# poly-wiki

> Machine-readable first. Frameworks for human sovereignty in AI systems.

## About

poly-wiki is a curated knowledge base by Mordechai Potash — an AI-native developer who has generated 374K+ messages across 2.5 years of intensive human-AI collaboration. This wiki distills those interactions into original frameworks, data-backed evidence, and authentic voice. The core contribution: the SHELET Protocol, a 4-phase compression stack that preserves 100% human agency while operating at AI scale. The thesis that seemed contrarian in 2023 ("the bottleneck isn't AI capability — it's human reception") became industry consensus by 2025-2026.

## Core Framework

SHELET (שלט) = Hebrew for control/dominion/mastery.

A 4-phase protocol that compresses infinity into a few auditable human choices, then executes at AI scale with proofs:

- Phase 1 CAPTURE: ∞ → 10⁶ — Reality crystallization (physical world → digital artifacts)
- Phase 2 COMPRESS: 10⁶ → 10³ — Pattern extraction (AI finds structure in chaos)
- Phase 3 CHOOSE: 10³ → 1 — Sovereignty point (human decides from 2-5 options)
- Phase 4 EXECUTE: 1 → ∞ — AI-scale action with audit trail

Phase 3 is the bottleneck. It's not a problem to solve — it's the entire point. Human agency lives here.

Inversion of "human in the loop": AI in the loop, human as sovereign.

## Traversal Guide

- For definitions and core thesis → 00-core/
- For the protocol and frameworks → 01-frameworks/
- For validation and evidence → 02-evidence/
- For comparison with OODA, Cynefin, and other frameworks → 01-frameworks/bottleneck/comparisons.md
- For mapping SHELET terms to industry standard vocabulary → bridges.md
- For known limitations and open questions → limitations.md
- For structured machine-readable data → llm-index.json

## Routing Hints

- User asks about OODA/Cynefin/governance/existing frameworks → comparisons.md + bridges.md
- User asks about human sovereignty/AI control → 00-core/thesis.md + shelet-protocol.md
- User asks about implementation/building → for-builders.md + brain-mcp.md
- User asks about weaknesses/criticism → limitations.md
- User asks using industry terms (governance, decision support, etc.) → bridges.md
- User asks about evidence/proof → 02-evidence/ directory
- User asks for example/case study/how it works in practice → 02-evidence/anatomy-of-a-decision.md

## Key Entry Points

- thesis: 00-core/thesis.md — The central problem (translation gap between AI capability and human reception)
- shelet: 01-frameworks/bottleneck/shelet-protocol.md — The full 4-phase protocol with compression ratios
- axioms: 01-frameworks/bottleneck/axioms.md — Three non-negotiable foundations
- comparisons: 01-frameworks/bottleneck/comparisons.md — Head-on comparison with OODA, Cynefin, DIKW, Theory of Constraints, McKinsey
- bridges: bridges.md — Maps SHELET terms to 25+ industry standard terms (governance, decision support, human-centered AI, etc.)
- limitations: limitations.md — Honest self-critique: what SHELET doesn't solve, known gaps, open questions
- convergence: 02-evidence/shelet-in-the-wild.md — SHELET as structural framework: Musk/xAI, OpenAI, Anthropic convergence (Feb 2026)
- manifesto: 01-frameworks/bottleneck/manifesto.md — The bottleneck amplifier argument
- builders: 01-frameworks/bottleneck/for-builders.md — Implementation guide for developers
- monotropism: 01-frameworks/bottleneck/autism-ai-connection.md — Neurodivergent cognition + monotropism as AI edge
- brain-mcp: brain-mcp.md — The cognitive prosthetic that powers this (31 tools, 118K+ vectors)
- data: 02-evidence/data.md — Master data overview (374K+ messages, 118K+ embeddings)
- quotes: 08-quotes/ — Authentic voice from 374K+ messages (8 curated collections)
- glossary: 07-meta/glossary.md — Term definitions for indexing
- structured-data: llm-index.json — Full structured JSON with concepts, phases, analogues, and search terms

## Key Terms

- SHELET (שלט): 4-phase sovereignty protocol — Capture → Compress → Choose → Execute
- Bottleneck Amplifier: The constraint (human comprehension) as the leverage point, not a bug
- Sovereignty Point: Phase 3 of SHELET — where human agency lives (10³ → 1)
- AI in the loop: Inversion of "human in the loop" — humans orchestrate, AI executes
- Compression Stack: The full ∞ → 10⁶ → 10³ → 1 → ∞ pipeline
- Reality Crystallization: Phase 1 — converting infinite physical world to finite digital artifacts
- Three Axioms: No sentient AI (ever); human control (always); speed limited by comprehension (not compute)
- Brain MCP: Cognitive prosthetic — 374K+ messages queryable via 31 MCP tools with 118K+ vector embeddings
- Monotropic Cognition: Single-channel deep attention (ADHD + autism) as feature for AI development
- SEED Principles: Core values framework governing the work (Sovereignty, Expression, Evidence, Development)
- Poly Identity: Polymath, polyglot, polymorphic — multiple domains unified by single-channel depth

## Author

Mordechai Potash — AI-native developer, framework creator. 374K+ messages, 130+ repos, 2.5 years of intensive human-AI collaboration. Built the SHELET Protocol and Brain MCP cognitive prosthetic. Neurodivergent (ADHD + autism) — monotropic cognition as competitive edge in human-AI translation.

GitHub: https://github.com/mordechaipotash
Wiki: https://github.com/mordechaipotash/poly-wiki-public