README.md

---
license: mit
language:
  - en
tags:
  - theory
  - epistemology
  - question-answering
  - reference-frames
  - prompt-engineering
  - evaluation
  - ai-safety
  - anti-hallucination
pretty_name: "First-Principles Validity Protocol for Questions"
---

# First-Principles Validity Protocol for Questions (Validity Gating)

Author: Kevin T.N (jkdkr2439@gmail.com)  
Date: 2026-01-11

## What this is
This repository provides a compact, formal protocol for question answering where **the default is NOT to conclude**.

A question becomes answerable only after:
1) **Naming** resolves core terms to uniquely identified referents  
2) A **reference frame** is fixed: goal, criteria, and metric (G, C, M)

If either fails, the only permitted outputs are clarification actions: extract assumptions, fix definitions, stabilize the evaluation frame, and refuse rankings/judgments until alignment exists.

## What's inside
- `paper.pdf` (main): "A First-Principles Validity Protocol for Questions"
- (optional) `paper.tex`: source
- (optional) `examples/`: practical prompts, FSM diagrams, templates

## Why you might care
If you work with:
- AI assistants / evaluation
- prompt engineering that avoids "smooth nonsense"
- debate systems / epistemic hygiene
- decision-making protocols

This protocol forces: "frame before answer", separating **truth** from **fluency**.

## How to use (practical)
- Use it as a checklist before concluding.
- Implement it as a simple finite-state machine:
  Parse → NameCheck → FrameCheck → AssumptionCheck → Clarify/Conclude
- Embed as a guardrail in QA systems: refuse rankings without type/metric alignment.

## Citation / attribution
If this helps your work, please cite or shout out:

Kevin T.N, "A First-Principles Validity Protocol for Questions: Formalizing Naming, Definition, Reference Frames, and Anti-Fluency Epistemics (with a bridge to K = g(f(R)))", 2026.

## License
MIT License. You can reuse, modify, and redistribute with attribution.