DAO: Decentralized Governance

In April 2016, Slock.it launched The DAO - an investment fund governed by code. Within 28 days, 20,000 people put $150M into a smart contract that was supposed to replace a board of directors with blockchain voting. Two months later, a hacker found a vulnerability and drained $60M. The Ethereum community voted for a hard fork to recover the funds - splitting the blockchain into Ethereum and Ethereum Classic. The first DAO failed, but the idea did not die: by 2025 more than $30B is held in DAOs, and protocols with billions in TVL are governed by votes in which less than 5% of token holders participate. How does governance without a CEO work - and why does it underperform its promises yet outperform expectations?

• **Uniswap Governance** controls a $3B+ treasury and sets parameters for a protocol with $5B TVL - yet the typical turnout is 5%, and the top 10 delegates control more than 50% of active votes
• **Gitcoin Grants** distributed $60M+ through quadratic funding - a mechanism where 1000 donations of $1 each give a project more matching than 1 donation of $1000, equalizing the influence of whales and regular users
• **Optimism RPGF** created bicameral governance: Token House (token-weighted) for protocol parameters + Citizens' House (SBT-based, 1 person = 1 vote) for funding public goods - the most ambitious experiment in on-chain democracy

Предварительные знания

• Standards: ERC-20, ERC-721, ERC-1155

Voting Mechanisms: How On-Chain Voting Works

In a traditional company, decisions are made by a board of directors - 5-15 people elected by shareholders. The process is opaque: meeting minutes are published with a delay, votes happen behind closed doors, and minority shareholders learn about decisions after the fact. **DAO (Decentralized Autonomous Organization)** moves the entire governance process to the blockchain: every proposal, every vote, every execution is a transaction anyone can verify.

The core of on-chain governance is the **Governor contract** (OpenZeppelin standard). It is a smart contract that manages the full lifecycle of a proposal: from creation to execution. Governor does not hold funds itself - it controls a **Timelock contract**, which is the actual owner of the treasury and has the rights to upgrade other protocol contracts.

**Timelock** is a critical security element. Between a proposal's approval and its execution, 24-48 hours pass. During this time, users who disagree with the decision can withdraw their funds from the protocol. If proposals were executed instantly, an attacker with enough tokens could drain the treasury in a single block.

**Quorum** is the minimum participation threshold. In Compound, the quorum is 400,000 COMP (~4% of total supply). If a proposal receives 399,999 COMP votes - it is rejected, even if 100% of votes were "for". Quorum protects against situations where 3 people holding 100 tokens make decisions affecting billions of dollars.

**Snapshot - gasless voting.** On-chain voting costs gas (~$5-20 per transaction). Snapshot solves this: votes are signed off-chain via EIP-712 typed signatures and stored on IPFS. The result is not binding - the protocol trusts a multisig to execute the community's will. More than 30,000 DAOs use Snapshot because gasless voting increases participation from 3-5% to 15-20% of holders.

Governance Tokens: 1 Token = 1 Vote and the Plutocracy Problem

Who votes in a DAO? Holders of **governance tokens** - special ERC-20 tokens that grant voting rights. Most protocols follow the principle of **1 token = 1 vote**: whoever holds more tokens has more influence. UNI (Uniswap), COMP (Compound), AAVE (Aave) are the governance tokens of the largest DeFi protocols.

Governance tokens differ from regular ERC-20 tokens. They implement the **ERC-20Votes** extension (OpenZeppelin), which adds checkpointing - saving balances at each block. When a proposal is created, the Governor fixes a snapshot block. Votes are counted by the balance **at the time of the snapshot**, not the current balance. This protects against the "buy tokens → vote → sell" attack.

**Voter apathy** is the main governance problem. In Compound, the average turnout is 3-5% of total supply. Reasons: gas costs $5-20 per vote, proposals are technically complex (changing an interest rate parameter from 0.035 to 0.038 per block), and small holders have no motivation - their 100 tokens will not affect the outcome with a quorum of 400,000.

**Whale dominance** - plutocracy in action. In MakerDAO, 4 addresses (Andreessen Horowitz, Paradigm, and two other funds) control enough MKR to unilaterally reach quorum on any proposal. In practice this means that the "decentralized governance" of a protocol with $8B TVL depends on decisions by 4 venture capital funds from San Francisco.

