Methodology & Scoring Transparency
This document explains the current (beta) scoring logic and outlines planned factors, their rationale, and anticipated safeguards. It will evolve as we incorporate feedback.
Beta Status: At present, score = pledge verification only. Weighting table below represents intended future distribution—not active yet.
Current Active Factor
- Pledge Verification (100%): Successful signing of the ethical alignment pledge with a supported wallet. Stored hash enables later verification without exposing raw pledge text publicly.
Planned Factors (Indicative Weights)
| Factor | Planned Weight | Intent | Risks / Mitigations |
|---|---|---|---|
| Pledge Verification | 20% | Baseline ethical declaration | Low signal strength → Down‑weights over time |
| Behavior Reputation | 40% | Consistent adherence signals | Gaming via sybil → Rate limits & anomaly detection |
| Interaction History | 20% | Quality / civility of AI interactions | Privacy concerns → Opt‑in, local preprocessing |
| Peer Endorsements | 20% | Web-of-trust amplification | Collusion → Weighted by endorser credibility |
Data Handling Principles
- Minimize: Collect only what’s necessary for a factor.
- Local-first: Preference for client-side processing before hashing / attesting.
- Revocable: Future revocation & decay to prevent permanent penalties.
- Explainable: Each factor will have a public rationale & scoring formula.
Known Limitations
- No decentralized attestation layer yet.
- Single-chain (Solana devnet) prototype: migration risk.
- No formal audit; experimental codebase.
- Potential early adopter bias—mitigated by future reweighting & provenance flag separation.
Versioning
A version tag will be attached to each credential once multi-factor scoring launches. Early badges retain a provenance=true attribute but not preferential scoring weight.
Minting Stats: 42 / 500 planned founding badges minted.