DAO Governance at a Crossroads: Dom vs Vitalik

A sharp debate over the future of decentralized governance has flared after Vitalik Buterin argued that today’s dominant DAO model has drifted away from its original purpose, prompting a pointed response from Dominic Williams, who says at least one network is already doing what critics claim is impossible.

Buterin’s intervention cuts straight to a growing unease across crypto governance. “We need more DAOs – but different and better DAOs,” he wrote, reflecting on how early visions of decentralized autonomous organizations imagined systems that could manage resources and coordinate action more robustly than governments or corporations. That ambition, he argues, has been diluted. What now passes for a DAO often amounts to “a treasury controlled by token holder voting,” a structure that “works” in a narrow sense but is “inefficient, vulnerable to capture, and fails utterly at the goal of mitigating the weaknesses of human politics.”

Despite the cynicism that has followed, Buterin insists DAOs remain essential. He lists concrete gaps that centralized institutions and current crypto tooling struggle to fill: better oracles for stablecoins and prediction markets, onchain dispute resolution for use cases like insurance, the maintenance of trusted lists such as secure applications and canonical interfaces, rapid coordination for short-lived projects, and long-term stewardship when original teams disappear.

His critique of token-based oracles is especially blunt. Where voting power is tied to tokens, “whales can manipulate the answer on a subjective issue,” and the system’s security ceiling is constrained by its market capitalisation. The problem, he says, is not greed but design. Building better systems is “not just a technical problem but also a social problem.”

To frame when decentralisation should average opinions and when it should empower decisive leadership, Buterin leans on his own “convex vs concave” framework. For concave problems, compromise beats coin flips, favouring robust mechanisms that resist capture. For convex problems, strong leadership can be useful, provided decentralised checks remain in place.

‘ICP’s Network Nervous System has processed 1000s of proposals & orchestrated the Internet Computer network with 100% automation’

Two obstacles loom over all of this: privacy and decision fatigue. Without privacy, governance becomes “a social game.” Without relief from constant voting, early enthusiasm decays into disengagement. Here, Buterin sees promise in modern tools. Zero-knowledge techniques can protect privacy. AI can reduce decision fatigue. Consensus-finding communication platforms can outperform noisy social feeds. Still, he is careful to draw a line: “AI must be used carefully.” It should “scale and enhance human intention and judgement,” not replace it, and certainly not be handed full control.

This is where Dom steps in with a challenge to the premise. Responding to Buterin’s critique, he wrote: “True? Modern DAO ‘design … is inefficient, vulnerable to capture, and fails utterly at the goal of mitigating the weaknesses of human politics’” before pointing to a counterexample. “ICP’s Network Nervous System has processed 1000s of proposals & orchestrated the Internet Computer network with 100% automation.”

The claim is more than rhetorical. On the Internet Computer, governance is handled by the Network Nervous System, a protocol-level mechanism that automatically executes accepted proposals without human intermediaries. Supporters argue this reduces discretionary power, limits capture, and turns governance into a continuous, machine-enforced process rather than episodic political theatre.

The exchange highlights a fault line opening across the sector. One camp, articulated by Buterin, sees current DAOs as a necessary but deeply flawed stepping stone that demands new designs built with privacy, communication, and human limits in mind. Another points to live networks where automation already replaces many of the governance frictions critics describe.

A sharp debate over the future of decentralized governance has flared after Vitalik Buterin argued that today’s dominant DAO model has drifted away from its original purpose, prompting a pointed response from Dominic Williams, who says at least one network is already doing what critics claim is impossible.

Buterin’s intervention cuts straight to a growing unease across crypto governance. “We need more DAOs – but different and better DAOs,” he wrote, reflecting on how early visions of decentralized autonomous organizations imagined systems that could manage resources and coordinate action more robustly than governments or corporations. That ambition, he argues, has been diluted. What now passes for a DAO often amounts to “a treasury controlled by token holder voting,” a structure that “works” in a narrow sense but is “inefficient, vulnerable to capture, and fails utterly at the goal of mitigating the weaknesses of human politics.”

Despite the cynicism that has followed, Buterin insists DAOs remain essential. He lists concrete gaps that centralized institutions and current crypto tooling struggle to fill: better oracles for stablecoins and prediction markets, onchain dispute resolution for use cases like insurance, the maintenance of trusted lists such as secure applications and canonical interfaces, rapid coordination for short-lived projects, and long-term stewardship when original teams disappear.

