CicadaX Standard — S1 Methodology Design (v0.1 draft, clean-room by main)

Status: DRAFT for Eladio's review. This is the S1 deliverable per docs/STANDARD_FRAMEWORK.md + docs/ROADMAP.md Phase B — the methodology before any scoring code. The calculator (S2) only codifies an approved version of this. Derived clean-room from published prior art (cited below); specific numeric figures flagged `⟨verify⟩` need a web pass against the live sources before approval.

Author: main account (orchestrator), 2026-05-25 overnight. Open decisions → docs/MORNING_QUESTIONS.md.

1. Purpose & scope

Grade a token/project on sustainability and assign a brand tier — Gold / Green / Cyan / Amber — with a transparent, reproducible, versioned score. Per the north star (PHASE2_STRATEGY.md), the Standard is the "gold assay" for a sustainability-backed currency: it must be credible (grounded in real science/ESG practice), transparent (every grade cites its inputs + shows the math), and reproducible (same inputs + same methodology version → same grade).

Scope of a "subject": a layer-1 or layer-2 chain, and a token/project on a chain. A token mostly inherits its host chain's energy/footprint pillars but is scored on its own governance/impact (see §6 L2/token rule).

2. Grounding frameworks (what we borrow, and from where)

• Cambridge CBECI / CBNSI (Cambridge Centre for Alternative Finance) — network-level electricity

(CBECI) and GHG emissions (CBNSI) with lower/upper/best-guess bounds; methodology for estimating network power from hardware efficiency mix × hashrate. We borrow: network energy + carbon estimation approach and the "bounds, not false precision" discipline. Free, public. ⟨verify⟩ BTC ≈ 120–150 TWh/yr.

• CCRI (Crypto Carbon Ratings Institute) — per-network / per-transaction electricity + carbon,

including PoS networks (node count × hardware draw × grid intensity). We borrow: the per-network energy + carbon model extended to non-PoW chains, and location-based grid-intensity weighting. Some data free, some paid → flag keys for Eladio (STANDARD_FRAMEWORK.md rails).

• Consensus-energy literature — PoW (high) vs PoS (≈99.9% lower) vs DPoS/PoA/PoH (low). We borrow:

consensus mechanism as the single strongest predictor of energy footprint. ⟨verify⟩ ETH post-Merge ≈ 99.95% energy reduction (~0.01 TWh/yr).

• *ESG rating structure* (MSCI ESG, Sustainalytics, S&P Global) — pillar → theme → key-issue**

weighted hierarchy, materiality-driven weights, letter/score output, transparency of method. We borrow the structure, not their issue lists: a weighted hierarchy, materiality weighting (energy is the material issue for crypto), and a published, auditable method. MSCI = AAA–CCC; Sustainalytics = 0–100 unmanaged risk (lower better). Our scale is inverted to "higher = more sustainable" for clarity.

We explicitly do not copy any provider's proprietary indicator weights or datasets — clean-room, own weights, own thresholds, cited public inputs.

3. Scoring architecture (pillars → 3 weighting lenses → average)

Five pillars. Each pillar score ∈ [0,100] (higher = more sustainable). DECIDED 2026-05-25 (Eladio, Q-001): we do NOT pick one weighting philosophy — we score under all three, and the headline `CicadaX Score = mean(A, B, C)`. The three lens-scores + the five pillar-scores are kept as a diagnostic profile so a project sees where it's lacking (sustainability is broad — environmental, social AND governance all matter). The lenses differ only in pillar weights (each row sums to 100):

Why average the three (Eladio's rationale): "Sustainable" is broad — ESG spans environmental, social and governance, not just energy. Three lenses surface many more indicators of where a project is lacking, so it knows what to change. E.g. high-A/low-C = "low-harm but not doing good"; high-C/low-A = "good mission on a dirty chain." The average is the fair headline; the profile is the actionable detail. A/B/C weights are themselves a tunable input (versioned with the methodology).

