Reputation Score

A key feature is that reputation scores are composable primitives, meaning they can be used as inputs by other smart contracts and protocols. This enables portable reputation, where a user's score from one application can be leveraged in another without starting from zero. For example:

• A lending protocol can automatically adjust collateral factors based on a borrower's on-chain history.
• A governance system can weight votes using a user's proven contribution score. This creates a network effect for trust, reducing redundancy across the ecosystem.

Unlike static credentials, a robust reputation score is dynamic, updating in near real-time based on new on-chain transactions and behaviors. It is also context-aware, meaning the calculation weights different activities based on the specific use case. For instance:

• Lending context: Heavy weight on repayment history and collateral management.
• Governance context: Heavy weight on proposal quality and voting participation.
• Trading context: Analysis of wash trading patterns and MEV-related activity. This ensures the score remains relevant and resistant to manipulation for its intended application.

The foundational data for a reputation score is sourced directly from immutable, public blockchain ledgers. This provides cryptographic proof of provenance for every data point. The scoring algorithm's logic (or its verifiable credentials) can often be audited, making the score transparent and falsifiable. Key aspects include:

• On-chain data: Transaction history, token holdings, smart contract interactions.
• Off-chain attestations: Verifiable credentials (e.g., from DAOs, KYC providers) can be incorporated via oracles.
• Auditable logic: The methodology is open for review, though specific model weights may be private to prevent gaming.

A core technical challenge is ensuring a score represents a unique entity and is resistant to Sybil attacks, where a single user creates many addresses to inflate influence. Reputation systems employ various mechanisms for uniqueness proof:

• Graph analysis: Mapping transaction relationships to cluster addresses likely controlled by one entity.
• Staking/PoH: Requiring a stake or proof of unique humanity (e.g., World ID) to mint a verifiable reputation credential.
• Activity consolidation: Scoring based on the aggregate behavior of a clustered entity, not individual throwaway addresses. This prevents the trivial forgery of reputation and maintains the metric's economic value.

A reputation score distills complex behavioral data into a quantitative metric, typically a numerical value or tier (e.g., 0-1000, or Bronze/Silver/Gold). However, it is often multi-dimensional, comprising several sub-scores or attributes. For example, a comprehensive score might include separate components for:

• Financial trustworthiness: Credit history, collateralization ratios.
• Governance participation: Proposal success, voting consistency.
• Technical reliability: Smart contract deployment audit status, bug bounty participation.
• Social contribution: Grants received, community engagement metrics. This allows for nuanced assessment beyond a single number.

The primary utility of reputation scores is to reduce information asymmetry and enable new financial and governance primitives. Concrete applications include:

• Under-collateralized Lending: Using a high reputation score as a substitute for excess collateral, enabling credit lines.
• Risk-Adjusted Parameters: Protocols dynamically adjusting loan-to-value ratios, liquidation thresholds, or insurance premiums based on user reputation.
• Sybil-Resistant Governance: Weighting voting power via proof-of-personhood and contribution history instead of mere token holdings.
• Curated Registries & Access: Granting access to whitelists, beta programs, or exclusive pools based on proven track records.

The frequency, consistency, and economic weight of an address's on-chain activity. This includes:

• Total transaction count and total value transacted.
• Time-weighted activity to prioritize recent behavior.
• Analysis of transaction types (e.g., swaps, mints, transfers). High, sustained volume from legitimate sources signals active and economically significant participation.

Measures the quality and sophistication of engagements with DeFi protocols, beyond simple transfers. Key factors include:

• Number of unique protocols interacted with.
• Complexity of interactions (e.g., providing liquidity, borrowing, staking, voting).
• Duration of engagements (e.g., long-term liquidity provision). Deep, diversified protocol usage indicates experienced and committed network participation.

Assesses the capital efficiency and solvency of an address. This is calculated using:

• Portfolio concentration and diversification across assets.
• Debt-to-collateral ratios for addresses using lending protocols.
• Liquidation history and proximity to liquidation thresholds.
• Profit/loss trends from trading and yield farming activities. Strong financial health reduces perceived counterparty risk.

Algorithms to detect and penalize Sybil attacks, where a single entity controls multiple addresses to manipulate scores. Methods include:

• Graph analysis of funding sources and transaction clusters.
• Behavioral fingerprinting to identify patterns common to bot networks.
• Asset provenance tracing to identify airdrop farming or wash trading. A high uniqueness score confirms the address represents a distinct, non-collusive actor.

Evaluates contribution to decentralized decision-making within DAOs and protocols. Metrics include:

• Proposal creation and voting frequency.
• Voting power (token-weighted) and delegation activity.
• Sentiment and consistency analysis of votes relative to the community. Active, informed governance participation signals long-term alignment with a protocol's success.

A network-derived metric based on the reputation of an address's peers. It leverages the adage "you are judged by the company you keep." Calculated by:

• Analyzing the reputation scores of frequent transaction counterparts.
• Mapping subgraph relationships within lending pools, NFT communities, or DAOs.
• Identifying associations with known malicious or sanctioned addresses, which negatively impacts the score.

An attacker creates a large number of pseudonymous identities (Sybils) to artificially inflate or manipulate a reputation score. This is a fundamental threat to any decentralized scoring system.

