Agency under uncertainty

The Assumption Registry is an in-development method for recording a consequential belief, its status, evidence for and against it, a proposed test, and the decision it may affect.

AI may organize evidence. It may not manufacture evidence.

What is inside

• Assumptions — record a belief, epistemic status, confidence, consequence if false, and review date. A Stale formula field is present; its behavior is unverified.
• Evidence — record an observation, source, strength, direction, directness, and related assumptions.
• Experiments — record a method and a prewritten disconfirming result before interpreting an outcome.
• Decisions — record a choice, date, reversibility, outcome, and related assumptions.

<aside> <img src="https://artemis.danielwahnich.dev/assets/mark/artemis-mark-paper.png" alt="https://artemis.danielwahnich.dev/assets/mark/artemis-mark-paper.png" width="40px" /> Method mode: no paid-AI step is required by the documented workflow. Sorting and current view configuration remain unverified. Example-row counts are declared build evidence, not live-enumerated observations.

</aside>

Current in-development structure

Assumptions

Evidence

Experiments

Decisions

Using it

The documented loop moves from a desire to assumptions, evidence, a proposed test, human judgment, and a recorded decision. It does not determine the decision.

A first pass