The Lido DAO governs Lido, the largest liquid-staking protocol on Ethereum. Users deposit ETH and receive stETH, a transferable token that accrues staking rewards while remaining usable across DeFi — so a staker keeps liquidity instead of locking ETH in a validator. The protocol is run by holders of the LDO governance token, who set the node-operator set, protocol parameters, and the fee — currently 10% of staking rewards, split between node operators and the DAO treasury. Because Lido stakes a very large share of all ETH, its governance is not just a protocol matter but a live question about the decentralization of Ethereum itself. (Lido docs)

stETH and the liquid-staking model

Solo staking on Ethereum requires 32 ETH, dedicated hardware, and a locked position. Lido removes all three constraints: any amount of ETH can be deposited, the protocol routes it to a curated set of professional node operators, and the depositor receives stETH that rebases daily with rewards and can be traded, lent, or used as collateral. That composability is why stETH became one of the most widely integrated assets in DeFi — and why it anchors the collateral base of protocols like Aave.

The trade-off is trust in Lido's operator set and smart contracts. Historically the operators were a DAO-curated whitelist; to widen participation Lido added the permissionless Community Staking Module (CSM), letting independent operators join with a bond rather than a governance vote. Node-operator selection, module parameters, and withdrawal mechanics are all controlled by LDO governance. (docs.lido.fi)

The centralization debate

Lido's success created the problem it is best known for. Because it pools stake from many users behind one protocol, Lido at times controlled roughly a third of all staked ETH — close to the one-third threshold at which a single actor can affect Ethereum's finality, and well within sight of the two-thirds level that matters for consensus safety. Critics argued that a single liquid-staking protocol approaching these thresholds undermines the very decentralization staking is meant to secure; a widely debated 2021 research-forum proposal even asked whether Lido should self-limit its share of the network. Lido's answer has been to decentralize within the protocol — more independent operators, the CSM, distributed-validator technology — rather than cap deposits. Rivals such as Rocket Pool make the opposite structural bet — a permissionless node-operator set from the start, at the cost of the curated professionalism Lido relies on. The debate remains one of the sharpest governance questions in Ethereum. (Lido research forum; Lido blog)

Dual Governance — giving stakers a veto

LDO holders decide, but the people most exposed to a bad decision are stETH holders, who may hold no LDO at all. Dual Governance, approved by an LDO vote in mid-2025 (about 53.6M LDO in favour against a 50M-LDO quorum, with a single dissenting vote) and rolled out to Ethereum thereafter, closes that gap by inserting a dynamic timelock that stETH holders can extend. It is one of the most closely watched governance mechanisms in DeFi because it grafts an exit-based safeguard onto a token-weighted DAO, and it is the largest live instance of optimistic governance: the second constituency can delay or leave, never propose. (The Block; Dual Governance 101)

Every governance action affecting Lido on Ethereum is routed through an Emergency Protected Timelock whose governance address is the DualGovernance contract. Even entirely unopposed, a proposal waits 3 days after submission before it can be scheduled and a further 1 day before it can execute. Dissent is registered by depositing stETH into a signalling escrow, and the escrowed share drives a five-state machine:

• Veto Signalling — once escrowed stETH crosses the first seal at 1% of stETH supply, execution is delayed on a sliding scale, from 5 days up to a maximum of 45 days as more stETH piles in, buying time to scrutinise the proposal.
• Rage Quit — if escrowed stETH crosses the second seal at 10% of supply, governance is frozen: no queued proposal can execute until the dissenting stakers have fully withdrawn their ETH. Stakers who disagree leave before the change binds them.

The thresholds and durations are not protocol lore; they are immutable constants in a deployed config provider contract, readable by anyone. Read on 7 August 2026 at Ethereum block 25,705,002:

That last-but-one row is the part the explainers tend to skip. The exit is real, but it is not fast: a staker who rage-quits waits at least two months after their withdrawal finalises before touching the ETH, and a DAO whose stakers keep rage-quitting stretches that toward six. The delay is deliberate — it prices repeated use of the escape hatch — but it means "exit" here is a slow, one-directional decision rather than a liquid one. (DualGovernanceConfig.sol)

The design was audited by multiple firms (Certora, OpenZeppelin, Statemind, Runtime Verification) and lives in the lidofinance/dual-governance contracts. It borrows the name and spirit of Moloch's rage quit, applied to a two-token structure: LDO proposes, stETH can veto or exit. (Dual Governance overview)

The on-chain record, one year in

Dual Governance is usually described in the conditional — what stETH holders could do. The deployed contracts also record what they have actually done, and the answer is nothing at all. Every figure below was read from Ethereum mainnet on 7 August 2026 at block 25,705,002; the addresses are published on Lido's deployed-contracts page.

The DualGovernance state machine was initialised on 8 August 2025 at 10:21:47 UTC and its persisted state has read Normal ever since. Three fields in getStateDetails() make that a definitive claim rather than a snapshot:

• normalOrVetoCooldownExitedAt is 0 — the machine has never left the Normal state, because that field is stamped on any exit from it and is never cleared.
• vetoSignallingActivatedAt is 0 — Veto Signalling has never once been entered. That field is set on entry and never reset. (DualGovernanceStateMachine.sol)
• getRageQuitEscrow() returns the zero address, which the contract's own documentation states means "no rage quits have occurred in the system". (DualGovernance.sol)

Meanwhile the timelock has been busy. getProposalsCount() returns 12, and every one of the twelve carries status Executed — none cancelled, none stuck. Each was scheduled exactly 3 days after submission, the minimum the timelock allows, from the first on 29 May 2025 to the twelfth on 20 July 2026. The safeguard has never been the reason a Lido proposal was late.