His critique of token-based oracles is especially blunt. Where voting power is tied to tokens, “whales can manipulate the answer on a subjective issue,” and the system’s security ceiling is constrained by its market capitalisation. The problem, he says, is not greed but design. Building better systems is “not just a technical problem but also a social problem.”

To frame when decentralisation should average opinions and when it should empower decisive leadership, Buterin leans on his own “convex vs concave” framework. For concave problems, compromise beats coin flips, favouring robust mechanisms that resist capture. For convex problems, strong leadership can be useful, provided decentralised checks remain in place.

‘ICP’s Network Nervous System has processed 1000s of proposals & orchestrated the Internet Computer network with 100% automation’

Two obstacles loom over all of this: privacy and decision fatigue. Without privacy, governance becomes “a social game.” Without relief from constant voting, early enthusiasm decays into disengagement. Here, Buterin sees promise in modern tools. Zero-knowledge techniques can protect privacy. AI can reduce decision fatigue. Consensus-finding communication platforms can outperform noisy social feeds. Still, he is careful to draw a line: “AI must be used carefully.” It should “scale and enhance human intention and judgement,” not replace it, and certainly not be handed full control.

This is where Dom steps in with a challenge to the premise. Responding to Buterin’s critique, he wrote: “True? Modern DAO ‘design … is inefficient, vulnerable to capture, and fails utterly at the goal of mitigating the weaknesses of human politics’” before pointing to a counterexample. “ICP’s Network Nervous System has processed 1000s of proposals & orchestrated the Internet Computer network with 100% automation.”

The claim is more than rhetorical. On the Internet Computer, governance is handled by the Network Nervous System, a protocol-level mechanism that automatically executes accepted proposals without human intermediaries. Supporters argue this reduces discretionary power, limits capture, and turns governance into a continuous, machine-enforced process rather than episodic political theatre.

The exchange highlights a fault line opening across the sector. One camp, articulated by Buterin, sees current DAOs as a necessary but deeply flawed stepping stone that demands new designs built with privacy, communication, and human limits in mind. Another points to live networks where automation already replaces many of the governance frictions critics describe.

A sharp debate over the future of decentralized governance has flared after Vitalik Buterin argued that today’s dominant DAO model has drifted away from its original purpose, prompting a pointed response from Dominic Williams, who says at least one network is already doing what critics claim is impossible.

Buterin’s intervention cuts straight to a growing unease across crypto governance. “We need more DAOs – but different and better DAOs,” he wrote, reflecting on how early visions of decentralized autonomous organizations imagined systems that could manage resources and coordinate action more robustly than governments or corporations. That ambition, he argues, has been diluted. What now passes for a DAO often amounts to “a treasury controlled by token holder voting,” a structure that “works” in a narrow sense but is “inefficient, vulnerable to capture, and fails utterly at the goal of mitigating the weaknesses of human politics.”

Despite the cynicism that has followed, Buterin insists DAOs remain essential. He lists concrete gaps that centralized institutions and current crypto tooling struggle to fill: better oracles for stablecoins and prediction markets, onchain dispute resolution for use cases like insurance, the maintenance of trusted lists such as secure applications and canonical interfaces, rapid coordination for short-lived projects, and long-term stewardship when original teams disappear.

His critique of token-based oracles is especially blunt. Where voting power is tied to tokens, “whales can manipulate the answer on a subjective issue,” and the system’s security ceiling is constrained by its market capitalisation. The problem, he says, is not greed but design. Building better systems is “not just a technical problem but also a social problem.”

To frame when decentralisation should average opinions and when it should empower decisive leadership, Buterin leans on his own “convex vs concave” framework. For concave problems, compromise beats coin flips, favouring robust mechanisms that resist capture. For convex problems, strong leadership can be useful, provided decentralised checks remain in place.

‘ICP’s Network Nervous System has processed 1000s of proposals & orchestrated the Internet Computer network with 100% automation’

Two obstacles loom over all of this: privacy and decision fatigue. Without privacy, governance becomes “a social game.” Without relief from constant voting, early enthusiasm decays into disengagement. Here, Buterin sees promise in modern tools. Zero-knowledge techniques can protect privacy. AI can reduce decision fatigue. Consensus-finding communication platforms can outperform noisy social feeds. Still, he is careful to draw a line: “AI must be used carefully.” It should “scale and enhance human intention and judgement,” not replace it, and certainly not be handed full control.

