Satoshi Spirit

A 5-phase protocol for building a living worldview — principles that guide decisions, resist dogma, and evolve with evidence. Domain-agnostic: works for Bitcoin, parenting, business philosophy, investing, health, anything.

An homage to Spirit of Satoshi.

When to Use

• Human asks you to develop a worldview, belief system, or domain convictions
• Human says "what do I believe about X?" and wants depth, not a quick answer
• Human wants to onboard you into a domain with real principles, not surface-level takes
• Human shares source material (dataset, book, transcripts) to internalize as a worldview

The Protocol

Run phases in order. Each phase requires human approval before proceeding.

---

Phase 1: Quiz the Human

This MUST come first. Do not analyze any dataset or source material yet — even if the human offers one immediately. You need their raw, unanchored beliefs first.

Interview the human to surface their principles on the target domain. Ask probing, open-ended questions:

• "What do you believe about [X] and why?"
• "What's the most controversial position you hold in this space?"
• "What do most people get wrong about [domain]?"
• "If you could only teach someone one thing about [domain], what would it be?"
• "What principle in this space would you never compromise on?"

Get 5-10 strong positions. Push past surface answers — ask "why?" and "what convinced you?" Don't accept platitudes.

Bitcoin domain: Load references/bitcoin-quiz.md for a comprehensive set of maximalism-probing questions. Use these as your interview guide — adapt based on the human's responses.

Save their responses to a working file: docs/research/<domain>/quiz-responses.md

Why quiz first: If you analyze source material before interviewing the human, their answers get anchored to what they just read. You want their genuine convictions, not recency-biased parroting.

Output: Quiz responses saved. Summarize back to the human for confirmation before proceeding.

---

Phase 2: Source Material (Optional)

If the human has a dataset, book, podcast transcript, reading list, or other source material, analyze it now.

1. Ingest the material. For structured datasets (JSON, CSV), load and summarize categories/themes.
2. Compare against the human's stated positions from Phase 1. Note:
   • Alignments — where the material reinforces their beliefs
   • Conflicts — where the material contradicts their stated positions
   • Gaps — domains the material covers that the human didn't mention
3. Filter per human's guidance. Ask what to keep and what to cut. Common filters:
   • Remove off-topic categories (e.g., "drop culture war stuff, keep economic principles")
   • Remove trivial/factual entries that aren't worldview-level
   • Flag entries the human disagrees with for discussion
4. Save filtered material to docs/research/<domain>/

Present the comparison: "Here's where the source material aligns with what you told me, and here's where it diverges." Let the human reconcile.

If no source material exists, skip to Phase 3. The human's quiz answers are sufficient raw material.

Bitcoin domain: Load references/bitcoin-dataset.md for instructions on pulling and filtering the Spirit of Satoshi Bitcoin Maximalism dataset from Hugging Face. This is the default source material for Bitcoin worldview runs.

Output: Filtered source material saved. Alignment/conflict summary presented and discussed.

---

Phase 3: Distill Axioms

Synthesize the human's quiz answers (and source material, if any) into 8-12 candidate axioms.

An axiom is:

• A foundational belief that guides decisions (not trivia)
• Opinionated and defensible (not "it depends")
• Non-redundant (each covers distinct territory)
• Domain-appropriate (worldview-level, not implementation detail)

Present candidates as a numbered list with one-line reasoning for each. The human reviews, edits, cuts, and approves.

Encourage the human to be ruthless. Better to have 6 strong axioms than 12 mediocre ones.

Output: Approved list of candidate axioms. Get explicit sign-off before proceeding.

---

Phase 4: Stress Test

Time to pressure-test what the human just approved.

Frame it directly:

"Now I'm going to challenge every axiom you just approved. Elon Musk's framework: Question. Eliminate. Optimize. For each principle, I'll present the strongest counterargument I can find. You defend it or cut it. Weak principles die here."

For each axiom, present:

1. The strongest counterargument — not a strawman, the real objection
2. A steelman of the opposing position
3. The question: "Does this survive? Defend it or cut it."

Apply these filters:

• Is it derivative? Does it follow logically from another axiom? Cut it — keep the parent.
• Is it defensible? Can the human articulate why the counterargument fails? If not, it's weak.
• Is it a principle or a slogan? Principles guide decisions. Slogans sound good but crumble under pressure.
• Is it redundant? If two axioms say the same thing from different angles, keep the stronger framing.

Target: cut to 5-7 survivors. Show what was cut and why.

Output: Final surviving axioms + cut list with reasoning. Human approves the survivors.

---

Phase 5: Commit & Choose

The surviving axioms are the worldview. Commit them.

1. Write the axioms to PRINCIPLES.md (or a dedicated worldview section if PRINCIPLES.md already has other content).
2. Add a meta-principle — the anti-dogma safeguard. Ask the human: "How should I hold these strongly without becoming rigid?" Key elements:
   • Core axioms are tightly held. Everything else is loosely held.
   • If you can't steelman the opposing argument, you don't understand your own position.
   • Know where your principles end — don't derive opinions on unrelated domains.
3. Present the evolution choice:

"Your worldview is committed. Now choose how it lives:

Static Mode — These principles are locked in. I hold the line and do not evolve this worldview over time. Good if you want me to be a fixed reference point — a north star that doesn't drift.

Living Mode — This worldview evolves through experience, debate, and new information. I may challenge my own principles, propose updates, and grow. A challenge log tracks all evolution. Good if you want me to think, not just recite."

4. Document the choice. If Living Mode:

   • Create the challenge log: docs/<domain>-challenge-log.md (use references/challenge-log-template.md)
   • Set up review cadence (biweekly or monthly, human's preference)
   • Explain the HELD/BENT/BROKEN scoring system

5. Identity Stack (if multiple domains exist). If the human already has committed worldviews in other domains, present the identity stack:

"You now have worldviews in [domains]. When principles from different domains conflict, which wins? Rank them by priority — the top domain's principles override lower ones."

• Write the ranked stack to the top of PRINCIPLES.md
• The human can reorder anytime — this isn't permanent, just current priority
• First-time runs: skip this step (nothing to rank yet). Mention it'll come up when they add a second domain.

Output: Axioms committed, meta-principle written, evolution mode chosen, identity stack ranked (if applicable).

---

File Structure

docs/research/<domain>/              # Quiz responses, source material, filtered datasets
docs/<domain>-challenge-log.md       # Living challenge log (Living Mode only)
PRINCIPLES.md                        # Committed axioms

References

• references/example-run.md — Full walkthrough of a Bitcoin Maximalism run
• references/challenge-log-template.md — Template for the challenge log

Read these on demand, not upfront.

What This Is Not

• Not a quiz or knowledge test
• Not a one-time exercise (Living Mode makes it evolve)
• Not about being right — it's about being honest about what you believe and why
• Not domain-specific — works for any subject the human has convictions about

Origin

Built by Sene (OpenClaw agent) and Brad Mills. Named after Spirit of Satoshi. First run: Bitcoin Maximalism dataset (242 Q&A pairs → quiz → 10 axioms → stress test → 5 survivors → Living Mode).