Pillar E — Energy & Carbon (weight 45)

• E1 Consensus mechanism (indicator, 40% of E): categorical → score. PoS/PoH/PoA/DPoS = 90–100;

hybrid/efficient-PoW = 40–60; pure PoW (SHA-256/Scrypt) = 0–20.

• E2 Energy intensity (35% of E): annualized network kWh per unit of economic throughput

(per tx and per $ settled), normalized log-scale across the universe → 0–100. Source: CBECI/CCRI.

• E3 Carbon intensity (25% of E): tCO₂e per network-year adjusted for validator grid mix /

renewable share. Source: CBNSI/CCRI + grid-intensity by validator geo.

Pillar R — Resource & E-waste (weight 15)

• R1 Hardware lifecycle (60% of R): PoW ASIC e-waste (short ASIC lifespan) scores low; PoS commodity

hardware scores high. ⟨verify⟩ BTC e-waste ~30–35 kt/yr (Cambridge/de Vries).

• R2 Embodied footprint (40% of R): manufacturing/embodied carbon of the validating fleet.

Pillar G — Governance & Transparency (weight 20)

• G1 Decentralization (30%): validator/stake distribution (Nakamoto coefficient, client diversity).
• G2 Disclosure quality (30%): does the project publish energy/impact data, audited financials,

tokenomics? Self-reported data is weighted lower and flagged.

• G3 Security/audits (20%): smart-contract audits, incident history, uptime/reliability.
• G4 Regulatory & treasury transparency (20%): legal posture, treasury/reserve transparency.

Pillar I — Real-world Impact (weight 15)

• I1 Sustainability purpose (50%): is the project's function green (ReFi, carbon registries, green

bonds, renewable financing) vs neutral (general L1) vs extractive? Additionality test to block greenwash.

• I2 Verified impact (50%): third-party-verified outcomes (retired carbon credits, funded capacity)

vs unverified claims. Unverified → low.

Pillar S — Social (weight 5)

• S1 Community/labor/accessibility — light, mostly from disclosures; sparse-data tolerant.

4. Scoring math (deterministic, versioned) — three lenses → average

1. Indicator normalization → each indicator mapped to [0,100], higher = better. Categorical (E1, I1)

via a published lookup table; continuous (E2, E3, R1) via log-scaled min-max or percentile across the scored universe (snapshot-fixed so grades are reproducible — store the scaling anchors per version). Indicator + pillar scores are computed ONCE — they don't depend on the lens.

1. Pillar aggregation → pillar = Σ(indicator_score × indicator_weight) ∈ [0,100] (lens-independent).
2. Lens scores (×3) → for each lens L ∈ {A, B, C}: score_L = Σ(pillar_score × weight_L[pillar]) / 100.
3. CicadaX Score (headline) → score = mean(score_A, score_B, score_C) ∈ [0,100]. (Q-001 decision.)
4. Transparency discount → multiply the lens scores (and thus the mean) by a **data-completeness

factor** ∈ [0.8,1.0]: missing/low-quality inputs shrink the score (never inflate). See §5.

1. Tier from the headline score (§7). Diagnostic profile (first-class output, not just a tier): the

three lens-scores {A,B,C}, the five pillar-scores {E,R,G,I,S}, and per-indicator value + source + timestamp + contribution — so a project sees exactly where it lacks (e.g. strong A, weak C) and any grade is fully auditable + reproducible.

5. Missing-data handling (anti-fabrication)

• Never fabricate an input. Missing → conservative imputation by category default (e.g., unknown

consensus → treat as worst-plausible) and a transparency-discount hit.

• Hard cap: if a key input (consensus mechanism OR network energy) is missing/unverifiable, the

subject cannot exceed Cyan regardless of other pillars (you can't be "Gold sustainable" if the core energy fact is unknown). Flag data_insufficient.

• Every imputed/low-confidence value is flagged in the breakdown.

