Architecture

High-level implementation as a compute layer

Our Protocol Stack

Data

This layer is about sourcing the attestation data related to reputation across a variety of use cases. This data maybe on-chain events and blockchain transactions (souldbound NFTs, token transactions or events) or off-chain (verifiable credentials, verifiable voting, social graph or events).

Once the data is sourced and authenticated, it will be transformed and stored for EigenTrust computation.

EigenTrust APIs (Reputation/Ranking models)

This layer will provide developers with APIs and SDKs to configure their own reptuation algorithms without having to worry about managing the data or learning and managing reputation and recommendation algorithms - EigenTrust APIs will take care of everything.

Developers can take control of the core value propositions and experiment with different type of reputation heuristics relevant for their use case, leveraging our public infrastructure.

Decentralized Compute Network

This layer will be responsible for running the distributed EigenTrust compute at scale and ensure availability, reliability and verifiability of the computation.

Architecture Flow

Architecture diagram

Raw Data is any application/community specific on-chain or off-chain data, ex: token transactions, follows/likes on social graph, soulbound tokens issued/received.

Aggregator–Transformer processes the raw data based on the desired reputation/ranking heuristics, and turns them into various input datasets for consumption by EigenTrust—local trust and seed trust—e.g. your follower list on web3 social, your trusted wallets based on certain type of on-chain transactions.

Algorithm/Parameters define the trust or ranking system for your app/marketplace, e.g. “if X follows Y on web3 social, assign unit weight in local trust from X to Y”, “if X sends >0.1ETH to Y, add to the local trust from X to Y the amount over 0.1ETH.”

Seed Input is a combination of trustworthy peers in the network; it could also be your own self. The users and/or app developer can define this set.

For example, for personalized recommendation on web3 social it could be the people you already follow (user-side input), community members with highest participation in a DAO (community-/app-developer-side input).