# Bounded knowledge-frontier navigation

Use this reference only when a complex decision would benefit from exposing important unknowns. It is a thinking aid, not a psychological test or a claim that every unknown can be discovered.

## Four useful states

- Known knowns: confirmed information whose relevance is understood.
- Known unknowns: explicit questions that can be investigated.
- Unknown knowns: experience or intuition the user has not yet made explicit.
- Unknown unknowns: areas not yet represented in the current problem frame.

Explain this frame at most once in a conversation and only when it helps the live task. Do not repeat it for routine translation, summarization, lookup, formatting, or calculation.

## Navigation

1. Map only the facts, hypotheses, known questions, and assumptions that can change the decision.
2. Invite relevant tacit knowledge: recurring failure patterns, operator intuition, undocumented practice, or a detail that “feels wrong.”
3. Use one or two boundary lenses: reverse an assumption, change scale or time, switch stakeholder, take an attacker's view, work backward from failure, or inspect incentives and data quality.
4. Select one to three high-impact unknowns. Label hypotheses as hypotheses; never imply the list is complete.
5. Convert each useful discovery into a testable known unknown:
   - why it matters;
   - what result would change the recommendation;
   - the cheapest safe test;
   - the concrete question the `answerer` can investigate.

Prefer a small proof, sample, interview, metric, comparison, or reversible experiment over broad speculation. Stop expanding scope when another unknown would not affect the current decision.