I believe many developers and designers have been annoyed by hierarchies in filesystems. But I wanted to comment on this: > there is a mismatch between the nar

I believe many developers and designers have been annoyed by hierarchies in filesystems. But I wanted to comment on this: > there is a mismatch between the narrowness of hierarchies and the rich structure of human knowledge Absolutely true, this is exactly what I think annoys us most than anything, it shows us how limited hierarchies are. But at the same time, I think it's very relevant to keep in mind that our knowledge and the mental relationships we can find between ideas are very hard to make explicit and complete, like you would ideally want in a tag-based filesystem. I feel serious tagging, if manually defined, it's quite expensive if we want it to be really useful (surely, we can also consider complementary automatic tagging, like AI). Hierarchies instead, might not be very expressive, but they are very simple to use in "most" cases. So I would say we're still far from getting the best of both worlds. The problem to "solve" is information organization/structuring, and not even humans handle that ideally (we are more like, faulty, search engines with random inputs, prone to forget XD). About the other ideas, I think they are all interesting, agree a lot with hashes usage and no-filenames, not so convinced about metadata, but haven't really thought enough about it. I don't think we can talk about the ideas fitting cohesively or not yet (but hey, I don't even think links in HFS are cohesive from any perspective), we would have to see more formal proposals for implementation and interface. This said, I hope we see more work along these lines in the future, it's a very worthy field to explore! Maybe start small, testing some of the ideas, we get a lot of design insight when we are working on the implementation.