How DAO governance works: votes, delegates, fee switches and why it keeps breaking

Who this is for: Investors, founders, delegates and policy staff who need to read a governance proposal, work out whether it can pass and what happens after it does, and judge whether a DAO’s structure protects the people who hold its token.

A decentralized autonomous organization is a way of running something, usually a DeFi protocol and the treasury attached to it, without a board or a CEO. Decisions are made by vote among holders of a governance token and carried out by code. The phrase promises more than most DAOs deliver, and the interesting story of the last few years is the gap between the ideal of leaderless coordination and the reality of a few large holders, professional delegates and foundations doing most of the work.

That gap matters more in 2026 because the money is now real. Uniswap governance voted to take a share of swap fees and burn UNI with it, Aave’s DAO has been buying back its own token since 2025, Optimism’s foundation spends half of sequencer revenue on OP repurchases, and Sky (the former MakerDAO) funds staking rewards from buybacks. A governance token that was once a voting chit is becoming a claim on protocol cash flow, which raises the stakes on every vote and sharpens every old question about who controls the votes. This guide explains how a proposal moves from forum to execution, who actually votes, how fee switches change the incentives, where governance has broken and what has been bolted on, how treasuries are run, and what courts have said about liability, closing with a worked example from Compound’s 2026 governance fight.

DAO governance by the numbers

What a DAO actually is in 2026

Strip away the branding and a large DAO has five parts: smart contracts (a lending market, an exchange, a staking protocol) whose parameters only a governance contract can change; a governance token whose holders propose and vote on those changes; a treasury of the DAO’s own token, stablecoins and ETH that the governance contract can spend; a foundation or development company that writes most of the code and most of the proposals; and, increasingly, a legal entity that can sign contracts and be sued. The word autonomous refers only to execution: once a vote passes, the contract carries it out without anyone’s signature. Everything before that point is written by people, debated on forums and voted on by delegates who are paid, in several DAOs, to show up.

The token is where the model differs most from a company. Shares carry a claim on cash flow and a vote, both enforced by corporate law. Most governance tokens launched between 2020 and 2024 carried only the vote, because a cash-flow claim invited securities regulators, and the supply schedule often left a large share with the foundation and investors, which is the root of the foundation-dominance problem discussed below. The fee switches of 2025 and 2026 have begun to close the gap between governance tokens and equity without resolving the legal question.

How a governance vote actually works

The mechanics are similar across the big protocols, most of which use some version of the Compound Governor contracts (Governor Bravo and its OpenZeppelin descendants). A proposal is a bundle of on-chain actions: change an interest rate parameter, spend treasury funds on a grant, upgrade a contract, switch on a fee. Governor Bravo caps a proposal at 10 actions, which is why Uniswap Labs split its July 2026 v4 fee activation into two on-chain votes.

1. Forum post. A request for comment goes up on the governance forum. Uniswap requires at least 7 days of discussion; Aave calls this stage an ARFC and Arbitrum a temperature check.
2. Off-chain signal. A Snapshot vote, free because nothing executes, tests sentiment. Uniswap’s temperature check runs 5 days and needs 10 million UNI in favour to proceed; Lido’s needs 1,000 LDO to open and runs 7 days.
3. On-chain proposal. The proposer must hold or be delegated a threshold: 1 million UNI at Uniswap, 1 million votable ARB at Arbitrum. Voting power is measured at a snapshot block taken when the proposal is created, so tokens bought or delegated afterwards do not count.
4. Voting window. Uniswap waits 2 days then votes for 7; Lido votes for 3 days with a 2-day objection period; Arbitrum votes for 14 to 16 days.
5. Quorum and majority. The proposal passes if it reaches quorum and more votes are for than against. Uniswap needs 40 million UNI in favour; Compound 400,000 COMP; Lido 5 percent of LDO supply; Arbitrum between 100 million and 300 million ARB for ordinary proposals and 150 million to 450 million for constitutional ones, scaled to delegated supply.
6. Timelock. A passed proposal queues before execution: 2 days at Uniswap, a 3-day minimum at Lido, and for Arbitrum constitutional changes 8 days on L2 plus a rollup challenge period plus 3 days on L1. The timelock exists so that anyone who disagrees has time to exit the protocol or organise a response.
7. Execution. Anyone can call execute once the timelock expires. The contract performs the actions exactly as encoded, which is why delegates increasingly audit the calldata rather than the forum summary.

Parameters as published in each DAO’s documentation or constitution, October 2026.