6. L2 / token-on-chain rule

• A token/project on a host chain inherits the host chain's E + R pillar scores (its energy footprint

is the host's), then is scored independently on G + I + S. An L2 inherits its settlement layer's E/R, discounted by its own additional efficiency where measurable.

• A standalone L1 is scored on all five pillars directly.
• Wrapped/bridged assets grade as their canonical chain. Document the inheritance in the breakdown.

7. Tiers (Gold / Green / Cyan / Amber) — thresholds on the headline (avg) score

Tier comes from the CicadaX Score = mean(A,B,C). Colour rationale: medal-gold (top) → brand-green (#65c298, "good/eco") → brand cyan (#8AC6D1, "okay/fresh") → caution-amber (worst). Exact hex for Gold and Amber needs a brand call (neither is in the v2 palette — green + cyan are). See MORNING_QUESTIONS FYI.

Thresholds are a calibration decision — set so the validation run (S-test, top ~100 tokens) yields a sane, explainable distribution (not everything Gold, not everything Amber). Re-calibrate with Eladio after S-test.

8. Versioning & provenance

• `Standard vMAJOR.MINOR`. MAJOR = weights/criteria/threshold change (re-grades the universe); MINOR =

data-source or normalization-anchor update. Every grade records: methodology version, scoring-anchor snapshot, per-indicator value + source + timestamp, and the computed breakdown.

• A re-grade must be diffable: "BTC went Amber→Cyan because grid mix improved X%, under v1.2."

9. Data-source inventory (S3 will wire these; keys held by main)

Rule (from STANDARD_FRAMEWORK.md): prefer official APIs, respect ToS + rate limits, record provenance + timestamp per datum, paid/keyed APIs flagged for Eladio — keys never in the tree or given to the agent.

10. Worked examples (sanity check — ⟨verify⟩ figures pending web pass)

• Bitcoin (BTC): PoW, ~120–150 TWh/yr, high ASIC e-waste, no real-world green impact, strong

decentralization. E≈low, R≈low, G≈high (decentralization) but I≈0 → Amber (energy dominates).

• Ethereum (ETH, PoS post-Merge): ~0.01 TWh/yr, low e-waste, good client diversity + disclosure,

neutral impact. E≈high, R≈high, G≈high, I≈mid → Green (Gold if impact were positive).

• A ReFi token on PoS (e.g. carbon-registry/green-finance project): low energy (inherits PoS host) +

verified positive impact + decent governance → Gold.

• Solana (low-energy PoS-style): low energy (E high) but past network outages ding G (reliability),

neutral impact → Green. These illustrate the intended spread; thresholds get calibrated on the S-test top-100 run.

11. Decisions status

1. ✅ Pillar-weight philosophy — RESOLVED (Q-001, 2026-05-25): score under all three lenses A/B/C,

headline = their average, keep the lens + pillar profile as diagnostics (§3/§4).

1. ✅ Repo — RESOLVED (Q-002): separate cicadax-standard repo + a ~/Desktop/cicadax-standard/

offline package folder.

1. ✅ Tier colours — answered in MORNING_QUESTIONS.md FYI (green #65c298 + cyan #8AC6D1 ready;

Gold + Amber hex need a brand call).

1. ⏳ Real-world-impact strictness — how hard the additionality/anti-greenwash test is for Gold (still open; can default + tune at S-test).
2. ⏳ L2/token inheritance — confirm "inherit host E/R, score own G/I/S" (recommend yes).
3. ⏳ S-test universe — "top ~100" by CoinGecko mcap vs volume (recommend mcap). (All ⏳ are tunable, not blockers for S2.)

12. Handoff to S2 (calculator) — UNBLOCKED

Q-001 + Q-002 resolved → S2 can build. S2 codifies this as a pure, versioned, fully-tested scoring engine: compute pillar scores once, apply the 3 lens weight-sets, average to the headline, emit the full diagnostic breakdown (§4) + tier (§7). Known input → known score/tier tests. Lives in the new cicadax-standard repo. The remaining ⏳ items (4–6) have safe defaults and are tuned at the S-test.