DAO governance is supposed to be decentralized decision-making. Thousands of token holders, hundreds of delegates, collective intelligence in action. That's the promise.

The reality looks different.

We just shipped Delegate Vote Quality scoring at ChainSights, analyzing every vote cast by thousands of delegates across 24 DAOs. The headline finding: more than 75% of all delegates score below 5 out of 10 on vote quality.

That doesn't mean they're not voting. Most of them are. That's precisely the problem.

Participation Is Not Governance

Every major DAO dashboard tracks participation rates. How many delegates voted? What percentage of voting power showed up? These numbers look healthy across most DAOs — 60%, 70%, sometimes higher.

But participation rate answers the wrong question. It tells you who showed up. It doesn't tell you how well they governed.

A delegate who reads every proposal, participates in forum discussions, and votes after careful deliberation gets the same participation score as someone who opens Snapshot, clicks "For" on everything in 30 seconds, and closes the tab.

In fact, the speed-clicker often scores higher — because they never miss a vote.

Four Signals That Measure What Matters

We built four signals to distinguish informed voting from checkbox behavior:

Deliberation measures the time between consecutive votes. A delegate who votes on five proposals within two minutes is speed-clicking, not deliberating. A delegate who spaces votes across hours or days is more likely to have actually read the proposals.

Independence looks at vote diversity. If a delegate votes "For" on 95% of all proposals, that's not independent judgment — that's rubber-stamping. Genuine governance requires saying "No" sometimes.

Focus tracks category concentration. Does a delegate vote across every category — treasury, protocol upgrades, grants, partnerships — or do they focus on areas where they have expertise? Domain experts who vote within their competence tend to make better decisions than generalists who vote on everything.

Originality compares voting patterns against the largest token holders. If a delegate consistently votes the same way as the top three whales, they may not be exercising independent judgment. They may just be following the leader.

Each delegate gets a composite Vote Quality Score (VQS) from 0 to 10.

What the Data Shows

Across thousands of delegates and 24 DAOs, the distribution tells a clear story:

Median VQS: 3.9 — the typical delegate scores below 4 out of 10
75th percentile: 4.9 — even the "above average" delegates barely crack 5
Only 4.4% score 8 or above — truly high-quality governance participation is rare

The variation between DAOs is striking. We identified three distinct archetypes:

Consensus DAOs

DAOs like Uniswap (median 3.6), ENS (3.2), and Lido (1.8) show low Independence and Originality scores across almost all delegates. This isn't necessarily bad governance — many of their proposals are uncontroversial operational decisions where consensus is expected. But it does mean the voting process adds little beyond a rubber stamp.

At Uniswap, we analyzed 94 delegates. The highest individual VQS was 3.3 out of 10. Not a single delegate scored above the midpoint.

Mixed DAOs

Arbitrum (4.2), Balancer (4.4), and Superfluid (4.4) show a broader spread. Some delegates think independently, others follow the crowd. These DAOs have a mix of contested and routine proposals, which naturally creates more diversity in voting patterns.

Pluralistic DAOs

CVX (7.8), Aavegotchi (7.1), and Giveth (5.2) show the highest vote quality. Their communities genuinely disagree on governance decisions, and delegates exercise independent judgment. These DAOs have the healthiest governance dynamics — not because everyone agrees, but because disagreement is normalized.

Why This Matters

DAOs collectively manage billions of dollars in treasury funds. The governance processes that control these funds are supposed to provide checks and balances. But if 75% of delegates are essentially rubber-stamping decisions, those checks don't exist in practice.

This creates real risks:

Governance capture becomes easier when most delegates vote predictably. A coordinated group only needs to influence the few independent voters to control outcomes.

Delegation fails its purpose. Token holders delegate to experts so they don't have to vote on everything themselves. If those experts aren't actually deliberating, the entire delegation model breaks down.

Participation metrics mislead. A DAO reporting "85% delegate participation" sounds healthy. But if most of that participation is checkbox voting, the real governance engagement might be closer to 15%.

What DAOs Can Do

The data points to several structural improvements:

Delegate accountability frameworks. Require delegates to explain their votes, especially when they vote against proposals. Public reasoning creates accountability and encourages deliberation.

Voting power decay. If a delegate consistently shows low vote quality, their delegated power could gradually redistribute to more engaged delegates. This creates incentives for quality, not just quantity.

Discussion requirements. Some DAOs are experimenting with mandatory forum participation before voting. If you haven't engaged with the proposal discussion, your vote counts for less. This directly targets the checkbox voting problem.

Quadratic voting. Reducing the outsized influence of large token holders makes every vote matter more, which in turn incentivizes more thoughtful participation from all voters.

See Your DAO's Scores

Every data point in this analysis is live and free on ChainSights. Pick any of the 47 tracked DAOs, click into the Delegate Vote Quality section, and see how individual delegates score across all four signals.

No login. No paywall. No email required.

chainsights.one/matrix

Because participation is easy. Governance is hard. We measure the difference.

ChainSights provides identity-first Web3 analytics. We help DAOs understand who actually controls their governance — not just which wallets voted.

Wallets lie. We don't.