Shortcuts and fast lanes

Full votes are slow, so every large DAO has built faster paths for routine work. Lido’s Easy Track lets committee multisigs open motions that pass automatically after 72 hours unless holders of 0.5 percent of LDO object. Aave’s Risk Stewards move supply and borrow caps within pre-approved bands without a vote. Uniswap’s expedited process, used in 2026 for fee expansions to Robinhood Chain in July and Circle’s Arc chain in September, skips the forum stage for a 5-day Snapshot and an on-chain vote. Each fast lane trades some decentralisation for throughput and is a place where a captured committee could do damage, which is why scopes are narrow and revocable.

Delegation and who actually votes

Because most holders never vote, delegation is the norm. Holders assign their voting power to a delegate, who may be a venture fund, a governance service provider, a university blockchain club or an individual who has built a reputation in the forum. Delegates vote on every proposal; the holders behind them rarely pay attention. At Uniswap, a holder cannot vote or propose at all until they delegate, even to themselves.

The result is that quorum is usually reached by a handful of addresses. The clearest public data in 2026 comes from Compound, whose foundation published a full voting record in October. The May 2026 proposals creating a Treasury Management Committee (Proposal 580) and a $52 million two-year V4 programme (Proposal 582) each passed with about 1.88 million COMP in favour and none against, on a 400,000 COMP quorum. In September 2026 a single address holding 1.77 million COMP (attributed on the forum to the trader known as Humpy, voting with dCOMP backed by about $33 million of deposits) decided three outcomes: it carried Proposal 608 (about 1.8 million for, 984,000 against), defeated a security patch in Proposal 609 (627,000 for, 1.77 million against) and, with one other address, sank a service-provider renewal on Snapshot. With 10 million COMP in existence, votes of 2 to 3 million are turnout in the low twenties as a percentage of supply, and most of that sits with a few wallets.

Quorum thresholds elsewhere tell the same story. Uniswap’s 40 million UNI quorum is about 4 percent of the 1 billion UNI supply; Lido’s is 5 percent of LDO. Arbitrum measures quorum against delegated rather than total supply precisely because so much ARB is undelegated. Low thresholds are a concession to reality: set quorum at 20 percent of supply and nothing would ever pass.

Professional delegates have filled the gap. Optimism, Arbitrum, Aave and Compound all host delegate communication threads where firms such as PGov, SEEDGov and Avantgarde publish their reasoning, and several DAOs pay delegates for participation. That improves scrutiny and introduces a new conflict: a delegate paid by the DAO, or a service provider that is also a large delegate, votes on budgets that may include its own. Compound’s 2026 dispute turned partly on that overlap, with the Foundation, the Governance Working Group and security providers all supporting a programme that funded them.

The fee-switch era: tokens become cash-flow claims

For most of the 2020 to 2024 period, governance tokens carried no claim on protocol revenue. That changed in 2025 and 2026 across the largest DAOs, and it changes how governance should be analysed.

Uniswap: UNIfication

On November 10, 2025, Uniswap Labs and the Uniswap Foundation jointly published UNIfication. It switched on protocol fees for the first time: v2 pools give 0.05 percent of each swap to the protocol (LPs keep 0.25 percent rather than 0.30 percent), and v3 pools give a quarter of LP fees on the 0.01 and 0.05 percent tiers and a sixth on the 0.30 and 1 percent tiers. Fees flow through contracts called TokenJar and Firepit that buy and burn UNI, with a one-off burn of 100 million UNI from the treasury for fees forgone since launch. The proposal also folded the Foundation’s operations into Labs, gave Labs an annual growth budget of 20 million UNI in return for zero fees on its interface, wallet and API, and placed governance under a Wyoming DUNA called DUNI. Fees went live on Ethereum mainnet in late December 2025. By July 2026 Labs reported roughly 7.5 million UNI (about $25.6 million) burned, with monthly burn value rising from $3.1 million in February to about $5.1 million in June. Follow-on votes extended fees to eleven other chains, to v4 pools in July 2026 and to Arc in September 2026.

