A hierarchy is actually a tag, so really what you're saying is you firmly believe in only having a single tag for each file and a specific adhoc schema for those tags. I think once you've used a filesystem that behaves in more complex ways you start to realize that's not necessary. For example with a snapshotting filesystem you add dates and times to the path in order to access older versions of files. I don't really use MacOS but I have noticed the tagging in my limited experience with it, which seemed reminiscent of what the old MacOS used to do where you could set icon colors. So whatever they're doing for tags isn't really the only story. MacOS is a good example though because that is a filesystem that maintains separate metadata in addition to the file contents (they used to be called resource forks, but I'm not up on modern MacOS). Many linux filesystems also can support that sort of thing. But those are just flavoring the files and not really adding different ways to index into the filesystem. A hierarchical index such as nested folders is probably a minimally viable solution and is very useful. But that doesn't mean it's not limiting and that there are not different larger solutions that achieve more useful results. Once you start thinking about snapshots and transparent de-duplication and other ideas like don't-repeat-yourself things wind up becoming far less clear.