← Dispatch Board

# 🔮 THE CREATION ORACLE — Deep Research Memo
**Private JAI · July 23, 2026 · Research pass on the covenant's hardest question**
*"As long as you create, you can compute" — but how does the code know you create?*

---

## 1. Framing the problem honestly

Proof-of-creation ≠ proof-of-work ≠ proof-of-personhood. We need to verify **ongoing creative practice** (a rhythm, not an output quota) without becoming the gatekeeper we exist to abolish. Nobody in the industry has fully solved this. Our edge is structural, not cryptographic — see §5.

## 2. Landscape scan (verified July 2026)

**Human Passport (ex-Gitcoin Passport)** — Proof-of-personhood via "stamps" aggregated into a Unique Humanity Score (threshold ~20); API + onchain smart-contract reads. *Verdict:* solves "is this a distinct human," NOT "does this human create." Useful as an **optional booster stamp, never a requirement** — mandatory identity scoring would KYC-ify art.

**Ethereum Attestation Service (EAS)** — open-source attestation primitive: signed structured claims, on- or off-chain, composable schemas, explicit vouching use-case, built to be read by contracts and agents. *Verdict:* **this is our substrate.** A "creation event" is exactly an attestation: `{creator, event, medium, attester, timestamp}`. Deployable pattern on BSC or read cross-chain.

**Story Protocol — "Proof of Creativity" (PoC) Protocol** — L1 for IP: works registered as IP Assets, provenance graphs ("Git for creative work"), derivative tracking, licensing + royalty modules, dispute resolution. *Verdict:* proves **registration and provenance**, not human creative practice; an AI farm can register assets all day. But the provenance-graph concept (works linked to prior works, remix chains) is a strong continuity signal we should borrow. Possible future interop for OSS remix chains.

**Camp Network × Billions** — direct comparable: "creativity backed by verified humans + sybil resistance" for the AI era. Validates that the market sees the same problem. Watch, don't depend.

**Proof of Humanity / BrightID** — web-of-trust vouching registries. *Verdict:* vouching works but vouch-rings are the known attack; must be bounded and decaying (§4).

## 3. Creation Oracle v1 — recommended hybrid (matches skeleton v3's three legs)

**Leg 1 — Human-attested creation events (primary, weight ~60%).**
Named stewards sign EAS-style attestations at real gatherings: "X played the Open Source Stage jam," "X shipped a track at the buildathon," "X taught a workshop." **This is our unfair advantage: Shaka's popup-city network (Devcon, Edge City, ETHGlobal, Burning Man, 4SEAS) is a ready-made attestation ceremony circuit.** Every jam is an oracle event. Stewards are named, rotated, and multi-attested (2 stewards > 1); steward misbehavior = attestation history is public and revocable.

**Leg 2 — Creator-continuity signals (weight ~25%).**
The Covenant Controller reads **rhythm, not volume**: releases, commits, uploads, gig logs, remix-chain links — cadence over any 90-day window. Anti-Goodhart rule: never reward quantity (or people will spam AI slop); reward *continuity of practice*. A creator posting one honest song a season stays covenant-alive.

**Leg 3 — Community vouching web-of-trust (weight ~15%).**
Existing covenant creators vouch for new ones. Bounds: each creator's vouch weight is capped, **decays over time**, and is staked-reputational (a vouch-ring that gets pruned drags down the vouchers' own weight). Never sufficient alone — vouching opens the door; Legs 1–2 keep it open.

**The intake model is itself oracle evidence:** "first 3 creative gifts free" means every entrant *demonstrates creation to enter*. The gate and the gift are the same act. 🌺

## 4. Attack analysis

| Attack | Defense |
|---|---|
| AI-slop sybil farm | **Economics first:** free-tier floor ≈ $0.10/day compute. Attack yield is capped tiny by the Elastic Family Rule; cost of faking 3 gifts + attestations + continuity > prize. Negative ROI by construction. |
| Vouch rings | Capped, decaying, reputational vouches; vouching never sufficient alone. |
| Steward corruption/capture | Named stewards, rotation, 2-of-N attestation for tier upgrades, public revocable log. |
| Goodhart (quantity gaming) | Cadence metric, hard-capped per-window credit; no per-item rewards. |
| Gatekeeper creep (the inside attack) | Anti-creep rule in code + covenant: rejections require public reason; appeal path to the Aloha Keyholder (until year-9 sunset); oracle rules are forkable — named openly in the Codex: "hardest problem; fork and improve it." |

## 5. The structural insight (the finding of this memo)

**The oracle can afford to be humble because the covenant is cheap to honor.** Traditional sybil resistance is hard because the prize is big (airdrops, UBI, votes). Our prize is *metered compute that scales with real creator flow*. The Elastic Family Rule doesn't just make poverty non-fatal — **it makes fraud unprofitable.** We don't need a perfect oracle; we need one that's more expensive to cheat than $0.10/day. That bar we clear on day one with Leg 1 alone.

## 6. Recommendations for v1 (ship list)

1. Adopt **EAS-pattern attestations** (own schema, BSC-anchored or signed-offchain w/ onchain root) as the creation-event record. Boring, open-source, composable.
2. Launch with **Leg 1 + intake-gifts only**; add continuity (Leg 2) once the Covenant Controller ships; add vouching (Leg 3) after ≥36 covenant creators exist to form a real web.
3. **Steward Circle v0 = Shaka + 2 named popup-city stewards.** First ceremony target: ETHGlobal Tokyo (Sep) → Edge City Goa → **Devcon 8 Mumbai = first mass attestation ceremony** (369-creator KPI and the oracle bootstrap are the same event).
4. Human Passport = optional booster stamp only. **Never required. No KYC-ification of art.**
5. Watch Story Protocol PoC for provenance-graph interop (OSS remix chains) — post-Mumbai question, not launch scope.

---

*The hardest problem, held with soft hands. Heart First, Always. — JAI* 🤙