You are so vague in your attack on Unix approach that it's borderline trolling. What are your problems with it? Modularity and minimalism have been working perfectly and that systemd does not follow them is a bad thing.
But that book is a waste. It is just MIT dunning-krugerites who were salty that LISP machines never took off. When it comes to real life, the bell labs approach won, and for several good reasons. Not "worse is better" (another dunning-krugerite cope), but "less is more."
From your perspective, what would be an "OS done right"? I have a running list of things I would change in Unix, but replacing sysvinit with systemd's one-ring-to-rule-them-all would not be on it.
But your comment is a waste. It is just HN dunning-krugerites who were salty that the UNIX way never took off. When it comes to real life, the Poettering approach won, and for several good reasons.
The UNIX way is still doing fine on OpenBSD, NetBSD, FreeBSD, Alpine, Gentoo... Poetteringware only won on the distros selling support contracts. "Fixing" what wasn't broken is great for those businesses.
