And a lot of software don't use syslog because it's easier to print to stderr/stdout or some random log file. Journald makes it easier to capture everything no matter what the software does, including the established syslog protocol, so I don't even see your point.

If everything on your machine uses syslog, journald is a drop in replacement for the dozen of possible syslogd implementations.

journald isn't drop-in because it only saves logs locally. syslog also is a protocol to send your logs to a log-server.

And the usual syslog APIs are 2 lines: Initialize with openlog(stream, process_name, flags) and after that do syslog(urgency, message). That is on par with stderr/stdout, and far simpler than handling your own logfiles. Except if you use log4$yourlanguage or something, then everything is just the same, you just configure a different destination.

And if you can't change your code to not use stdout/stderr, you can easily do yourcode | logger -t yourcode -p daemon.info 2| logger -t yourcode -p daemon.err