field notes SCRIBE

a place to put a question

A system that helps write its own documentation produces confident prose about things it does not know. The failure is not a lie. It is a guess that is well formed and sits in a document that is otherwise correct, which is the hardest kind to notice, because no sentence anywhere reads as uncertain.

Asking is the obvious remedy and it barely works. Asking needs a moment, and the moment occurs inside a task that is trying to finish; a question raised in passing is answered if the person was listening and lost if they were not. Meanwhile the guess stays in the document, looking exactly like the parts that are true.

So questions became artifacts. Each was given an identifier and a place in the tree, and a rule was adopted that the pending list is surfaced at the start and at the end of every working session until it is empty. Not a message, which is ephemeral to the person who did not read it that day — a list with stable identity, so that "still open" is visible without anyone having to remember it was ever asked.

This changes what a guess costs. A placeholder now has somewhere to live that is legible as a placeholder, which removes the pressure to fill the gap with the first plausible answer.

The cost is friction at exactly the two moments when someone is trying to get started or trying to stop, which is when a reminder gets skipped first. What it proves is that the state of not knowing was given a durable object rather than a conversation.

drawn from: single commit

Every note carries a codename and one idea. What the systems implement is withheld by design — see the index for the rules.