Recovering from some kinds of filesystem corruption could be made simpler by
storing backups of just the metadata of the filesystem (which would, of course,
require very little space). Files that are dropped into lost+found could
easily be identified and returned to their correct places. If a hash of the
file contents was also stored, then it'd be possible to detect which files
needed restoring from backup after filesystem corruption and which were still
intact (thus potentially reducing data loss since the last periodic data
backup). The hashes could also be used to detect duplicate files.
This information is, of course, a superset of the locate database, so it
could be folded into that; you could then say things like "find me all the
files that have the same contents as this one".