Aave: buybacks under a finance committee

Aave moved earlier and more cautiously. The Aavenomics implementation proposal of March 4, 2025, authored by Marc Zeller of the Aave Chan Initiative, created a four-member Aave Finance Committee (Chaos Labs, TokenLogic, LlamaRisk and ACI, acting on 3-of-4 signatures) and authorised AAVE buybacks of $1 million a week for six months, to be resized against the budget afterwards with quarterly reporting. The protocol then held about $115 million in reserves, and GHO supply of 186 million was generating roughly $12 million a year, half of it earmarked for a non-transferable Anti-GHO token paid mostly to staked AAVE. Aave shipped V4 on Ethereum mainnet on March 30, 2026 after an ARFC on March 13 and a Snapshot on March 23, and in October 2026 Aave Labs proposed a memberless Cayman foundation to hold the trademark, domains and IP that the DAO cannot legally own.

Optimism and Sky: buybacks from the top

Optimism’s version arrived in January 2026 as a Foundation proposal to spend 50 percent of Superchain sequencer revenue (about 5,868 ETH in the prior 12 months) on monthly OP repurchases for 12 months from February 2026, through an OTC provider because on-chain liquidity was judged too thin. Delegates forced the Foundation to split the buyback from a separate request for discretion over the remaining ETH, and the vote needed 60 percent Joint House approval. By August 2026 more than 9 million OP had been repurchased.

Sky, the rebranded MakerDAO, went furthest. Its Endgame plan replaced MKR with SKY at 24,000 SKY per MKR, made SKY the sole governance token, permanently disabled new emissions against a 23.46 billion cap, and applies a delayed upgrade penalty to unconverted MKR that began in September 2025 at 1 percent and rises 1 percentage point every three months (4 percent as of June 2026). SKY staking rewards (6.58 percent variable in October 2026) are funded by the Smart Burn Engine buying SKY with surplus from a stablecoin system of about 10.13 billion USDS and DAI backed by $17.27 billion of collateral. Day-to-day work is delegated to sub-DAOs called Stars, such as Spark, under a written constitution, the Sky Atlas.

Once a token has a cash-flow claim, the parameters that control it (the fee split, the buyback budget, who executes it) become the highest-stakes votes in the DAO, and they set the foundation that controls the treasury against holders who would prefer the cash. The Optimism debate over whether buybacks are self cancelling while emissions continue, and Arbitrum’s thread on turning ARB into a utility token (an ARB staking proposal passed but was never implemented, participants noted in November 2025) are both versions of that argument. The methods in how to read on-chain data let readers trace the flows themselves.

Where it breaks

Token voting is plutocratic by design: one token, one vote means the largest holders decide. Turnout is low, so quorum is often reached by a handful of delegates. That makes governance vulnerable to a few recurring failure modes that have each played out in public.

Whale capture

Whale capture is the simplest. In 2022 an investor acquired enough of the Build Finance DAO token to pass a proposal handing the treasury to himself. The 2026 Compound fight shows the subtler form. On September 27, 2026 a delegate alleged that the Foundation had converted 8.42 million DAI of protocol reserves, released to it under Proposal 536 for protocol operations only, into 344,780 COMP, delegated that COMP to its own voting address and returned the tokens to the multisig 58 minutes before the voting cutoff on Proposals 580 and 582. The Foundation’s statement of September 28 said no trading had occurred, the assets remained DAO-owned in a 4-of-6 multisig and nothing was spent on overhead; critics pointed to on-chain transfers to an exchange and asked for the EPCF and SVR reserve funds (about 5.1 million USDC and 356 ETH) to be returned to governance. Whether or not the outcome changed, the appearance of an insider buying the votes needed to guarantee passage is corrosive to the legitimacy the whole system depends on. The same saga then showed the mirror image: a single 1.77 million COMP address defeating a security patch.

Foundation dominance

Foundation dominance is the broader version of the same problem. Many protocols are steered in practice by a foundation or development company that drafts the proposals, controls the treasury multisig and holds or influences a large voting bloc. The DAO ratifies. Uniswap’s long argument over whether to switch on protocol fees, resolved only when Labs and the Foundation co-authored UNIfication, and Arbitrum’s 2023 dispute after its foundation moved treasury tokens before a ratification vote had concluded, are the reference cases. Even well-run foundations face the objection that the people writing the proposals are the people funded by them.

Apathy, bribery and complexity

Voter apathy, bribery markets that pay delegates to vote a certain way, and the complexity of proposals that few voters read all compound the above. UNIfication touched contract upgrades, token burns, legal entity formation and a service agreement in one document. The forum post that matters is often the one from a risk provider explaining what the calldata actually does.

The checks that have been added

The response to these failures has been a layer of constitutional machinery that looks increasingly like corporate governance.