This is where Dom steps in with a challenge to the premise. Responding to Buterin’s critique, he wrote: “True? Modern DAO ‘design … is inefficient, vulnerable to capture, and fails utterly at the goal of mitigating the weaknesses of human politics’” before pointing to a counterexample. “ICP’s Network Nervous System has processed 1000s of proposals & orchestrated the Internet Computer network with 100% automation.”

The claim is more than rhetorical. On the Internet Computer, governance is handled by the Network Nervous System, a protocol-level mechanism that automatically executes accepted proposals without human intermediaries. Supporters argue this reduces discretionary power, limits capture, and turns governance into a continuous, machine-enforced process rather than episodic political theatre.

The exchange highlights a fault line opening across the sector. One camp, articulated by Buterin, sees current DAOs as a necessary but deeply flawed stepping stone that demands new designs built with privacy, communication, and human limits in mind. Another points to live networks where automation already replaces many of the governance frictions critics describe.

A sharp debate over the future of decentralized governance has flared after Vitalik Buterin argued that today’s dominant DAO model has drifted away from its original purpose, prompting a pointed response from Dominic Williams, who says at least one network is already doing what critics claim is impossible.

Buterin’s intervention cuts straight to a growing unease across crypto governance. “We need more DAOs – but different and better DAOs,” he wrote, reflecting on how early visions of decentralized autonomous organizations imagined systems that could manage resources and coordinate action more robustly than governments or corporations. That ambition, he argues, has been diluted. What now passes for a DAO often amounts to “a treasury controlled by token holder voting,” a structure that “works” in a narrow sense but is “inefficient, vulnerable to capture, and fails utterly at the goal of mitigating the weaknesses of human politics.”

Despite the cynicism that has followed, Buterin insists DAOs remain essential. He lists concrete gaps that centralized institutions and current crypto tooling struggle to fill: better oracles for stablecoins and prediction markets, onchain dispute resolution for use cases like insurance, the maintenance of trusted lists such as secure applications and canonical interfaces, rapid coordination for short-lived projects, and long-term stewardship when original teams disappear.

His critique of token-based oracles is especially blunt. Where voting power is tied to tokens, “whales can manipulate the answer on a subjective issue,” and the system’s security ceiling is constrained by its market capitalisation. The problem, he says, is not greed but design. Building better systems is “not just a technical problem but also a social problem.”

To frame when decentralisation should average opinions and when it should empower decisive leadership, Buterin leans on his own “convex vs concave” framework. For concave problems, compromise beats coin flips, favouring robust mechanisms that resist capture. For convex problems, strong leadership can be useful, provided decentralised checks remain in place.

‘ICP’s Network Nervous System has processed 1000s of proposals & orchestrated the Internet Computer network with 100% automation’

Two obstacles loom over all of this: privacy and decision fatigue. Without privacy, governance becomes “a social game.” Without relief from constant voting, early enthusiasm decays into disengagement. Here, Buterin sees promise in modern tools. Zero-knowledge techniques can protect privacy. AI can reduce decision fatigue. Consensus-finding communication platforms can outperform noisy social feeds. Still, he is careful to draw a line: “AI must be used carefully.” It should “scale and enhance human intention and judgement,” not replace it, and certainly not be handed full control.

This is where Dom steps in with a challenge to the premise. Responding to Buterin’s critique, he wrote: “True? Modern DAO ‘design … is inefficient, vulnerable to capture, and fails utterly at the goal of mitigating the weaknesses of human politics’” before pointing to a counterexample. “ICP’s Network Nervous System has processed 1000s of proposals & orchestrated the Internet Computer network with 100% automation.”

The claim is more than rhetorical. On the Internet Computer, governance is handled by the Network Nervous System, a protocol-level mechanism that automatically executes accepted proposals without human intermediaries. Supporters argue this reduces discretionary power, limits capture, and turns governance into a continuous, machine-enforced process rather than episodic political theatre.

The exchange highlights a fault line opening across the sector. One camp, articulated by Buterin, sees current DAOs as a necessary but deeply flawed stepping stone that demands new designs built with privacy, communication, and human limits in mind. Another points to live networks where automation already replaces many of the governance frictions critics describe.