The reason is visible in the size of the bar. stETH total supply at the same block was 9,432,901 stETH, so the first seal — the point at which a proposal is delayed by even five days — requires roughly 94,329 stETH to be escrowed, and freezing governance outright requires about 943,290 stETH. The signalling escrow currently holds 0.0000114 stETH, and getRageQuitSupport() reads 0.00000000012%. In practice the escrow is empty, and has been for a year.

Two honest readings of that record compete, and the wiki takes neither on faith. One: the veto is a deterrent working as designed, shaping which proposals are drafted at all, and a mechanism that never fires can still be doing its job — the optimistic-governance case. Two: a threshold of ~94,329 stETH held by a dispersed, mostly passive holder base is a coordination problem no one has yet solved, and the veto is nominal. Nothing on-chain distinguishes the two, and the twelve uncontested proposals do not decide it either way. What the record does settle is narrower and worth stating plainly: the largest deployed exit-based safeguard in DeFi has never been used. Re-read on 3 September 2026 at block 25,894,634, the machine still answers the same way: getPersistedState() and getEffectiveState() both return Normal, vetoSignallingActivatedAt and normalOrVetoCooldownExitedAt are both still zero, getRageQuitEscrow() is still the zero address, and the signalling escrow still holds 0.0000114 stETH. Because those three fields are stamped on the event and never cleared, the claim is settled by the contract's own storage rather than by the month it happened to be read in – which is why it is stated here with the block that proves it.

One dated caveat belongs alongside it. The timelock still carries emergency protection: an activation committee and an execution committee that can put the system into a 30-day emergency mode and execute outside the normal flow. isEmergencyModeActive() reads false, and getEmergencyProtectionDetails() gives an expiry of 20 June 2027, after which those committee powers lapse on their own. Until then, Dual Governance runs with a training-wheel override that its final form will not have. (EmergencyProtectedTimelock; committee addresses)

Lido DAO in court: the general-partnership question

Lido DAO is the named defendant in the leading US test of whether a DAO is a general partnership – the question that decides whether an unwrapped DAO's participants answer personally for its obligations. Samuels v. Lido DAO, No. 3:23-cv-06492-VC (N.D. Cal.), was filed on 17 December 2023 by an LDO buyer alleging the token was an unregistered security under Section 12(a)(1) of the Securities Act. It names Lido DAO itself alongside four institutional holders: Paradigm Operations, Andreessen Horowitz (AH Capital Management), Dragonfly Digital Management and Robot Ventures.

Judge Vince Chhabria's order of 18 November 2024 (Dkt. 115) denied every motion to dismiss but Robot Ventures'. It held that the plaintiff had adequately alleged Lido DAO is “a general partnership within the meaning of California law,” and that the other three investors were adequately alleged to be members of it and “therefore liable for Lido's conduct.” The order is careful about how far down that reaches. It expressly distinguishes Sarcuni v. bZx DAO, 664 F. Supp. 3d 1100 (S.D. Cal. 2023), which had accepted a partnership comprising every holder of the defendant DAO's token, and holds instead that “every LDO holder, on the other hand, hasn't automatically joined the partnership … because they don't all necessarily have the ability to meaningfully participate in DAO governance.” What is alleged to make a partner is the capacity to participate in governance, not the holding of LDO.

Nothing has been decided on the merits. This is a pleading-stage ruling – the court held only that the claim may proceed – and the order notes that the existence of a partnership is a question of fact for a later stage, and that the partnership may prove narrower or broader once discovery is done. The case is now on appeal, docketed in the Ninth Circuit as Samuels v. AH Capital Management, No. 25-5701 on 9 September 2025, and the district court has stayed proceedings pending that appeal, ordering joint status reports every 120 days; the most recent was filed on 18 June 2026. Both dockets were open when read on 7 September 2026. The wider picture, including the wrappers DAOs adopt to make the question moot, is on DAO legal structures.

It is worth reading alongside Dual Governance above. The mechanism exists because the constituency most exposed to an LDO vote – stETH holders – has no say in it; the litigation asks the converse question, which LDO holders are associated closely enough with the enterprise to carry its liabilities. Both turn on the same underlying fact: in a token-weighted DAO, participation is a spectrum rather than a membership roll.

How Caper approaches this

Dual Governance is a retrofit: Lido started as a token-weighted DAO and later bolted on a second constituency (stETH) with a veto and an exit, because the holders bearing the risk had no direct say. A caper reaches the same destination from the opposite direction — it is built so voice and exit are the same right from day one. There is no separate escrow, no 1%/10% thresholds to organise, and no second token class: every member's voting weight and their pro-rata claim on the treasury at exit are one and the same number, combining the stake they hold with the participation they have shown. A member who dislikes where a caper is heading does not need to accumulate a veto quorum; if they have taken part at all — bought, sold or voted, each of which mints the soulbound record exit spends — they can exit for their share immediately, on any day. Lido's mechanism is the more general fix for the enormous installed base of one-token DAOs; a caper is what that safeguard looks like when it is native rather than grafted on. The exact exit math is on the linked governance pages, verified against the contract — no additional guarantees are implied here.

References

• Lido documentation — protocol, staking modules, and governance reference.
• Lido deployed contracts — the mainnet addresses for DualGovernance, the config provider, the signalling escrow and the Emergency Protected Timelock (primary).
• Lido research forum — proposals and the decentralization debate (primary source).
• Dual Governance 101: Explainer and Dual Governance: An Overview — the veto/rage-quit mechanism and thresholds.
• lidofinance/dual-governance — the audited contracts; DualGovernanceStateMachine.sol and DualGovernanceConfig.sol carry the state-transition and delay-growth logic quoted above.
• The Block, Lido DAO votes to enable dual governance (2025) — the approving vote.