DAOs and governance tokens

Many DeFi protocols, layer-2 networks and token projects are run, at least formally, by token holders who vote on upgrades, fees and treasury spending. For a lender, investor, auditor or counterparty, the governance setup answers practical questions: who can change the contract I depend on, how fast, and who controls the money.

What a DAO is in practice

A decentralized autonomous organization (DAO) is a group that coordinates through smart contracts and token-weighted votes to control a protocol, a treasury, or both. The label covers very different arrangements:

• On-chain binding governance. A governor contract counts votes and, if a proposal passes, executes its code directly, for example changing an interest-rate model or sending funds from the treasury. Compound's Governor Bravo is the classic design that many protocols copied.
• Off-chain signaling. Holders vote on an off-chain platform such as Snapshot, where signed messages cost no gas. The result is carried out by a multisig whose signers promise to follow it. Cheap and flexible, but execution depends on trusting the signers.
• Hybrid. Off-chain "temperature checks" first, then an on-chain vote for anything that moves funds or code.

Typical decisions include protocol parameters (collateral factors, fees), contract upgrades, listings of new assets, grants and budgets, and pay for delegates and service providers.

Token voting mechanics

Most DAOs use one token, one vote. Voting power is read at a snapshot block, usually the block when the proposal was created or shortly after, so that tokens bought or borrowed after that point carry no votes on that proposal. In Compound-style tokens, votes count only if they have been delegated, either to the holder's own address or to someone else. Delegation passes voting power without moving the tokens, which lets passive holders lend their weight to active delegates.

A governor contract enforces a set of parameters. Compound's widely copied Governor Bravo setup illustrates them (in February 2025 Compound replaced the contract with an OpenZeppelin-based governor whose parameters can be changed by vote):

Other DAOs choose different values. Uniswap's governor, for instance, requires a quorum of 40 million UNI in favor. Parameters can themselves be changed by vote, so a professional should read the current values on-chain rather than rely on documentation.

The proposal lifecycle and timelocks

A well-run change typically passes through five stages: forum discussion, an off-chain temperature check, an on-chain proposal, the voting period, and a timelock queue before execution.

The timelock is the most important safety feature in the chain. During the delay, users who dislike the change can withdraw their funds, and a guardian or security council (if one exists) can cancel a malicious proposal. A protocol with a 2-day timelock gives depositors 48 hours of warning; one that can execute instantly gives none.

Several variations try to fix the weaknesses of plain token voting:

• Optimistic governance: a designated party proposes routine actions that pass automatically unless enough holders object within a window. Faster, but relies on holders watching.
• Vote-escrow models: holders lock tokens for a period to receive voting power that grows with lock length. Curve's veCRV, with locks of up to four years, is the best-known example; it rewards long-term holders but creates new markets for votes (see the lesson Governance attacks and failure modes).
• Bicameral designs: Optimism splits power between a Token House (token holders) and a Citizens' House (identity-based, one person one vote) so that tokens alone do not decide everything.

Multisig councils and delegated authority

Even "fully on-chain" DAOs usually keep a small group with special powers, because a 7-day vote is too slow to respond to an exploit. Security councils and guardian multisigs can pause contracts, cancel queued proposals or push emergency upgrades. Arbitrum's Security Council, for example, has 12 members and needs 9 of 12 signatures for emergency actions.

These councils are a necessary compromise, but they concentrate power. A professional should establish:

• who the signers are, whether they are independent of the founding team, and how they are elected or removed;
• the signing threshold and how signers store their keys (see the module on custody and key management);
• exactly what the council can do without a token vote, and whether its actions are themselves subject to a delay;
• for off-chain-vote DAOs, whether the multisig has ever deviated from a vote result.

Treasury management

DAO treasuries can be large, but their headline value is often misleading because much of it is held in the DAO's own token. Consider a DAO reporting a USD 200 million treasury: USD 180 million in its native token and USD 20 million in stablecoins, with annual spending of USD 15 million. If the token falls 60%, the treasury is worth USD 92 million (72 + 20). Selling native tokens to fund operations would push the price down further, so the realistic runway is the stablecoin portion: about 16 months.

Headline figures on treasury dashboards can mislead in other ways too. They may include tokens still locked in vesting contracts, tokens already promised as grants or incentives, and positions in DeFi protocols valued at prices that could not be realized in a sale. A professional should rebuild the figure from on-chain balances, subtract commitments, and value native tokens with a liquidity discount.

Sound treasury practice for a DAO resembles corporate treasury policy:

Voter apathy and concentration

Token voting rarely looks like broad democracy. Most holders never vote, many tokens sit on exchanges or in DeFi contracts, and the tokens that do vote are often concentrated among founders, early investors, foundations and a few delegates. In July 2024, Compound proposal 289, which would have moved about USD 24 million of COMP out of the treasury, passed with only 57 addresses voting (the case is covered in the lesson Governance attacks and failure modes).

Low turnout has two effects. Quorums become hard to reach for ordinary proposals, and a determined minority can pass contested proposals when others are not watching. Delegation helps with turnout but concentrates power in a small number of delegates, some of whom are paid by the DAO itself.

DAOs have tried several remedies. Delegate programs pay recognized delegates a stipend tied to participation and published voting rationales. Some DAOs lower quorums to keep routine business moving, which also lowers the cost of capture. Others move routine parameter changes to optimistic processes and reserve full votes for treasury and upgrade decisions. None of these removes the underlying fact that voting power follows token holdings, and token holdings can be bought or borrowed.

Useful metrics when assessing a DAO:

• turnout as a share of circulating supply, and how often quorum is reached;
• the share of votes held by the top 5 and top 10 voters or delegates;
• the minimum number of voters needed to reach quorum (a voting "Nakamoto coefficient");
• tokens held by the team, investors and foundation, and their unlock schedule;
• whether exchanges or custodians vote with client tokens.

Governance participation can also carry legal consequences for the voters themselves, a point developed in the next lesson.

Exam focus: on-chain binding vs off-chain signaling (Snapshot plus multisig) vs hybrid governance; snapshot blocks and delegation; governor parameters (proposal threshold, voting delay, voting period, quorum, timelock) and what each protects against; why the timelock matters to depositors; optimistic, vote-escrow (veCRV) and bicameral (Optimism) variants; security councils and what to check about them (signers, threshold, scope, delay); treasury runway when holdings are mostly the native token; turnout and concentration metrics.