seems to be more likely a thinly veiled (actually not veiled at all...given comments of the principles) attempt to fragment the POSIX world - and forcing projects with limited resources to make a Hobson's choice of whether to support systemd based Linux or POSIX standards exclusively. It breaks write once - compile/run anywhere - that was generally available for those who made sure their applications were POSIX compliant. This means that a lot of software that was available across Linux, and Unix flavors (BSD) will now be exclusively available on one or the other - thus fragmenting the *nix world.
Software is not separate from the ethics that surrounds it. This approach and apparent rabid anti-interoperability view is arrogant and self-serving at the expense of cooperation and choice. Furthermore, the monolithic architecture, obfuscated binary logs, and centralized configuration are antithetical to the Unix way - and makes a linux system as difficult to deal with as a Windows system from an automation and management perspective, and raises concerns in terms of security (the greater the complexity in a system, the greater the opportunity for bugs - and thus the greater the attack surface).
Finally it throws away many many years of experience/knowledge acquired by system admins, developers, and users about how a *nix system operates and is configured. This fragmentation of the human factors aspect will by its very nature cause faults/issues during operation.
So - for a host of reasons, I believe it is technically - and more importantly - ethically wrong.
There is actually one more good thing I can think of: it will spawn new distros, software projects to provide alternatives of various applications in the stack, and perhaps new operating systems altogether - with a renewed focus on design simplicity (KISS) and all of the benefits that come from that. Once a system becomes too complex to understand - are you sure you can trust it?
So to recap: systemd has two things going for it; exposure of Linux APIs, and the power to breath life into the further exploration of alternatives in the OS/application layer.