Ethereum DAOs face overhaul as Vitalik warns token voting has failed

Summary

• Buterin argues most DAOs have devolved into token-controlled treasuries that are inefficient, vulnerable to whales, and far from Ethereum’s original governance vision.​
• He highlights five core DAO use cases: robust oracles, on-chain dispute resolution, shared “safe lists,” rapid short-term funding, and long-term project maintenance.​
• Vitalik proposes a convex/concave framework, private ZK voting, AI assistance (not control), and better communication tools to reduce capture, popularity contests, and decision fatigue.

Ethereum (ETH) co-founder Vitalik Buterin has outlined proposals for restructuring decentralized autonomous organizations (DAOs) in the cryptocurrency ecosystem, according to statements published by the developer.

Buterin stated that the Ethereum ecosystem requires more DAOs but argued that current implementations have diverged from the original design goals that informed the network’s development. According to his analysis, contemporary DAOs primarily function as treasuries controlled through token-holder voting mechanisms, a structure he characterized as inefficient and vulnerable to influence by large token holders.

Early Ethereum development incorporated DAOs as code-based systems operating on decentralized networks, intended to manage funds and decisions through automated protocols. The current token-voting model has led some users to question the effectiveness of DAO governance structures, according to Buterin’s statements.

The developer identified several areas where collective decision-making remains necessary for decentralized finance operations. Oracles, which supply external data to blockchain networks, represent a critical component for stablecoins, prediction markets and other DeFi applications, according to the analysis.

Current oracle designs face limitations, Buterin stated. Token-based oracles allow large holders to influence outcomes, particularly on subjective questions. The cost of attacking such systems cannot exceed their market capitalization, creating challenges for protecting large amounts of capital without imposing high fees, according to the assessment. Human-curated oracles reduce some vulnerabilities but compromise decentralization principles.

Additional challenges exist in on-chain dispute resolution for complex smart contracts such as insurance products, where subjective judgment is required. DAOs also maintain shared lists of trusted applications and verified contract addresses, which risk fragmentation without proper coordination mechanisms, according to Buterin.

The developer outlined five core use cases for improved DAO systems: enhanced oracle systems for stablecoins and prediction markets; on-chain dispute resolution for complex smart contracts; shared lists to protect users from fraudulent applications; rapid coordination for short-term community-funded projects; and ongoing maintenance when original development teams discontinue involvement.

Buterin proposed a “convex versus concave” framework for evaluating DAO designs. Concave problems benefit from compromise and averaged inputs, requiring systems resistant to capture and financial attacks. Convex problems reward decisive action and clear direction, where leadership can function effectively with decentralized oversight to prevent abuse, according to the framework.

Privacy emerged as a significant concern, with Buterin stating that lack of privacy can transform governance into popularity contests. Decision fatigue represents another challenge, as frequent voting reduces participation over time, according to the analysis.

The developer identified several technological approaches worth pursuing, including zero-knowledge proofs for private participation; limited deployment of multi-party computation or fully homomorphic encryption; software tools to reduce voting frequency; artificial intelligence systems to assist human judgment; and communication platforms designed for consensus-building.

Buterin cautioned against granting full control to large AI models, stating that AI should support human decision-making either at the organizational level or through user-controlled tools that execute votes on behalf of individuals.

Projects developing new oracle or governance systems should treat such work as a core priority rather than a secondary feature, according to Buterin, who stated this approach is necessary for maintaining decentralization across applications built on the Ethereum network.