Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

New Online Communities: Graph Deep Learning on Anonymous Voting Networks to Identify Sybils in Polycentric Governance

Quinn DuPont

TL;DR

DAOs face significant Sybil challenges due to anonymous voting, which can distort governance. The authors develop a graph deep learning pipeline that learns node embeddings with a GCNN and clusters them via FAISS to identify Sybil activity in Snapshot governance data, achieving a 2-5% reduction in the voting graph without requiring new identity systems. Framed within Ostrom-inspired Digital Common Pool Resources and polycentric governance, the work links neoinstitutional theory to practical forensics for digital asset commons. The approach offers scalable, privacy-preserving avenues for improving governance integrity, with implications for policy, regulation, and the design of resilient decentralized communities.

Abstract

This research examines the polycentric governance of digital assets in blockchain-based Decentralized Autonomous Organizations (DAOs). It offers a theoretical framework and addresses a critical challenge facing decentralized governance by developing a method to identify Sybils, or spurious identities. Sybils pose significant organizational sustainability threats to DAOs and other, commons-based online communities, and threat models are identified. The experimental method uses an autoencoder architecture and graph deep learning techniques to identify Sybil activity in a DAO governance dataset (snapshot.org). Specifically, a Graph Convolutional Neural Network (GCNN) learned voting behaviours and a fast vector clustering algorithm used high-dimensional embeddings to identify similar nodes in a graph. The results reveal that deep learning can effectively identify Sybils, reducing the voting graph by 2-5%. This research underscores the importance of Sybil resistance in DAOs, identifies challenges and opportunities for forensics and analysis of anonymous networks, and offers a novel perspective on decentralized governance, informing future policy, regulation, and governance practices.

New Online Communities: Graph Deep Learning on Anonymous Voting Networks to Identify Sybils in Polycentric Governance

TL;DR

DAOs face significant Sybil challenges due to anonymous voting, which can distort governance. The authors develop a graph deep learning pipeline that learns node embeddings with a GCNN and clusters them via FAISS to identify Sybil activity in Snapshot governance data, achieving a 2-5% reduction in the voting graph without requiring new identity systems. Framed within Ostrom-inspired Digital Common Pool Resources and polycentric governance, the work links neoinstitutional theory to practical forensics for digital asset commons. The approach offers scalable, privacy-preserving avenues for improving governance integrity, with implications for policy, regulation, and the design of resilient decentralized communities.

Abstract

This research examines the polycentric governance of digital assets in blockchain-based Decentralized Autonomous Organizations (DAOs). It offers a theoretical framework and addresses a critical challenge facing decentralized governance by developing a method to identify Sybils, or spurious identities. Sybils pose significant organizational sustainability threats to DAOs and other, commons-based online communities, and threat models are identified. The experimental method uses an autoencoder architecture and graph deep learning techniques to identify Sybil activity in a DAO governance dataset (snapshot.org). Specifically, a Graph Convolutional Neural Network (GCNN) learned voting behaviours and a fast vector clustering algorithm used high-dimensional embeddings to identify similar nodes in a graph. The results reveal that deep learning can effectively identify Sybils, reducing the voting graph by 2-5%. This research underscores the importance of Sybil resistance in DAOs, identifies challenges and opportunities for forensics and analysis of anonymous networks, and offers a novel perspective on decentralized governance, informing future policy, regulation, and governance practices.
Paper Structure (23 sections, 11 equations, 25 figures, 3 tables)

This paper contains 23 sections, 11 equations, 25 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Deciphering DAOs: Advancing Research in Digital Collective Action
  3. Emergence and Evolution of DAO Studies
  4. Quantitative and Empirical DAO Research
  5. Existing Research Challenges for DAO Studies
  6. Online Communities
  7. Motivation and Governance in the Transition from Web 2.0 to Web3
  8. Towards Polycentric Governance
  9. Digital Common Pool Resources
  10. Graph Deep Learning to Identify Sybils
  11. Ethical considerations
  12. Sybil Resistance and Defence in Depth for Community Design
  13. Threat Model
  14. Defining Sybils and Deanonymizing Social Networks
  15. Experimental Sybil Identification Methodology
  16. ...and 8 more sections

Figures (25)

  • Figure 1: Herfindahl-Hirschman Index
  • Figure 2: Gini Coefficients
  • Figure 3: n-top Players Analysis
  • Figure 4: Unique Voting 'Strategies' used by 100 DAOs.
  • Figure 5: Minimum cost to purchase governance proposals in three example DAOs.
  • ...and 20 more figures