Notes on craft, method, and what we can measure
Where we write up the reasoning behind the tools: the methodologies they lean on, the research they cite, and what happens when we test them — including the tests that go against us.
All posts
5 postsDramatica's four throughlines, explained without the jargon
The most complete and least approachable story theory ever written. Its central claim — that a story models one mind working through a problem — is worth having even if you never touch the software.
We killed our best feature. The judge that killed it was an AI.
Our flagship tool made surgical, word-level edits to fix the most common complaint about AI prose, without rewriting the author's sentence. It lost a blind test in all but one case. Then we looked at who had been doing the judging.
The Chubbuck technique, whole — and why writers should learn it from her, not from us
Most writers meet Ivana Chubbuck's work through one borrowed idea. The full technique is twelve steps, built for actors in a room. What it gives a novelist, and why to go and do the real thing.
Why your story bible gets ignored
You filled in every character and rule, and the AI still gave your bald character hair. Not a bug in your bible — it is what a story bible actually is underneath, and what the interface implies that the model never promised.
Truby's seven steps, and the two everyone skips
The seven-step core of John Truby's structure, what each step is really asking, and why a draft that stalls in the middle has usually failed at step three rather than step five.
Told when there is something to read
The voice-preservation list gets one email when the study has an answer — including if the answer is that our technique did not beat a plain frontier model.