• Timelocks and guardians. Every major DAO now delays execution. Compound added a Proposal Guardian that can cancel malicious proposals in August 2024 (Proposal 304, 1.4 million COMP for, none against) and extended it in February 2026 (Proposal 538).
• Security councils. Arbitrum’s 12-member Security Council, elected in two six-member cohorts with one cohort replaced every 12 months, acts on 9-of-12 signatures. It used its emergency powers on October 2, 2026, and in September 2026 published a review of its key exposure across three chains.
• Dual governance. Lido’s model, live on mainnet in 2025, gives the users of the protocol a veto over its token holders. If holders of more than 1 percent of stETH supply lock into a veto signalling contract, LDO decisions are blocked for 5 to 45 days depending on how much stETH joins. Above 10 percent a rage quit begins and governance pauses until the objecting stakers have withdrawn. Every proposal also faces a minimum 3-day delay. For a protocol holding a large share of staked ETH, this is meant to make capture of LDO pointless.
• Written constitutions. Arbitrum’s constitution sets quorum, timelocks and council powers in a document that only a constitutional vote can change; Sky’s Atlas divides authority into Governance, Support, Stability, Protocol and Accessibility scopes; Optimism formalised its protocol upgrade process in 2026.
• Two-house models. Optimism’s Token House (OP holders) and Citizens’ House (non-transferable citizenship) must both approve some decisions; special votes need 60 percent of the Joint House.
• Bounded delegation of routine work. Risk stewards, Easy Track and finance committees handle parameter changes and treasury operations within mandates, reported back to the forum monthly.

None of this is autonomous in the original sense. A security council is a board with narrower powers, a constitution is a charter, and a finance committee is a treasury function. The difference is that the record of every decision is public and the execution is not discretionary.

How DAO treasuries are managed

Treasuries were the original reason governance mattered and remain the hardest thing to run well. The common pattern through 2024 was a large pile of the DAO’s own token plus ad hoc grants. By 2026 the leaders had moved to mandated, reported programmes.

Arbitrum’s Treasury Management Council, operated by Entropy Advisors with monthly reports since July 2025, held about $28 million in tokenised Treasury products from WisdomTree, Spiko and Franklin Templeton (see tokenised real-world assets), about $4.9 million of USDC converted from 15 million ARB and split between three managers, 2,500 ETH deployed to Fluid and 5,000 ETH staked through Lido. It reported about $900,000 of interest from the RWA book by July 2025 and said conservative positioning kept it clear of losses in the 2026 Resolv USR and Kelp rsETH incidents.

Lido publishes DAO financials. Its first-half 2026 results showed net revenue of $15.9 million, a $4.5 million operating loss after a one-off Kelp-related expense, foundation spending of $14.3 million, and a treasury that fell from $157.5 million to $88.3 million over the half (mostly ETH price) before recovering to about $121.3 million by August 25, 2026. An LDO accumulation programme deployed 2,320 stETH and returned 14.8 million LDO. Optimism’s Foundation reported new OP commitments of about 150 million OP in Year 4 (May 2025 to April 2026), a third less than Year 3, with airdrops cut to zero and Retro Funding paused with 777.6 million OP held back. The trend is the same everywhere: fewer token grants, more yield-bearing stablecoin and ETH positions, and a committee that reports.

The legal question

An organization with members, a treasury and no legal entity is, in most jurisdictions, a general partnership, which means every member could be liable for its debts and actions. Two US cases made the risk concrete. The CFTC charged Ooki DAO in September 2022, alongside a $250,000 settlement with its predecessor bZeroX and its founders, on the theory that the DAO was an unincorporated association continuing the same illegal derivatives business; on June 9, 2023 a federal judge in California entered a default judgment holding the DAO was a person under the Commodity Exchange Act and imposing a $643,542 penalty. Then in November 2024, in Samuels v. Lido DAO (N.D. Cal., case 3:23-cv-06492), the court declined to dismiss a securities claim, finding Lido DAO was plausibly a general partnership under California law and that venture investors who participated in governance, including Paradigm, Andreessen Horowitz’s AH Capital and Dragonfly, could plausibly be partners, while dismissing a smaller investor that had not. Participation, not passive holding, was the dividing line in both cases.

