Sorry - that sounds sarcastic, I'm interested - and I think that the discussion is useful for others . For example - I took a picture on a beach in cornwall in 2004 of my mate Neil, who I see every year or so.
it's keyworded 'pednevounder' (beach) 'cornwall' 'friends' 'Neil Brockbank' and filed in a folder:
I need to find some shots of Neil as he want's stuff for the band website. I don't even remember the day on the beach, let alone that Neil was there or that I took a picture of him. But all I have to do is to bung 'Neil' in the search and it's found (together with a load of other shots I've forgotten).
I'm sure you can do the same - but I'd love to know how?
Your system sounds to me EXACTLY what I used until a couple of years ago - the idea of going back to it fills me with horror! But perhaps I'm missing something fundamental here?
For web you send it to the relevant web page and delete it - as you would do for output to clients. The actual mods are held in a version, the output files themselves can then be deleted. In each case the versions used for the job can be held in a special 'project' or 'folder' within the DAM - but with no disk files associated with them - on the other hand the disk files can be re-created easily.