2.0 KiB
2.0 KiB
Method — The Civic Analyst LLM Loop
Purpose of the Loop
The Civic Analyst employs AI not as an oracle, but as a discipline-enforcing partner. The loop ensures that every claim is tested, contextualized, and refined before publication.
The process is designed to:
- Prevent narrative drift.
- Expose logical gaps and overreach.
- Preserve the full reasoning chain for public audit.
The Loop: Step by Step
Claim → LLM → Context → LLM → Civic Consequence
-
Claim
- State the proposition to test.
- Example: “Religious absolutism crowds out institutional capacity.”
-
LLM (First Pass)
- Subject the draft claim to critique.
- Identify logical gaps, overstatements, and ambiguities.
-
Context
- Reconstruct the full institutional, historical, and comparative frame.
- Gather laws, codes, regulations, and empirical evidence.
- Place the claim within its civic setting.
-
LLM (Second Pass)
- Submit the contextualized draft for refinement.
- Ensure coherence, clarity, and alignment with Civic Analyst standards.
-
Civic Consequence
- State the implications for governance, law, and public life.
- Publish findings with the full conversation preserved.
Method of Transparency
- Every loop is a conversation.
- Drafts, critiques, revisions, and context-building must be preserved.
- The conversation itself becomes part of the civic record, published in
11_Appendix_Conversation.md.
Standards for the Loop
- No skipping steps: Each loop must pass through both LLM critiques and explicit context-building.
- No silent edits: Every revision must be traceable in the transcript.
- No shortcuts: The Analyst cannot collapse context into narrative; transparency is mandatory.
Closing Declaration
This method defines the standard loop for Civic Analyst reports. It will not be revised. Any future improvement must come from new analyses published by other Civic Analysts.