In a decentralized world with anonymous users, it can be challenging to trust others for transactions, services, and various interactions. Openrank provides a solution by offering a score between 0 and 1 that indicates a user's trustworthiness based on their engagement within the community. Available on the Farcaster and Lens protocols, Openrank assigns grades to users based on their interactions with others.

How Openrank Works

Openrank employs an algorithm known as EigenTrust to generate these scores. Let's dive into how this algorithm functions and the value of a decentralized ranking and reputation protocol in web3 social media.

Applications of Openrank

Social Networks

Open social graphs require a ranking and algorithm layer for search, discovery, and recommendations without centralized moderation. Web3 social graph protocols generate peer-to-peer engagement data (likes, follows, comments, etc.) that can be used to create reputation graphs through algorithms like EigenTrust. This results in a ranking or score for users, which can be customized by developers. An open ranking layer enables personalized community experiences, reduces spam, and allows users and developers to choose their algorithms.

Marketplaces

Permissionless marketplaces need a verifiable reputation layer to address fraud and scams. A reputation system can provide contextual peer-to-peer attestations, making transactions safer and more efficient. An open ranking layer brings transparency, reduces trust and safety costs, and opens innovation for fraud detection. Consumer and developer marketplaces can use these rankings for better search and discovery.

Consumer Apps and Wallets

The growth of on-chain users necessitates ranking and reputation systems for consumer apps. An open ranking layer allows personalized feeds and recommendations, enhances reputation graph composability across apps, and focuses on social discovery for users.

Governance and Public Goods Funding

Reputation systems help address trust, voting, and resource allocation in governance. An open ranking layer allows communities to use reputation for governance and funding decisions, capturing peer-to-peer trust signals, and creating reputation graphs that reflect domain expertise.

EigenTrust: The Algorithm Behind Openrank

The EigenTrust algorithm is a reputation management system used in peer-to-peer (P2P) networks. Developed by Sep Kamvar, Mario Schlosser, and Hector Garcia-Molina, EigenTrust aims to assign a unique global trust value to each peer in the network based on their history of interactions, thus reducing the prevalence of inauthentic or malicious behavior.

How EigenTrust Works:

Imagine a social media platform where users share content, follow each other, and leave reviews or likes on posts. The platform wants to highlight trustworthy users whose content is valuable and reliable.

• Local Trust Values:

  • Alice interacts with Bob and David.

  • Alice finds Bob's posts very informative and gives him a high trust value of 0.9.

  • Alice occasionally finds David’s posts useful, giving him a trust value of 0.6.

• Normalization:

  • Alice's trust values are normalized to ensure fairness.

  • Bob’s normalized trust value: 0.9 / 1.5 = 0.6.

  • David’s normalized trust value: 0.6 / 1.5 = 0.4.

• Global Trust Values:

  • The platform combines all local trust values using an algorithm similar to Google’s PageRank.

  • This involves creating a matrix of normalized trust values and using the power iteration method to find the principal eigenvector, representing the global trust values.

• Transitive Trust:

  • The algorithm assumes that trust is transitive.

  • Alice trusts Bob directly, and Bob trusts Charlie with a value of 0.8.

  • Therefore, Alice is likely to trust Charlie indirectly.

  • Alice’s transitive trust in Charlie is calculated as:

    • Alice’s normalized trust in Bob (0.6) * Bob’s trust in Charlie (0.8) = 0.48.

• Convergence:

  • After several iterations, the trust values stabilize, reflecting each user’s reputation.

  • For example, Bob might end up with a global trust value of 0.7, David might have a global trust value of 0.5, and Charlie might achieve a global trust value reflecting both his direct interactions and indirect trust through others like Alice and Bob.

Benefits for a Social Media Platform

• Community-Driven Trust: Trust is built collectively by the community rather than imposed by a central authority, leading to a more democratic and robust system.

• Efficient Content Moderation: The platform can efficiently highlight quality content and reduce harmful content, improving user experience and engagement.

Key Features of the Algorithm

• Decentralized: Designed for decentralized P2P networks, requiring no central authority to compute or enforce trust values.

• Resilience to Malicious Peers: By aggregating trust values from multiple peers, the algorithm reduces the influence of malicious peers who might try to artificially inflate their trust scores.

• Scalability: The algorithm can handle large networks due to its iterative and distributed nature.

For more technical details about Openrank and its development, check out Ilemi's Articles.