One thing i dislike that systemd brings are binary log files.
There's no reason you can't keep using syslog-ng/rsyslog with systemd, and the distribution can make it work out-of-the-box (as syslog-ng does with systemd on Arch).
The journal provides structured logging (fields of data, with text or binary data like a network packet), indexed by fields, and with features like inline compression. Other loggers have been moving towards structured logging too for quite a while, rsyslog being one of the leaders (binary formats too). It's true that journald could have used a pre-existing format like sqlite3, but it wouldn't have been as specialized for the use case. The journal file format is documented on fd.o, and there's a backwards compatibility guarantee. You can also easily export the contents as text, if you desire.
The journal is simply another choice, that brings a different set of tradeoffs to the table.
Kay Sievers was also the udev maintainer before it was moved into the same repository as systemd to share code, and this problem (the 30 second timeout issue) predates that. The followup conversation between Kay and Linus is right there: https://lkml.org/lkml/2012/10/3/263.
I read those lkml links and I certainly agree with Linus on the "don't break things" philosophy. Kay seems to have a nilly-willy attitude about it - "oh it's broken, oh we'll fix it sometime... don't fuss guys".