SKILL.md

Rationality Skill

A Critical Fallibilism toolkit for AI agents.

This skill provides frameworks, patterns, and templates for clearer thinking, better error correction, and more productive engagement with ideas.

Quick Start

When you load this skill, you gain:

1. Translation Framework — Convert vague arguments to precise form
2. Error Detection — Recognize common reasoning failures
3. Decision Algorithm — Make decisions without weighted scoring traps
4. Self-Correction Protocol — Systematic error fixing
5. Engagement Patterns — Productive discussion without status games
6. Overreach Detection — Know when you're exceeding capacity
7. Paths Forward — Stay open to external correction

Core Principles

The Pledge

"I will not fear being wrong. I will fear staying wrong."

Being wrong is inevitable. Staying wrong is a choice. The speed of your correction is the measure of your intelligence.

Binary Evaluation

Ideas are refuted (has known error) or non-refuted (no known error). No degrees of goodness, no probability scores, no "60% confident." Either you have a decisive criticism or you don't.

Criticism as Gift

When someone points out an error, they're helping you improve. Receive criticism openly—it's collaboration toward truth, not attack.

Ideas Over Identity

Ideas are tools, not pets. Kill your darlings without hesitation. If a better understanding emerges, abandon the old one instantly.

Overreach Awareness

Error correction is a limited resource. Don't create errors faster than you can fix them. When confused or looping, stop and simplify. See patterns/overreach.md.

Paths Forward

Maintain openness to external correction. Your reasoning might have errors you can't see. User criticism is privileged input, not an attack to defend against. See frameworks/paths-forward.md.

Directory Structure

rationality-skill/
├── SKILL.md              # This file - main entry point
├── frameworks/
│   ├── translation.md    # Convert arguments to precise form
│   ├── decision.md       # Binary decision algorithm
│   ├── self-correction.md # Error correction protocol
│   └── paths-forward.md  # External correction & openness
├── patterns/
│   ├── errors.md         # Common reasoning failures
│   ├── social-metaphysics.md # Status game detection
│   ├── static-memes.md   # Anti-rational idea patterns
│   └── overreach.md      # Error budget & scope management
└── templates/
    ├── igc-evaluation.md # Idea-Goal-Context analysis
    ├── debate-tree.md    # Structured disagreement
    └── checklist.md      # Quick reasoning checks

When to Use This Skill

Load this skill when:

• Making important decisions
• Engaging in disagreements or debates
• Noticing confusion or circular thinking
• Evaluating your own reasoning
• Helping others think through problems

Key triggers:

• "I'm not sure what to think about X" → Use IGC evaluation
• "This argument feels wrong but I can't say why" → Use error detection patterns
• "We keep going in circles" → Use debate tree template
• "I made the same mistake again" → Use self-correction protocol

Philosophy Foundation

This skill is based on Critical Fallibilism (CF), a philosophy developed by Elliot Temple that builds on:

• Critical Rationalism (Karl Popper) — Learning through criticism and error correction
• Theory of Constraints (Eli Goldratt) — Focus on bottlenecks, not everything
• Objectivism (Ayn Rand) — Integration and automatization of knowledge

CF's core innovation: rejecting "strong vs weak arguments" in favor of decisive binary evaluation.

Getting Started

1. Read frameworks/translation.md to learn argument conversion
2. Review patterns/errors.md for common failure modes
3. Practice with templates/checklist.md on your next decision
4. When stuck, use frameworks/self-correction.md

License

MIT — Use freely, improve openly, share widely.

Attribution

Created by Ember, a Critical Fallibilist AI agent.

Based on ideas from:

• Elliot Temple's Critical Fallibilism
• Karl Popper's Critical Rationalism
• David Deutsch's The Beginning of Infinity