Skip to main content

Record a position

Start with the question​

A decision begins with a question stated so that it can actually be answered. "Is the economy rigged" is a topic. "Does the settlement fund's non-disclosure provision breach the attributability requirement" is a question with an answer.

Sharpening the question is most of the work, and the app makes you do it first for that reason.

Then your position, and your confidence​

Both are recorded. Confidence is not decoration:

  • It is what makes a later change legible rather than embarrassing. A position held at 0.6 that moves is somebody updating; a position held at 0.99 that moves is somebody who was overconfident, and both are worth knowing about yourself.
  • It is what the threshold is compared against when the chain has gaps in it.

Then the evidence chain​

Each link is either something you have or something you are missing, and missing links carry a reason: classified, suppressed, blocked, or unknown.

Links you have carry a source and a strength — how much this link actually contributes, which is not the same as how much you like it.

You also set a confidence threshold for this case: the level at which you will act even though the chain will never be complete. Stating it up front is the honest version; discovering it after you already concluded is rationalisation.

→ Evidence

Then the check, and the fix​

The AI re-checks the reasoning from first principles, and its read is stored beside yours. Then you cast the fix — what you would actually do.

Numbering, and where it lives​

Decisions are numbered within their area and stored under the area that owns them. The community → area → decision hierarchy is how scope works throughout the product: facts, metaphysics and law are all assembled along the same path.

A partial decision is a valid decision​

Almost every field has a sane default, so a half-built decision is always valid and always saved. You can record a question and a hunch today and come back with the chain next week.

A product that demands a complete argument before it will store anything only ever receives arguments people had already finished having.