**Alternative: reputation-based governance.** Instead of purchasable tokens - reputation that cannot be transferred. **Soulbound Tokens (SBT)** are non-transferable NFTs proposed by Vitalik Buterin in 2022. An SBT is permanently bound to an address: it cannot be bought, sold, or transferred. Governance via SBT: a vote depends on contributions to the ecosystem (code, participation in discussions, testing), not on wallet size.

**Optimism RPGF (Retroactive Public Goods Funding)** is a practical example of non-plutocratic governance. Funding recipients are determined not through token voting, but through the **Citizens' House** - a chamber where holders of non-transferable SBT (Citizen badges) vote. The vote does not depend on wealth: each citizen = 1 vote, and that vote cannot be bought or sold.

Quadratic Voting: One Person, One Vote in a Token World

Token-weighted voting (1 token = 1 vote) creates plutocracy: whoever has more money has more power. Democracy (1 person = 1 vote) requires identification - but on the blockchain a single person can create a million addresses. **Quadratic Voting (QV)** is an elegant compromise between these extremes: the cost of votes grows quadratically. The first vote costs 1 token, but casting 10 votes requires 100 tokens.

**Gitcoin Grants** is the most successful application of a quadratic mechanism. Gitcoin uses not QV but **Quadratic Funding (QF)**: the matching fund allocated to a project depends not on the sum of donations, but on the **number of unique donors**. If 1000 people donate $1 each, the project receives more matching than if 1 person donates $1000. Across rounds from 2020-2024, Gitcoin distributed more than $60M via QF.

**Sybil attack - the Achilles' heel of QV.** If one vote is cheap but costs grow quadratically - it is more profitable to split 100 tokens across 100 addresses (100 votes at 1 token each) instead of one address (10 votes for 100 tokens). QV only works with **Sybil resistance** - the guarantee that one address = one person.

**Gitcoin Passport and WorldID** are two approaches to Sybil resistance. Gitcoin Passport collects "proof of humanity": linking Twitter, GitHub, ENS, Google - each action adds to a score. WorldID (Worldcoin project) uses iris scanning with a special device (Orb), generating a zero-knowledge proof: "I am a unique human" without revealing identity. By 2025, WorldID had verified more than 6 million people.

**Conviction Voting** is an alternative mechanism where the strength of a vote grows over time. The longer you support a proposal, the stronger your vote - as if conviction were converted into influence. A proposal passes not at a specific moment, but when the accumulated conviction exceeds a threshold. This eliminates fixed voting periods and protects against "last-minute" attacks - buying votes right before a deadline.

**Futarchy** is a radical idea from economist Robin Hanson: "vote on values, bet on beliefs". Instead of voting on specific decisions, participants trade prediction market tokens: "token price of the protocol IF decision A is adopted" vs "price IF decision B is adopted". The market selects the decision that maximizes value. MetaDAO on Solana is the first implementation of futarchy, where proposals are adopted based on market prices.

Delegation: Representative Democracy on the Blockchain

Voter apathy is the main enemy of DAOs. Typical turnout: 3-8% of token supply. The reasons are clear: understanding a technical proposal about changing the `reserveFactor` parameter from 0.25 to 0.30 in Compound requires deep DeFi knowledge, and the reward for voting correctly is zero. **Vote delegation** addresses this problem by bringing representative democracy to the blockchain: instead of voting yourself, you delegate your voting power to a **delegate** - someone you trust.

Delegation in ERC-20Votes works through the `delegate(address)` function. Your tokens stay in your wallet - you can sell them at any time. But voting power transfers to the delegate. Importantly: **by default you cannot vote**, even if you hold governance tokens. You must first delegate - even just to yourself (`delegate(msg.sender)`).

**Delegate platforms** have built infrastructure around delegation. **Agora** shows delegate profiles with voting history, reasoning for each vote, and on-chain statistics. **Karma** calculates reputation scores for delegates based on: participation rate (% of votes cast), forum activity, and proposal creation. A delegate with a 95% participation rate and detailed reasoning for each vote is more valuable than one who votes once every six months.

