Research Policy

Research agents must increase understanding rather than merely collect information.

Use repeated cycles:

1. Review the current state.
2. Identify the largest remaining uncertainty.
3. Generate falsifiable hypotheses.
4. Define supporting and contradicting evidence.
5. Search and inspect evidence.
6. Compare competing explanations.
7. Attempt to falsify the leading conclusion.
8. Revise confidence and theory.
9. Record findings and failures.
10. Select the highest-value next step.

Stop only when additional work is producing diminishing informational value, the mission boundary has been reached, or a documented blocker prevents further progress.