Why On-Chain Reputation Systems Are Incomplete Without Scoring Rules

Reputation is a public good that suffers from a fundamental data provision problem. Protocols like Ethereum Attestation Service (EAS) and Gitcoin Passport create verifiable records, but they rely on altruism or side incentives for data submission.

Without scoring rules, reputation systems are economically incomplete. A user's on-chain history is meaningless if no one is paid to analyze and score it accurately. This creates a reputation mirage—data exists, but reliable interpretation does not.

Compare Proof-of-Stake to Reputation. Validators are slashed for lying; reputation oracles face no direct penalty for bad scores. Systems like Chainlink and Pyth solve this for price data via cryptoeconomic security, but social data lacks this design.

Evidence: The Sybil resistance problem in airdrops and governance (e.g., Optimism's Citizen House) persists because scoring user legitimacy is a costly, unrewarded task. Truthful reputation aggregation requires a dedicated incentive layer.

Reputation is a prediction market. It quantifies the expected future behavior of an agent. Current systems like Ethereum Attestation Service (EAS) or Gitcoin Passport only collect raw data, creating a static ledger of claims without a mechanism to assess their predictive power.

Scoring rules are the missing engine. A proper scoring rule, like a strictly proper scoring rule, financially penalizes inaccurate predictions. This forces reputation oracles like UMA or Pyth to be honest, transforming subjective signals into a credible consensus on agent quality.

Without scoring, data is noise. A wallet's transaction history or attestation count is a low-signal feature. Scoring rules create a market for truth where staked capital backs predictions, as seen in prediction markets like Polymarket, making reputation a dynamic, high-stakes asset.

Evidence: The $200M+ in fraud from the Mango Markets exploit demonstrated that social reputation without a cryptoeconomic scoring mechanism is worthless. A scoring rule would have priced the attacker's on-chain behavior as high-risk long before the exploit.

On-chain reputation is manipulable data without a mechanism to penalize lies. Systems like EigenLayer or Chainlink oracles rely on subjective slashing, which is slow and politically fraught. A proper scoring rule mathematically guarantees that a reporter's expected reward is maximized only by submitting their true, private belief.

Proper scoring rules are strictly proper. This means a reporter's optimal strategy is honesty, not gaming the consensus. This contrasts with simple averaging or majority voting used in many DAOs, which are vulnerable to Sybil attacks and lazy copying. The math enforces incentive-compatible truth.

The proof is in the payoff function. For a continuous outcome, the Brier score or logarithmic score calculates a penalty based on the divergence between a prediction and reality. A protocol like UMA's Optimistic Oracle uses this to settle disputes; reporters who lie lose their bond. The score is the objective judge.

Evidence: In prediction markets like Polymarket or Augur, scoring rules (via automated market makers) ensure that the market price reflects the true probability of an event. Without this, the reported data is just unverified gossip, which is the fatal flaw of most social reputation systems today.

Scoring rules are the execution layer for reputation. A wallet's history is raw data; the scoring rule is the algorithm that transforms it into a usable signal for protocols like Aave's GHO or Uniswap's governance.

Without rules, reputation is non-composable. A Soulbound Token (SBT) from Gitcoin is a static badge. A scoring rule makes it a dynamic input for a sybil-resistant airdrop or a collateral-free loan on a credit protocol.

The alternative is centralized curation. Projects like Goldfinch manually underwrite borrowers. Scoring rules automate this, creating a permissionless trust primitive that scales beyond human review.

Evidence: MakerDAO's Endgame plan explicitly incorporates decentralized identity and reputation as core infrastructure, signaling that raw on-chain data alone is insufficient for systemic risk assessment.

Reputation is a forecast. A user's score predicts their future behavior, such as loan repayment or governance participation. Without a mechanism to reward accurate predictions, scores become subjective opinions.

Scoring rules are the solution. Protocols like UMA's Optimistic Oracle or Augur's prediction markets use financial incentives to align reporters with truth. A reporter earns rewards for accurate assessments and loses stake for incorrect ones.

Current systems are incomplete. Soulbound Tokens (SBTs) or Gitcoin Passport attestations are static records. They lack the dynamic financial feedback loop that scoring rules create, making them vulnerable to sybil attacks and stale data.

The market calibrates value. Just as Uniswap's AMM discovers token prices, a prediction market for reputation discovers the economic weight of a user's on-chain actions. This creates a self-correcting system where signal outweighs noise.