Notes on making a story remember
What we are learning building an AI text adventure that keeps a long story consistent: what breaks, how to check it yourself, and what is still open.
Theory of mind: why the best stories are about who knows what
Theory of mind is how we track what others know and believe. Fiction runs on it, and AI narrators get it wrong by default. Here is why that gap matters.
Read the articleHow to tell whether an AI story really remembers: five tests
Every AI story app says it remembers. Five tests you can run in one evening show what yours keeps, what it drops, and what a fix costs.
Read the articleWhy AI stories ignore your corrections, and what makes a fix hold
You fix a wrong detail, and two turns later the old version is back. Why corrections slip in AI stories, and what it takes for a fix to hold.
Read the articleWhy AI stories forget, and what it takes to make one remember
Long AI stories drift and contradict themselves around chapter ten. The cause is not a small context window, and the fix is not a bigger model.
Read the articleWhat is interactive fiction? A plain guide
Interactive fiction covers parser games, gamebooks, visual novels and AI text adventures. What each one actually is, and how to pick where to start.
Read the articleInteractive fanfiction: replaying a story you already love
Fanfiction asks what if. Interactive fanfiction lets you find out turn by turn, inside the original world. What makes it work, and what usually breaks it.
Read the articleWhy we are training our own models
Nostyss runs on a general-purpose model today: the right way to start, the wrong place to stay. Where the fine-tuning stands, and why owning the model matters.
Read the article