About Moviography
You remember far more than you think. Not the films — the evenings. Who took you, where you sat, what somebody said on the way home, and which one you weren't old enough to get into and got into anyway.
Nobody built anywhere to keep those. So: this.
How it comes back
Almost nobody can answer “what films have mattered to you?” cold. It is too large a question and it points at the wrong thing — at a list, at taste, at what you would defend in an argument.
Ask instead about a person, a place, or a stretch of time, and something different happens. The memories arrive on their own, and they arrive with detail nobody could have retrieved on demand: the smell of the lobby, who was crying, the drive home, the argument in the car park afterwards.
That works because a film was never watched in a vacuum. The experience and the picture are bound together, and the picture holds on to the experience long after you have stopped thinking about either. Pull one and the other comes with it.
So the archive asks small questions and follows what they open. One answer names a person; that person turns out to have been at four other things; one of those was a cinema that closed in 1998, and now there is a whole year you had not thought about in decades. It unspools.
The point is the surprises. Not a list of what you have seen — you can get that anywhere — but the shape of your own life along an axis you had never looked at it from.
What it holds
Memories, not ratings
A memory can hang off a film, a person, a place, or nothing at all. “The smell of the video store on a Friday” belongs in an archive before anyone knows what to file it under.
Whatever you've actually got
Dates stay as vague as you are. “The early nineties” stays the early nineties — it never gets quietly rounded into a year you didn't say. Can't remember the title? “The one with the dog, at the drive-in” is a real entry and can stay that way forever.
Evidence that argues back
A ticket stub is a witness, not a decoration. It can flatly contradict you, and when it does both accounts stay — with yours still on the page. It's your life; the receipt doesn't get to overrule you.
Two ways in
Dump a memory the second it surfaces, no title or date required. Or sit down deliberately and dig through a subject — your grandparents, video stores, the ones that scared you. Neither is the real one, and it's never finished either way.
Where the AI is
Not where you'd expect, and not in charge of anything.
It doesn't do the interviewing. The questions here are written — by a person, on purpose, and revised over time like anything else worth writing. Which one you get asked next is decided by arithmetic: the archive knows who turns up constantly and has never been described, which decades are thin, which subjects you walked away from. None of that needs a model, and a written question can be fixed when it's bad. A generated one can't.
The model's only job is reading what you answer — pulling out the people, places, films and dates so you don't have to file them yourself. It proposes; you dispose. Nothing it suggests enters your archive until you accept it, and anything it touched stays labelled forever.
Which means the whole thing works with it switched off. That isn't a fallback — it's what it is. Turning it on makes the filing better and changes nothing else.
Who owns it
You do — in the boring, literal sense: your archive is a single file on your own disk.
Copy it to back it up. Delete it to delete it. Leave it to somebody. It opens in free software that already exists, with no subscription and no company needing to still be in business. The specifics are just below.
Questions you'd reasonably have
What format is my archive in?
One SQLite database file, plus a folder holding whatever you've uploaded, exactly as you uploaded it. Nothing converted, nothing re-encoded, nothing proprietary.
SQLite because it's the most boring, most durable, most widely readable database on earth — it's in your phone, your browser and your car, and the format is on the US Library of Congress's list for long-term preservation, which is a very dull sentence and exactly the point. If Moviography disappears tomorrow, your archive opens in software that's already on the machine you're reading this on.
How do I get it out?
Copy the file. That's the whole procedure — it isn't an export bolted onto a database somewhere else. The file is the archive and it's already on your disk. A tidier one-command bundle not built yet will wrap it up with a plain-text copy so you won't need SQLite either, but you aren't waiting on us to own your data. You already do.
What if the movie's name is gone?
Write what you have. “The one with the dog, at the drive-in, I was maybe six” is a real entry and can stay that way forever. Plenty of archives are mostly gaps. That's still an archive.
What if I'm wrong about something?
Congratulations, you have a human memory. Your mother will eventually tell you it was 1994 and she took you, not your father. Moviography keeps both, marks them as disagreeing, and leaves your version on the page.
Do I have to fill it all in?
There's nothing to fill in and no bar to fill. No percentage complete, no streak, no profile 60% done. You add things when things occur to you, and then you go and do something else.
Is there a feed? Followers? Likes?
None of those. But your archive isn't a locked box either — everything is private until you decide otherwise, and then you choose exactly what shows: a whole archive, one memory, or just the fact that you saw something without saying a word about it. Sharing is something you do deliberately, to people you pick. There's nothing to perform and no algorithm to please.
What happens to it when I die?
It's a file. Someone inherits it the way they'd inherit a photo album, and it opens without a subscription, a password reset, or a company still being in business. That's not a side effect of the design; it's most of why the design is like this.