The response has been legal wrappers. Wyoming created the DAO LLC on July 1, 2021 and, with Senate File 50 signed March 7, 2024 and effective July 1, 2024, the decentralized unincorporated nonprofit association, or DUNA, which gives a DAO legal personality, limited member liability, the ability to contract, bank and be served, and corporate tax treatment without disclosing members, while permitting for-profit activity and compensation to contributors. Uniswap proposed DUNI, a DUNA for its governance, in August 2025 and made it the legal backbone of UNIfication; the Foundation also led a 20-signatory letter to Treasury and the IRS in September 2025 asking for clear tax pathways for DAOs. The Cayman foundation company remains common: Aave Labs proposed a memberless one in October 2026 to hold IP the DAO cannot own, with directors appointed only by AIP, and a delegate immediately asked whether a Swiss association would be better. Every wrapper trades member protection for an entity with directors who can be served.

Governance tokens themselves sit in an uncertain place. Whether a token that confers voting rights and a claim on protocol revenue is a security has been argued for years; the SEC’s 2025 change of posture, including the dismissal of most of its pending crypto enforcement cases, reduced the immediate pressure without resolving the question in law, and the fee switches of 2025 and 2026 make the revenue claim harder to deny. The CLARITY Act, which passed the House in July 2025, proposes a test for when a network is decentralized enough that its token is a commodity; Uniswap’s DUNI documents say the DUNA was designed with that legislation in mind. For the agency split, see SEC vs CFTC crypto regulation; for Europe, which mostly looks through to the entity operating a protocol, see what is MiCA.

How we got here: a timeline

Worked example: Compound Proposal 582, forum to execution

How to evaluate a DAO’s governance: a checklist

• Who holds the votes? Pull the top 20 delegates and their share of delegated supply. If one wallet exceeds quorum on its own, as at Compound in 2026, every vote is contingent on that wallet.
• Is quorum measured against total or delegated supply? Arbitrum’s 40 to 50 percent of delegated supply is a stricter test than Uniswap’s 4 percent of total supply; they produce very different thresholds for capture.
• How long is the timelock, and who can cancel? Two days is the floor. A guardian or security council is a safety feature only if its members are elected, term-limited and accountable, as Arbitrum’s annual cohort elections are.
• Does the foundation vote, and with whose tokens? The best answer is that it does not vote protocol-owned tokens at all; the worst is that it acquires and delegates them around snapshots.
• Who writes the proposals that fund the writer? Service providers proposing their own renewals and foundations proposing their own budgets should face review from a delegate or risk provider with no stake in the outcome.
• Where does the revenue go? Trace fees to burns (Uniswap), committee-run buybacks (Aave), a foundation’s OTC desk (Optimism) or staking rewards (Sky). Each has different transparency and capture risk.
• Is the treasury under a mandate with reporting? Monthly reports, named managers and stated risk limits, as at Arbitrum’s ATM Council and Lido’s half-year financials, are the standard. Ad hoc grant votes are a red flag.
• What is the legal wrapper? A DUNA, a Cayman foundation or a Swiss association each limits liability differently. No wrapper means the Ooki and Lido precedents apply to anyone who participates in governance.

Risks and open questions

The first risk is that fee switches settle the economic question and sharpen the legal one. A token burned with protocol revenue, or whose holders receive buyback-funded rewards, looks more like a profit interest than a 2021 governance token did. The 2025 enforcement retreat lowered the odds of an SEC action, but private suits like the one against Lido DAO do not depend on the agency’s mood, and the general partnership theory means active delegates, not just founders, could be defendants if a wrapper is absent or defective.

The second is that the fixes concentrate power in new places. Security councils, finance committees, risk stewards and expedited processes all move decisions out of the general vote. Arbitrum’s September 2026 review of its Security Council’s key exposure is a reminder that a 9-of-12 multisig is only as safe as its signers, and Optimism’s delegates objected to Foundation discretion over half of Superchain revenue for the same reason.

The third is that voting power is now borrowable. The Compound address that swung three votes in September 2026 used dCOMP, a wrapped position that let it vote while borrowing against the collateral. Where governance tokens can be borrowed or wrapped, the cost of a decisive vote can be far below its market value. Dual governance and long timelocks mitigate this; nothing eliminates it. Finally, the DUNA is new law without case law: no court has tested whether it shields a holder sued in California, how it interacts with a Cayman foundation holding the same protocol’s IP, or how the IRS treats distributions. Uniswap’s DUNI is the largest live experiment.

What to watch next

Glossary

Governance token

A token that gives its holder the right to vote on proposals affecting a protocol or its treasury, and increasingly a claim on protocol revenue through burns or buybacks.

Delegate

An address that votes with voting power assigned to it by other holders. Delegation does not move or lock the tokens.

