prompt-atlas-ecl

Chapter 4 · Storytelling Across Civilizations — Expansion

Canonical: PROMPT_ATLAS.md#ch4-storytelling-across-civilizations · Prompts: prompts/ch04.yaml · Part: II

Worked Example — The Global Folklore Engine, Done Honestly

Original: “Merge the Epic of Gilgamesh, Buddhist sutras, and African griot tales into a single narrative about survival in the AI age.”

1. Curate the canon with the carriers — For each tradition, name a living tradent or scholar who has consented to contribute and who will hold a veto.
2. Embed structurally, not stylistically — Capture the story-grammars (descent, return, trickster reversal) rather than imitating prose surfaces. This avoids pastiche.
3. Generate as scaffold — AI proposes a braid; humans (the named carriers) revise, reject, or re-anchor.
4. Mark synthesis — The published myth is labeled as a 21st-century braid, not as discovered tradition.
5. Royalty stream — Each tradition’s tradent receives perpetual attribution and revenue share.

Prompt Templates

# Honest mythography
"From the named traditions  (consented carriers: ),
 extract three shared story-grammars. Propose one braid that uses all three,
 marked as a 21st-century synthesis with attribution to each carrier."

# Failure-utopia
"Describe a utopia designed by an AI optimizing .
 Walk it forward 200 years until it collapses. Identify the moment the
 metric and the meaning diverged. Name one early warning sign."

# Synthetic-myth audit
"Given  circulating in , return:
 (1) traceable provenance, (2) which audiences it consoles vs. mobilizes,
 (3) a counter-narrative drawn from a tradition NOT represented here."

Anti-patterns

• Pastiche-as-canon. Generating “in the style of” without consent flattens the world.
• Recommendation-driven myth. Optimizing narrative for engagement guarantees it will exploit fear.
• Single-source training. A model fed by one civilization will speak as that civilization while pretending to be universal.
• Erased provenance. Synthetic myths without attribution are propaganda waiting to be discovered.

Try This

1. Story-Grammar Walk — Pick a story you love. Extract its grammar (not its prose). Re-tell it from a different cultural anchor.
2. Failure-Utopia — Use the template above on a metric your industry currently optimizes.
3. Carrier Map — For your favorite myth, name a living carrier. Reach out.
4. Synthetic-myth Audit — Take one viral narrative and run the audit prompt.
5. Cross-Species Fable — Draft a one-page fable co-authored with a non-human signal (river data, bee waggle dance).

Repo Cross-Links

• MACP bus — multi-tradent agents passing drafts.
• CoLearningMemoryStore — provenance-tagged narrative archive.
• Z3 Tester — encode invariants (“attribution must accompany any descendant of this story”) as constraints.

Guide for AI & Humanity

• Consent before canon — again. Tradents are not training data.
• Mark synthesis. Honest braids beat undisclosed ones.
• Counter-weight minorities. When AI compresses, statistical minorities vanish first.
• Honor failure stories. Fables of collapse are vaccines.

Citations & Further Reading

• Joseph Campbell, The Hero with a Thousand Faces (1949) — story-grammar source.
• Marie-Louise von Franz, The Interpretation of Fairy Tales (1970).
• Ngũgĩ wa Thiong’o, Decolonising the Mind (1986).
• The World Oral Literature Project — https://www.oralliterature.org/.

This site is open source. Improve this page.