Maybe the answer is tags/attributes for filtering, instead of nested directories? Could be more tedious but given the image recognition ArTiFiCiAl iNtElLiGeNcE maybe it could be easier
That's what I'm thinking about too. Most directories are describing what's already available for the computer to extract: the place of something being created, the time of it, the source where it came from (e.g. who sent this as email attachment). Only a minority needs to be explicitly added, like emotional meaning ("my favorites"), or external organization ("experiment 1", "driver's licence"). While we can't use the computer for the latter, it's nonsense that we leverage computers so little for the former.
I think it is. Directory hierarchies are trees, tags are graphs. Trees are great when you need to optimise the physical retrieving time, but don't map well to stuff like pictures. For example, tags of who is in the picture are way better than making folders by person. Same thing for places.