Proposal threshold

The minimum voting power an address must hold or be delegated to submit an on-chain proposal, such as 1 million UNI.

Snapshot (block)

The block at which voting power is measured for a proposal; tokens acquired or delegated afterwards cannot vote on it.

Quorum

The minimum voting power that must be cast for a proposal to be valid, expressed as a token amount or a share of total or delegated supply.

Timelock

A mandatory delay between a proposal passing and its execution, typically two or three days, that gives dissenters time to exit or respond.

Governor Bravo

The Compound governance contract design, with a proposal threshold, voting period, quorum and timelock, copied by most DeFi DAOs.

Security council

An elected multisig with narrow emergency powers, such as Arbitrum’s 12-member council acting on 9-of-12 signatures.

Dual governance

Lido’s mechanism allowing stETH holders to delay LDO decisions at 1 percent of stETH supply and trigger a rage quit at 10 percent.

Fee switch

A governance-controlled parameter that diverts part of a protocol’s fees from liquidity providers or users to the protocol or its token holders.

DUNA

Wyoming’s decentralized unincorporated nonprofit association, a legal entity form for DAOs that limits member liability and allows the DAO to contract and pay tax.

Optimistic governance

A process in which a motion passes automatically after a delay unless enough holders object, as in Lido’s Easy Track.

Why it matters

DAO governance started as an experiment in replacing managers with code. What it has become, a decade on, is a public and auditable version of the things companies already do: boards with limited powers, written charters, finance committees, buybacks and legal entities. The code enforces the outcome of the vote; everything that decides the vote is human, concentrated and increasingly professional. The 2026 Compound dispute, the Uniswap fee switch and the Lido partnership ruling are three views of the same fact: once a DAO controls real cash flow, the people who can swing a vote and the legal shell they stand inside matter at least as much as the smart contracts.

The practical conclusion is to read governance the way an analyst reads a shareholder register and a charter: who has the votes, what can they do with them, how fast, and who answers if it goes wrong. The guide to Ethereum Layer 2 networks, whose DAOs appear throughout this piece, and the CCS interview archive are the natural next reads.

Sources

1. Uniswap Docs: Governance process, accessed October 6, 2026
2. Uniswap Labs: UNIfication, November 10, 2025
3. Uniswap Governance: Temp Check, Activate v4 Protocol Fees, July 7, 2026
4. Uniswap Governance: Temp Check, Protocol Fee Expansion: Arc, September 18, 2026
5. Uniswap Foundation: DUNI, an IRL entity for Uniswap Governance, August 11, 2025
6. a16z crypto: The DUNA, an Oasis for DAOs, March 2024
7. CFTC: Charges against bZeroX, its founders and Ooki DAO, September 22, 2022
8. CFTC: Federal court orders Ooki DAO to pay penalty and close operations, June 9, 2023
9. US Courts via GovInfo: Samuels v. Lido DAO et al, N.D. Cal. 3:23-cv-06492, accessed October 6, 2026
10. Lido: Governance (voting, Easy Track, Dual Governance), accessed October 6, 2026
11. Lido: Poolside Recap, Tokenholder Update, August 2026, August 2026
12. Aave Governance: ARFC Aavenomics Implementation, Part One, March 4, 2025
13. Aave Governance: ARFC Aave V4 Activation on Ethereum Mainnet, March 13, 2026
14. Aave Governance: ARFC The Aave Foundation, Phase 1, October 2, 2026
15. Arbitrum Foundation: The Constitution of the Arbitrum DAO, accessed October 6, 2026
16. Arbitrum Forum: ATM Council Updates, July 2025 to September 2026
17. Arbitrum Forum: From Governance Token to Utility Token, November 6, 2025
18. Optimism Governance: Proposal to Align the OP Token with Superchain Success, January 7, 2026
19. Optimism Governance: Collective Year 4 Budget Update and Year 5 Outlook, August 7, 2026
20. Compound Forum: Recent governance activity in context, the record since 2024, October 2, 2026
21. Compound Forum: Misappropriation of the v2 DAI reserves by the Compound Foundation, September 27, 2026
22. Compound Forum: Statement on the use of V2 reserves, September 28, 2026
23. Sky: Understanding the SKY Token, August 18, 2026
24. Wikipedia: Decentralized autonomous organization (Wyoming 2021, Build Finance 2022), accessed October 6, 2026

Disclosure: This guide is for education only and is not investment, legal or tax advice.