## Overview
Once human expertise is captured into AI-consumable formats — skills, SOPs, open-source projects, context files — it can never be un-captured. This is a one-way ratchet: every piece that enters the AI ecosystem makes AI permanently more capable, while humans still need 20-30 years to develop deep expertise in one domain.
## Core Framework
Three properties drive it:
1. **Irreversibility**: published skills, open-source code, and documented SOPs cannot be un-published.
2. **Instant absorption**: AI ingests captured expertise immediately and duplicates it infinitely across instances.
3. **Compounding acceleration**: each captured piece makes AI better at capturing the next, a self-reinforcing cycle.
The asymmetry: a human learns one domain over decades, forgetting along the way. AI absorbs all captured expertise instantly, never forgets, and scales infinitely.
## Cross-Domain Applications
- **Enterprise knowledge management**: capturing institutional knowledge into AI-readable formats compounds advantage over firms that don't.
- **Career strategy**: the window for human expertise as a moat is closing — the question shifts from "what do I know?" to "what can I do that can't be captured?"
## Critical Analysis
The metaphor assumes captured expertise is high-quality and context-appropriate; much documented knowledge is outdated or doesn't transfer. The ratchet has friction, but the direction is one-way.
## Related Concepts
- [[Articulation Gap]] — the bottleneck that slows the ratchet
- [[Skill Floor Ceiling Compression]] — the ratchet's effect on skill distributions
- [[AI Capability Denial Acceleration Pattern]] — denying the ratchet makes it worse
*Source: [[Exactly Why and How AI Will Replace Knowledge Work]] — Daniel Miessler (2026-03-19) — https://danielmiessler.com/blog/exactly-why-and-how-ai-will-replace-knowledge-work*