I have totally separated the physical storage organisation from the logical/functional organisation. I have made my physical organisation efficient for physical storage and my logical organisation efficient and effective for human retrieval.
Physical (on disk):
Inbox (files after copy from camera card)
Inbox/<date> (files after import in PSU)
RAW/bucketxxxx (archive location for original files (not just raw files, but mostly), moved to final archive location from within PSU)
Derivatives (files as intermediate for e.g. pano, HDR, focus stacking, etc.)
Output (files ready to be put on their final format, be it web, print, email, etc.)
Logical (in PSU):
Import/<date> (files imported in import step, i do this next to the standard "State-In Last Import" to retain the original import date indefinitely)
Categories (i add categories to the files, most notably people names form y family, relatives and friends, but also sets of files like pano sets, HDR sets, etc. (i created my own categories for this))
Also, while i'm an amateur, i never (like: never) throw away images, not even test files of other "bad" ones. The archived files are backed up on my NAS and in 2 cloud backup services (1 US and 1 EU). My total archive in PSU is now a hair over 215K images.