“Why don't I just toss all that stuff in a big box (so roughly sorted by time) and go through it in the rare event that I need something someday?”
This is basically what I do with nearly all my paper records. It's a sort of JIT filing system. “Writes” into the document store take very little time. I just dump everything into a folder for the current month. I keep a ring buffer of 12 months worth of docs. When a dirty buffer comes up as current, I quickly sort through its contents for the rare item that needs to be preserved for longer than its already year-long storage, file it in a more persistent place, then toss the rest.
This comes at the expense of making “reads” from the document store more expensive. But as you point out, fetching docs from this system is relatively rare. It's more than worth the time (and cognitive overhead) to have to occasionally flip through a couple date-delimited folders to fetch a document I need to retrieve.
This is basically what I do with nearly all my paper records. It's a sort of JIT filing system. “Writes” into the document store take very little time. I just dump everything into a folder for the current month. I keep a ring buffer of 12 months worth of docs. When a dirty buffer comes up as current, I quickly sort through its contents for the rare item that needs to be preserved for longer than its already year-long storage, file it in a more persistent place, then toss the rest.
This comes at the expense of making “reads” from the document store more expensive. But as you point out, fetching docs from this system is relatively rare. It's more than worth the time (and cognitive overhead) to have to occasionally flip through a couple date-delimited folders to fetch a document I need to retrieve.