Reputation Scoring

What it is

Reputation scoring aggregates signals about a wallet address — transaction history, protocol interactions, attestations, social activity, past defaults or slashing — into a single score or set of scores meant to approximate trustworthiness, contribution level, or creditworthiness. It's the on-chain analog of a credit score or a seller rating, used where identity can't be directly verified but behavior can be observed.

How it works

1. A scoring protocol defines the inputs it will consider: wallet age, transaction count and diversity, holdings of specific credentials or attestations, governance participation, DeFi repayment history, social graph signals, and so on.
2. Data is pulled either directly from on-chain state (readable by anyone) or from off-chain integrations (linked social accounts, KYC providers, off-chain attestations) that get anchored on-chain.
3. An algorithm — sometimes a simple weighted sum, sometimes a more complex model — combines these inputs into a score, often normalized to a range like 0-100.
4. The score is published or made queryable, either as a public on-chain value or through an API/oracle that other protocols can call.
5. Consuming applications set thresholds: "you need a score above X to access this undercollateralized loan," or "your allocation in this grants round is weighted by your reputation score" (as seen in projects like Orange Protocol or Galxe's credential system).
6. Scores are usually recomputed periodically, since reputation is meant to reflect an evolving history, not a one-time snapshot.

Why designers use it

• Enables differentiated treatment (better loan terms, higher trust limits, weighted votes) without requiring full legal identity disclosure.
• Rewards genuine long-term participants over one-off or bot accounts, since real reputation is costly and slow to build.
• Creates a portable trust layer that multiple applications can consume instead of each one building its own siloed history.

Failure modes

• Sybil farming of the score itself: if the inputs are things attackers can cheaply fabricate (fake transactions, wash trading, throwaway social accounts), the score becomes gameable and stops reflecting real trust.
• New user penalty: legitimate new users have no history and score poorly, effectively locking out exactly the people the ecosystem wants to onboard.
• Opaque or centralized scoring logic: if the algorithm isn't transparent or is controlled by one entity, users can't understand or contest their score, and the entity gains outsized influence over who gets access to opportunities.
• Score portability mismatch: a score built for one context (grant funding) may be misapplied to another (credit risk) where the underlying signals don't actually predict the outcome, leading to bad decisions downstream.

What to check before using it

• Are the underlying inputs to the score genuinely hard to fake at scale, or can they be farmed cheaply?
• Is the scoring algorithm transparent enough that users can understand and improve their score?
• Does the system have a reasonable path for legitimate new users who lack history?
• Who controls updates to the scoring model, and can that control be abused or captured?
• Is the score being used for the purpose it was actually validated for, or repurposed into a context it wasn't designed for?