A Comprehensive Study of Governance Issues in Decentralized Finance Applications

Wei Ma, Chenguang Zhu, Ye Liu, Xiaofei Xie, Yi Li

TL;DR

Decentralized Finance (DeFi) applications, while promising transparency and user control, suffer from significant governance issues. This paper introduces a novel taxonomy for DeFi governance, analyzes thousands of audit reports to identify common governance problems, and develops a tool to automatically detect inconsistencies between DeFi project whitepapers and their actual code implementations.

What Problem Does This Paper Address?

DeFi aims to revolutionize finance by removing intermediaries and empowering users. However, the rapid growth of DeFi has outpaced the development of robust governance mechanisms. This has led to various issues, including:

  • Vulnerable Governance: Applications lacking clear governance structures are difficult to maintain and update, while others are susceptible to malicious attacks due to poorly designed governance.
  • Opaque and Fraudulent Practices: Developers sometimes deviate from their whitepaper promises, leading to discrepancies that can harm users and erode trust.
  • Lack of Standardization: The absence of a standardized approach to DeFi governance makes it challenging for stakeholders to understand, assess, and improve these systems.

This paper addresses the critical need to systematically understand, categorize, and detect governance issues in DeFi applications to foster trust and promote the healthy growth of the ecosystem.

Methods

The authors employ a multi-faceted approach:

  1. Taxonomy Development (RQ1): They conduct a comprehensive literature review of academic papers and industry reports to develop a detailed taxonomy of DeFi governance. This taxonomy categorizes governance issues based on governance design, content, and implementation.
  2. Audit Report Analysis (RQ2): They collect and analyze 4,446 audit reports from 17 reputable Web3 security companies. Using their developed taxonomy, they categorize and analyze governance-related issues, identifying common themes and their severity. Natural Language Processing (NLP) techniques, specifically BERTopic, are used to cluster and identify the most frequent topics within governance issues.
  3. Whitepaper-Code Inconsistency Detection (RQ3): They develop a novel prototype tool that leverages Large Language Models (LLMs), specifically ChatGPT, to automatically detect inconsistencies between DeFi project whitepapers and their smart contract code. This tool focuses on tokenomics configurations like fee and interest rates.

Datasets

  • Audit Reports: A dataset of 4,446 audit reports from 17 reputable Web3 security companies (e.g., Certik, OpenZeppelin).
  • Academic Literature and Industry Blogs: A collection of relevant research papers and industry insights used for taxonomy development and understanding governance concepts.
  • DeFi Project Whitepapers and Source Codes: Whitepapers and public source codes from eight selected DeFi projects were used to validate the inconsistency detection tool.

Key Insights and Ideas

  • Disparity Between Code and Whitepapers is Key: A significant observation is that the gap between smart contract code and DeFi whitepapers plays a central role in many governance issues. This highlights a critical trust and transparency problem.
  • Ownership and Incentive Mechanisms Dominate: The most common governance issues identified in audit reports revolve around ownership structures and incentive schemes.
  • High Severity of Governance Issues: Approximately 38% of high-severity issues identified in audit reports are related to governance, underscoring its critical importance.
  • Acknowledged but Not Fixed: While most governance issues are either fixed or acknowledged by development teams, a substantial number of serious issues are acknowledged but not fixed, indicating a potential gap in follow-through.
  • Need for Standardized Governance Frameworks: The research emphasizes the lack of standardized governance development models in DeFi, similar to traditional software engineering, which hinders robust design and implementation.

Key Results

  • Governance Taxonomy: A detailed taxonomy for DeFi governance issues was developed, providing a structured approach to analysis.
  • Common Governance Issues: Ownership and incentive mechanisms were identified as the most prevalent governance issues.
  • Inconsistency Detection Tool: A prototype tool was developed to automatically detect inconsistencies between DeFi whitepapers and code. This tool achieved an F1 score of 56.14% and a recall of 80% on eight tested DeFi projects.
  • Prevalence of High-Severity Issues: Around 38% of all high-severity issues identified in the audit reports were governance-related.

Limitations and Future Work

  • Data Bias: The study acknowledges potential selection bias in the sample of DeFi application projects analyzed, as the DeFi ecosystem is constantly evolving.
  • Subjectivity in Topic Modeling: The use of AI models like BERTopic introduces some level of subjective judgment in extracting and interpreting themes.
  • Inconsistency Detector Limitations: Variations in whitepaper and source code formatting can pose challenges for the automated inconsistency detection tool.
  • Future Work: The authors suggest several avenues for future research, including:
  • Developing standardized governance frameworks and methodologies for DeFi.
  • Further investigating fairness in tokenomics and the balance between centralization and decentralization.
  • Creating more robust verification methodologies for governance system design.
  • Enhancing the inconsistency detection tool to handle diverse formats and styles.
  • Exploring the legal significance of whitepapers in the DeFi ecosystem and the need for regulatory oversight.

Summary updated

BibTeX
@article{ma2025comprehensive,
  author = {Wei Ma and Chenguang Zhu and Ye Liu and Xiaofei Xie and Yi Li},
  title = {{A Comprehensive Study of Governance Issues in Decentralized Finance Applications}},
  journal = {ACM Trans. Softw. Eng. Methodol.},
  year = {2025},
  doi = {10.1145/3717062},
}
Figure