• Mechanism: An entity spins up thousands of wallets to simulate organic, trustworthy behavior (e.g., small successful transactions, governance participation) across a network.
• Goal: To gain disproportionate influence, such as skewing a creditworthiness assessment, manipulating a decentralized curation market, or gaming airdrop eligibility.
• Mitigation: Often requires costly signaling (like proof-of-stake), social graph analysis, or persistent identity solutions to increase the attack's economic cost.

Attackers deliberately execute on-chain transactions designed to corrupt the input data used to calculate reputation scores.

• Example: A lending protocol uses "successful repayment history" as a score component. An attacker could create a circular lending scheme between their own wallets, generating a false history of reliable debt repayment without real economic risk.
• Impact: The scoring model ingests garbage data, leading to inaccurate outputs that can be exploited for financial gain (e.g., securing an under-collateralized loan).
• Defense: Requires robust feature engineering that filters out wash trading, self-dealing, and other forms of meaningless on-chain noise.

Many reputation scores rely on oracles to fetch off-chain data (e.g., traditional credit scores, KYC status, real-world asset ownership). These oracles become a critical attack surface.

• Vector: If the oracle is compromised or provides incorrect data, the resulting reputation score is fundamentally flawed. An attacker might bribe or hack an oracle node to report false positive data for their address.
• Systemic Risk: A single point of failure in the oracle design can undermine the entire reputation system's security guarantees.
• Solution: Employ decentralized oracle networks with multiple independent nodes and cryptographic proofs of data correctness.

An adversary reverse-engineers the scoring algorithm by probing the system with various transactions and observing the resulting score changes.

• Process: Through repeated queries, the attacker approximates the model's weights and thresholds (e.g., "10 DeFi swaps increase score by X, 5 NFT mints by Y").
• Outcome: Once the model is understood, the attacker can optimize for the score rather than genuine good behavior, efficiently gaming the system at minimal cost. This renders the score meaningless as a trust signal.
• Prevention: Techniques include differential privacy, adding random noise to outputs, or using zero-knowledge proofs to compute scores without revealing the underlying logic.

The entity or DAO that controls the scoring parameters, model updates, or whitelists holds significant power, creating a central point of failure.

• Risks Include:
  • Admin Key Compromise: A malicious actor gains control of the upgradeable contract's admin keys and can arbitrarily set scores.
  • Governance Capture: A token-based governance system is taken over by a malicious coalition to vote in scoring rules that benefit them.
  • Censorship: The governing body can blacklist addresses, effectively setting their reputation to zero outside of transparent on-chain logic.
• Mitigation: Time-locked upgrades, decentralized governance with robust safeguards, and immutable scoring contracts for core logic.

A high-dimensional reputation score can itself become a privacy vulnerability, potentially linking a user's pseudonymous addresses or revealing sensitive financial behavior.

• Threat: By analyzing the nuanced components of a score (e.g., specific protocol usage, transaction timing patterns, asset holdings inferred from interactions), an observer can cluster addresses believed to belong to the same entity or make inferences about a user's wealth and strategy.
• Consequence: Breaches the pseudonymity expectation of many blockchain users and could lead to targeted phishing, extortion, or regulatory scrutiny.
• Countermeasure: Zero-knowledge reputation proofs allow a user to prove their score meets a threshold (e.g., >700) without revealing the underlying transaction history or the exact score value.

The property of a system that prevents a single entity from creating multiple fake identities (Sybils) to manipulate outcomes. Reputation scores are a primary tool for achieving Sybil resistance in decentralized governance, airdrops, and access control by linking identity to provable, costly on-chain history.

• Key Mechanism: Makes identity creation expensive by requiring a history of verifiable, valuable actions.
• Example: Using transaction volume and age to filter out low-cost, newly created wallets from a governance vote.

A persistent, verifiable representation of an actor (user, DAO, protocol) constructed from their immutable transaction history on a blockchain. It is the foundational data layer for calculating a reputation score.

• Components: Includes wallet addresses, transaction patterns, asset holdings, and interaction history with smart contracts.
• Contrasts with traditional, off-chain identity (KYC) by being self-sovereign, permissionless, and pseudonymous.

A consensus mechanism where token holders vote to elect a limited set of validators. Reputation scores can inform these votes, moving the system beyond pure token-weighted voting (Proof of Stake).

• Application: Voters may delegate stakes to validators with high reputation scores for reliability and honest historical performance.
• Goal: Enhances network security and governance by aligning economic stake with proven, long-term participation.

The assessment of a borrower's creditworthiness in decentralized finance, often using reputation scores derived from on-chain behavior as a proxy for traditional financial history.

• Data Points: Includes collateralization history, repayment of previous loans, and overall wallet activity.
• Use Case: Enables under-collateralized lending by allowing protocols like Goldfinch or Maple Finance to assess entity risk beyond just locked assets.

A token that is permanently bound to a specific wallet address and cannot be transferred or sold. They are often proposed as a technical primitive for encoding reputation and attestations on-chain.

• Function: Acts as a verifiable, tamper-proof record of achievements, memberships, or credentials.
• Relation to Reputation: A reputation score can be computed by aggregating the meaning and provenance of multiple SBTs held by a single identity.

The method of calculating a participant's voting power in a DAO or protocol. Moving beyond simple token voting (Token-Centric Governance), reputation scores can be used to weight votes, granting more influence to proven long-term contributors.

• Mechanism: A governance power formula might combine token balance with a reputation score derived from participation history.
• Goal: Mitigates plutocracy and rewards sustained, valuable engagement over mere capital allocation.