They are definitively NOT easy to manage: - you can't know their current state, since anything is deployed one command at a time, and then other commands to up

They are definitively NOT easy to manage: - you can't know their current state, since anything is deployed one command at a time, and then other commands to upgrade and so on, while OSes like NixOS/Guix Systems are just a network of symlinks rebuilt from a config upon any change; - you can't replicate them from scratch, even if you automate, no matter if Ansible/Salt, homegrown scripts, fog or something else, a proper replication demand a gazzilion of things to be manually prepared from a custom ISO or the current official ISO + all you change and you need to keep anything up, again NixOS/Guix System replicate from the config and making a custom ISO as a helper is just few lines of the very same language config; - classic package manager badly accept user modifications, something have done for sources (FreeBSD ports, Gentoo/make.conf and so on) but essentially nothing for binaries, just some rigid and limited automation by Debian with "alternatives" and "configure", again a nightmare to get something custom; - backups? There is NO DAMN REASON to backup an OS, data of course, no matter if homes or /var stuff BUT definitively not an OS, it must be reconstruct and frequently since anything derive more or less quickly and to be reliable you need to been able to replicate with the minimum effort. Just try to compare a NixOS/NixOps managed infra vs a RH/kickstarted or Debian/Preseeded one). Gen-Z probably do not even know what's a file https://www.theverge.com/22684730/students-file-folder-directory-structure-education-gen-z https://www.theverge.com/22684730/students-file-folder-direc... but that's a totally different matter, classic systems including current commercial big ones are archaic crap.