**Professional delegates** are a new role in the crypto ecosystem. Organizations like StableLab, Flipside, and independent delegates (such as @pennblockchain) accumulate delegation from thousands of addresses, building significant voting power. Some protocols (Optimism, Arbitrum) pay compensation to active delegates - from $2,000 to $20,000 per month. This creates a class of "professional governors" that begins to resemble a board of directors.

**Principal-agent problem** - a delegate may vote against the interests of delegators. In ENS DAO, one large delegate (with >1M delegated ENS) voted against returning funds from the treasury, causing a scandal - delegators could not retroactively revoke the vote, only cancel delegation for future votes. Unlike parliamentary elections (every 4 years), in a DAO delegation can be revoked instantly - but in practice 95% of delegators never check how their delegate votes.

**Optimism RPGF** is the most ambitious governance experiment. Optimism created a bicameral system: **Token House** (UNI-style, token-weighted) handles protocol parameters and grants, while **Citizens' House** (SBT-based, 1 citizen = 1 vote) distributes retroactive funding for public goods. Across 3 RPGF rounds, $80M+ was distributed among hundreds of projects - from Ethereum client developers to educational content creators.

**ENS DAO** - a lesson in real governance. In November 2021, ENS airdropped governance tokens. In the first year: 40+ proposals, 10,000+ unique voters, treasury >$1B. But in 2023 a crisis erupted: the delegate with the largest delegation demanded the firing of a developer over tweets. The vote showed that "decentralized governance" can be just as politicized and chaotic as traditional organizations.

Key Takeaways

• **The Governor contract** manages the proposal lifecycle: Pending → Active → Succeeded → Queued → Executed. Timelock (24-48h) protects against instant execution; quorum protects against decisions being made with low participation. Snapshot enables gasless off-chain voting
• **1 token = 1 vote** creates plutocracy: top-10 addresses control 40-60% of votes, turnout is 3-8%, and governance attacks via flash loans are real. Snapshot on the previous block and ERC-20Votes checkpointing are the key protections
• **Quadratic Voting** (cost = n²) mathematically levels influence: a whale with 1M tokens gets 1,000 votes, while 1,000 users with 1K tokens get 31,623 votes. But QV is vulnerable to Sybil attacks without proof of personhood (Gitcoin Passport, WorldID)
• **Vote delegation** raises turnout from 5% to 30-35%, but concentrates power among 50-100 professional delegates - a trade-off between participation and decentralization. Liquid democracy, Agora, and Karma are the infrastructure of a new governance class
• The first DAO in 2016 failed due to a vulnerability and split Ethereum - but the idea survived the failure. By 2025, $30B+ in DAOs, bicameral systems (Optimism), SBT-based voting, and quadratic funding show that on-chain governance is evolving - from naive plutocracy toward more complex models of democracy

Related Topics

DAO connects token standards with economic models and legal regulation:

• ERC Standards — Governance tokens are ERC-20 with the ERC-20Votes extension (checkpointing). Soulbound Tokens (SBT) for reputation-based governance use non-transferable ERC-721. LP tokens from DeFi pools are often delegated in governance
• Game Theory in Blockchain — Voter apathy, whale dominance, and governance attacks are game theory problems. Quadratic voting changes the payoff matrix, making Sybil attacks a rational strategy. Conviction voting adds a temporal dimension to strategic voting
• Tokenomics — Governance token design directly determines DAO health: team vesting schedule, airdrop distribution, inflation rate - all affect vote concentration and voter apathy
• Regulation and Law — DAOs have no legal status in most jurisdictions. Wyoming (2021) was the first to recognize DAOs as LLCs. Governance tokens may be classified as securities (SEC vs LBRY, 2023). Legal uncertainty is the main barrier to institutional participation

Вопросы для размышления

• The DAO in 2016 failed, but the Ethereum community voted for a hard fork to recover the money - effectively reversing blockchain "immutability". Was that decision correct? What if the amount had been not $60M but $600M?
• Quadratic voting mathematically levels influence, but requires Sybil resistance (proof that one address = one person). Which identification system would you choose - and are you willing to sacrifice anonymity for fair voting?
• Professional delegates in DAOs earn $2,000-$20,000 per month and control billions in voting power. How do they differ from traditional lawmakers or board members? Has DAO solved the problem of centralized governance - or just renamed it?

Связанные уроки

• gt-02