Vitalik Buterin Just Called Out DAOs—And He’s Right
The Ethereum co-founder says it’s time to stop playing treasury games and start building governance infrastructure.
You’ve seen the pattern: a new project launches, promises decentralised governance, distributes tokens, and then... nothing changes. The same whales vote on the same proposals while everyone else checks out. The DAO becomes either a rubber stamp for insiders or gridlocked.
Vitalik Buterin said the quiet part out loud.
In a post yesterday, Buterin argued that most DAOs are broken. They’re “a treasury controlled by token holder voting” that are “inefficient, vulnerable to capture and fail utterly at the goal of mitigating the weaknesses of human politics.”
He’s not wrong.
The Real Problem with DAOs
The current DAO model follows a simple playbook: distribute governance tokens, let holders vote on proposals, manage a treasury. It’s decentralised technically - there’s no single CEO, but it doesn’t improve on traditional governance. Often it makes things worse.
Why? Because most DAO governance suffers from three fatal flaws:
Low participation: Only the most dedicated (or most financially motivated) token holders bother to vote. Everyone else has decision fatigue.
Whale dominance: Large token holders can effectively control outcomes, making the “decentralisation” theatre.
Social gaming: Without privacy, governance becomes about optics and politics rather than substance.
These aren’t bugs to be fixed with minor tweaks. They’re features of the current design.
What DAOs Should Do Instead
Instead of managing treasuries, Buterin argues DAOs should tackle specific infrastructure problems:
Better oracles - Who determines what happened in the real world? Current oracle networks have centralisation risks and subjective disputes (like insurance claims) need dispute resolution mechanisms.
Onchain courts - When contracts fail or disputes arise, we need decentralised adjudication systems that can make nuanced judgments, not just execute code.
Long-term stewardship - Projects need governance structures that outlive their founding teams and can maintain critical infrastructure over decades.
Anti-scam registries - Shared lists and standards that communities can maintain together to protect users.
These problems could be solved by DAOs better than by centralised companies or chaotic free-for-alls. But they require rethinking what a DAO is.
Convex vs. Concave: A Framework for Better Governance
Buterin frames this using his “convex vs. concave” governance model.
Concave problems are ones where compromise beats a coin flip. Think: “What colour should the bike shed be?” Averaging many opinions produces reasonable results. For these, maximise input from diverse sources.
Convex problems require decisive action. Think: “Which technical architecture should we bet on?” Here you need strong leadership and decentralisation should provide accountability rather than trying to crowdsource the decision itself.
Most DAOs treat everything as a concave problem and end up with lowest-common-denominator mediocrity. The strongest governance systems recognise which type of problem they’re solving and adapt accordingly.
The Privacy and AI Question
Buterin identifies two key enablers for better DAOs:
Privacy tools - Zero-knowledge proofs, secure multi-party computation and homomorphic encryption could let people participate in governance without turning it into a social performance. You could verify that a vote was cast correctly without revealing who voted which way.
AI assistance - Not AI running DAOs (which Buterin cautions against), but AI helping participants analyse proposals, delegate votes to local models and reduce decision fatigue. Think of it as a governance copilot rather than autopilot.
These aren’t theoretical. The technology exists or is close. What’s missing is the design thinking to put it together.
The $17.5 Billion Question
DAO governance tokens currently have a market cap of at least $17.5 billion. Projects like Aave and Optimism have successfully used onchain governance for real protocol decisions and public goods funding.
But Buterin’s challenge is sharp: if you’re building new infrastructure - oracles, governance systems, anything that matters - DAO design should be 50% of your job, not 10%.
Why? Because Ethereum’s decentralised base layer only matters if the applications built on top are also decentralised. If every app ends up controlled by a handful of insiders or captured by special interests, we’ve just recreated Web2 with extra steps and worse UX.
The Bet
This isn’t the first time Buterin has pushed for better governance. But the timing matters. We’re seeing:
Mature DAOs struggling with participation and legitimacy
New infrastructure needs emerging (AI agents, cross-chain protocols, identity systems)
Technology capabilities catching up to governance aspirations
The next generation of DAOs won’t look like the current crop of token-voting treasuries. They’ll be purpose-built for specific problems, combine privacy and transparency strategically, and match their governance structure to the type of decisions they need to make.
Or they’ll fail, and we’ll keep pretending that putting “governance tokens” in people’s wallets somehow equals democracy.
Buterin is betting we can do better. The question is whether builders are listening.