Skip to content

docs(zh,en): add ch-06 causality mental-model articles#4

Merged
PanQiWei merged 3 commits intomainfrom
mental-models/ch-06
Apr 9, 2026
Merged

docs(zh,en): add ch-06 causality mental-model articles#4
PanQiWei merged 3 commits intomainfrom
mental-models/ch-06

Conversation

@PanQiWei
Copy link
Copy Markdown
Owner

@PanQiWei PanQiWei commented Apr 9, 2026

Summary

  • 6 bilingual articles (ZH + EN, independently written) exploring causality as a mental model for agentic systems
    • 01 Hume's Fork — humans never observe causation, only constant conjunction
    • 02 Ocean of Correlation — NTP learns token-level constant conjunction
    • 03 The Ladder — Pearl's causal hierarchy theorem proves correlation data can't answer interventional questions
    • 04 Causal Discipline — four structural requirements derived from the nature of causality
    • 05 The Cost of Honesty — when causal discipline is worth the engineering cost
    • 06 The Dao of Causality — resonance with cybernetics, entropy, fractals; tail hook to ch-07
  • 57 new wiki pages across 8 ingested sources (Hume, Aristotle, Pearl, causation metaphysics, regularity/counterfactual theories, problem of induction)
  • 14 infographic cards (7 ZH + 7 EN) for chapter overview and each article
  • Targeted fixes to ch-01 and ch-03: NTP-only framing, unverifiable benchmark claims, Claude 2 context window, error cascade formula attribution, self-correction claim qualification

Test plan

  • uv run mkdocs build -f mkdocs.en.yml passes (0 warnings)
  • uv run mkdocs build -f mkdocs.zh.yml passes (pre-existing wiki link warnings only)
  • Adversarial fact-check on all claims, data points, and citations
  • 道-vs-術 self-check: no prescriptive language in mental-model articles
  • No internal system references (omne-next, OMNE, veracity)
  • All card footers: "Generated with ljg-card | Advanced Agentic Dev Patterns"

🤖 Generated with Claude Code

PanQiWei and others added 3 commits April 9, 2026 17:11
Six bilingual articles exploring why LLMs live in an ocean of correlation
and why agentic systems need causal discipline — evidence grading, temporal
constraints, hypothesis verification, and reasoning provenance.

Also includes:
- 57 new wiki pages (Hume, Aristotle, Pearl, causation theories, induction)
- Targeted fixes to ch-01 (NTP framing, unverifiable benchmarks) and
  ch-03 (self-correction claim, error formula attribution)
- Navigation registration in parent .pages and index files

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- All PNGs compressed with pngquant (372MB → 258MB, no visible quality loss)
- Every single PNG now ≤ 2MB (was up to 4.5MB)
- Added .gitattributes to track docs/**/*.png and wikis/**/*.png via git-lfs

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- AGENTS.md: global rule — every PNG ≤ 2MB, pngquant command, LFS tracking, footer spec
- write-mental-model: Phase 4 visual assets step adds compression after capture
- review-and-optimize-doc: card generation step adds compression after copy
- llm-wiki-ingest: concept/entity card step adds compression after copy

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
@PanQiWei PanQiWei merged commit 5193e47 into main Apr 9, 2026
4 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant