I enjoyed reading the newsletter mini-article on which this discussion thread is based. I can't fault you for spending more time discussing Novell than VMS. I'm guessing that most people have never heard of VMS. Too bad, but that's DEC's fault.
One point that you mention at the end of the article was that many Unix sysadmins give blanket access to files because of their ignorance of file system security. That was also a big problem in VMS. Anyway, regarding Unix, I have spent, and continue to spend, a lot of time experimenting with file access settings. The default settings from the installation of any Unix that I've seen always seems to reflect a time when security was seen as an unneccessary burden. Back in the 1960s and 1970s many multiuser systems were thought to be inhabited by a community of trusted users. The idea of hiding or securing system files from these users would have been considered insulting. That was then and this is a different world, but the default file access settings in a Unix system still reflects this open door approach.
Some of the things that I'm trying on my home computers is to make most of the top level directories owned by root:users with drwxr-x--- permission. This keeps the directories inaccessible from accounts that are neither root nor in the users group. Accounts like nobody, ftp, mail, and other function specific accounts. I'm still playing with this.
Another idea is to make the /etc/*.conf files inaccessible to nonsystem accounts. After all, why let curious users look at the system configuration?
I've just about determined that the 'requirement' to have the /etc/passwd file readable by world/other is simply to allow certain unneccessary utilities like finger to work. Therefore I've set that file to world/other noaccess while maintaining the root:root file ownership with no ill effects. I haven't tried this on a production system yet. I don't believe in letting users run finger. I'm about to play with restricting access to the /proc directory. Normally privileged users can see just too darn much about the system and running processes when /proc is configured with its typical access settings. I haven't actually tried changing this yet. Maybe the system will refuse to boot. I don't know. Anyway, this is the sort of thing that I enjoy testing.
Given all of that I still say that ACLs can be useful when you have a very detailed file access situation. Normally, however, the use of Unix secondary groups can provide a pretty high level of file access flexiblity. Even if you never allow world/other any access to data files, for instance, you still have two levels of access to play with in the old style file access control. The file is owned by a user account and a user group. A little imagination can help you get a lot of mileage out of this simple scheme.
So I reiterate from my original post: Use ACLs only when you absolutely need them. ACLs can quickly become very complicated. Use ACLs sparingly and you will save yourself a lot of headaches troubleshooting resource access problems.