DAO Governance Failures: What MakerDAO, Uniswap, and Compound Teach About Decentralization

The theory of decentralized autonomous organizations is elegant. Replace hierarchical corporate governance with on-chain voting: token holders propose and ratify decisions, code executes automatically, no single point of failure, no CEO to fire. The DAO promised governance without gatekeepers — a system as transparent as its smart contracts.

The practice has been considerably more complicated. The major DeFi protocols that pioneered practical DAO governance — Compound, MakerDAO, and Uniswap — have accumulated several years of operational history now, and that history contains a series of instructive failures and near-failures that reveal structural problems in the DAO model that go beyond implementation details.

## Compound: The Proposal That Drained $80 Million

In September 2021, Compound experienced what became known as the "bug in governance" incident. A routine upgrade proposal — Proposal 062 — contained a bug in the newly deployed Comptroller contract that caused it to distribute COMP tokens incorrectly to users who supplied or borrowed certain assets. Approximately $80 million in COMP was distributed before the error was identified.

The instructive part is not the bug itself — software bugs are inevitable. The instructive part is what happened next. Under Compound's governance framework, fixing the error required a new governance proposal to pass, which took the standard seven-day voting period plus a two-day timelock. During those nine days, the buggy contract continued operating, and on-chain actors who understood what was happening extracted as many tokens as possible.

The incident exposed a fundamental tension in DAO design: the same mechanisms that make governance resistant to centralized control also make emergency response slow. Compound's 48-hour voting period and two-day timelock — designed to give token holders time to review and respond to proposals — became a nine-day window during which millions continued to be lost.

Several protocols subsequently implemented guardian roles — trusted multisig signers with the ability to pause contracts in emergencies — as a practical compromise. This is sensible, but it is explicitly a concession that fully decentralized governance cannot handle emergencies at the speed they require.

## MakerDAO: Governance Complexity and the Voter Participation Problem

MakerDAO's governance is among the most sophisticated in DeFi. The dual-token structure (MKR for governance, DAI for the stablecoin), the Maker Protocol Technical Foundation, the Core Units structure, and the subsequent transition to "Endgame" — all represent serious attempts to build durable, functional governance.

The persistent problem has been voter participation. Governance proposals at MakerDAO routinely pass with less than 10% of circulating MKR participating. Major votes determining risk parameters for the entire DAI system — collateralization ratios, stability fees, liquidation thresholds — have been decided by a handful of large MKR holders voting with concentrated positions.

The Curve Wars-era "governance maximalism" problem emerged: large holders of MKR (and analogously, large holders of governance tokens across DeFi) could exert outsized influence on protocol parameters in ways that benefited their other positions. A holder with both a large MKR stake and significant vault exposure has incentives to vote for lower collateralization ratios that increase their borrowing capacity — at the risk of the whole system.

This is not unique to MakerDAO. It is a structural feature of token-weighted voting in any protocol where governance token holders have financial positions that governance decisions directly affect.

MakerDAO's "Endgame" restructuring, proposed by founder Rune Christensen and substantially implemented through 2024-2025, attempted to address this through "SubDAOs" with specialized governance, an enhanced MKR lockup mechanism to align incentives with long-term protocol health, and a partial centralization of core protocol management. The trade-offs involved were explicitly acknowledged: Endgame reduced pure decentralization in exchange for better operational efficiency and incentive alignment.

## Uniswap: The Treasury and the Governance Cartel

Uniswap accumulated one of the largest treasuries in DeFi — over $3 billion in UNI tokens at various peak valuations — and for years could not effectively deploy it because governance proposals consistently failed to reach quorum or were blocked by a small number of large holders.

The 2023 Uniswap "fee switch" debate illustrated the dynamics. Enabling protocol fees — charging a small percentage of trading fees to go to the treasury rather than entirely to liquidity providers — was technically straightforward. The economic arguments for it were reasonable. But the vote failed repeatedly, partly because large UNI holders (venture capital firms and early investors with large positions) had conflicting interests: they benefited from the UNI treasury growing in nominal value but had complex off-chain arrangements with liquidity providers.

Andreessen Horowitz, which held a substantial UNI position, voted against the fee switch in 2023. Their rationale was published; it was thoughtful. But the incident highlighted that large VC investors with governance token positions vote in ways that reflect their overall portfolio interests, which may not align with what is best for the protocol or its users.

Uniswap's governance also revealed the "governance cartel" problem: the same handful of delegates — a16z, Gauntlet, Penn Blockchain, other institutional actors — appear repeatedly in close votes. Smaller token holders face rational apathy: the cost of researching proposals and voting exceeds the expected marginal impact of any individual vote when whale-scale positions dominate outcomes.

## Structural Problems and Attempted Solutions

The failures across these three protocols reveal consistent structural problems in current DAO governance:

**Token-weighted voting concentrates power.** Large holders have disproportionate influence, and the interests of large holders frequently diverge from the interests of small holders and protocol users. This is not decentralization; it is a different form of oligarchy.

**Low participation produces fragile legitimacy.** When major decisions pass with 5% of tokens voting, the claim of "community governance" becomes difficult to sustain. Low participation also makes protocols vulnerable to flash loan governance attacks — borrowing tokens briefly to acquire voting power for a single proposal.

**Emergency response requires centralization.** The Compound incident is not unique. Multiple protocols have implemented guardian multisigs, emergency pause functions, and timelocks with override mechanisms — all of which re-introduce some form of trusted authority.

**Governance is a full-time job.** Understanding the technical, economic, and risk implications of a complex protocol governance proposal requires expertise that most token holders do not have. Professional delegates, while improving decision quality, create a de facto professional governance class.

Attempted solutions include: conviction voting (accumulated voting power over time, reducing susceptibility to large token dumps), quadratic voting (reducing whale dominance at the cost of Sybil resistance), optimistic governance (proposals pass unless challenged within a window), and explicit separation of governance token holders from protocol users.

None of these solutions has fully resolved the core tension: decentralized governance is a collective action problem, and collective action problems are solved either by concentrating authority or by creating strong incentive alignment — neither of which maps neatly onto egalitarian token distribution.

## What the Failures Actually Mean

The right response to DAO governance failures is neither to abandon decentralized governance as inherently unworkable nor to declare any particular failure an existential crisis. These protocols are, despite their governance problems, still operating — managing billions in locked value with no central counterparty.

What the failures indicate is that DAO governance in its current form is best understood as a constraint mechanism — a way of making unilateral changes by any single actor very difficult — rather than as an efficient decision-making system. For protocol parameters that change rarely and benefit from conservative revision, this is useful. For operational decisions requiring speed, expertise, and ongoing management, it is a poor fit.

The most mature DeFi protocols are converging toward hybrid models: decentralized ownership combined with professional operational teams operating within DAO-approved frameworks. This is less philosophically pure than the original DAO vision. It is also more likely to produce durable, functional protocols.

The governance experiments of the past five years have generated real knowledge about what works and what does not. That knowledge is being incorporated into new protocol designs. The lesson is not that DAOs failed. It is that the first generation of DAO governance was less sophisticated than the problem requires.

DAO Governance Failures: What MakerDAO, Uniswap, and Compound Teach About Decentralization

// COMMENTS

ON THIS PAGE