0xSybil

Establishing a robust Sybil defense is an important aspect of a healthy community and fair democracy. We teamed up to prevent the attacks on our democracy, by analyzing open on-chain data.

We use the big data approaches to detect similarity patterns in wallet history, an extract the clusters of similar accounts. All the accounts that have a high-risk of sybil attack are marked in Ethereum Attestation Service, to make it a public good for all the Ethereum community.

We are aiming to build a tool that communities, DAOs, and protocols can use to detect fraudulent activity in their ecosystem and share it with everyone by making it publicly available on-chain.

We took a real snapshot from the Talent Protocol, which was manually verified that it was affected by a Sybil attack(https://twitter.com/TakeOffWeb3/status/1672274180484595714). We then analyzed all 179 voters and their on-chain transaction history. While manually were found 38 fake accounts, we identified more than 50 highly suspicious and connected accounts. Any suspicious addresses identified during the analysis were saved on-chain using EOAS. The results of this analysis are displayed at the frontend, where owners of suspicious addresses can challenge the attestation results using world coin proof of human verification.

The project consists of two main parts:

1. A Python code that retrieves a snapshot and examines all address cross-chain activities for similarities, clustering them into risk accounts. Transactions from different voters are analyzed for similarities in terms of the types of assets involved and the timing of those transactions. We use various available APIs, including Airtsack for fetching Poaps and Zerion for transaction history across different chains.
2. A NextJS-based application that offers an interface and API to interact with the Ethereum attestation service, world coin, and our own dataset.