If we remember files by context, why do we still force structure?
While building a file-related product, I keep coming back to the same question:
If people remember files by context,
why do we still force structure?
People don’t remember files by folders.
They remember them by situations — something from a project, something they opened recently, something they downloaded and meant to revisit.
But file systems are still built around fixed structure.
Folders assume a file belongs somewhere specific.
Tags try to help, but still require manual upkeep.
The problem is, context doesn’t stay still.
What mattered last week might not matter now.
Something useful in one project quickly becomes noise in another.
Over time, the structure stops reflecting how you actually work.
That’s usually when everything falls back to search, recent files, or memory.
Which makes me wonder:
are we actually organizing files,
or just building systems we eventually stop relying on?
Curious how others deal with this —
do you maintain structure over time, or mostly rely on search + recency?
I ran into something similar building analytics dashboards — users would set up these elaborate folder structures for saved reports, then always find what they needed by searching "that chart from last Tuesday's meeting."
The thing I learned is that save-time organization feels productive but it's basically busywork. The real question is whether you can get away with zero structure at write-time and just make the retrieval layer contextual enough that nobody misses folders.
That’s a really interesting way to put it “save-time organization as busywork” resonates a lot.
I’ve been thinking in a similar direction while building Voyager.
Instead of trying to get people to organize things upfront, I’m exploring what it looks like to let structure emerge from how files are actually used over time.
So not zero structure, but not something you have to maintain manually either.
More like something that stays aligned with context — recency, activity, what you’re working on rather than where you decided to put it once.
Still early, but curious how far that “no structure at write-time” idea can go in practice.