Rage-quit is the right of a DAO member to leave a decision they disagree with by burning their governance stake and withdrawing a proportional share of the treasury before the disputed action executes. It is the clearest expression of a broader design principle – exit rights – that gives minorities a credible alternative to being outvoted: instead of fighting a losing vote, you take your money and go. The primitive was introduced by Moloch DAO in 2019 and has since propagated across investment, grant, and treasury-heavy DAOs. For where this fits among the other levers a DAO uses to hold power accountable, see DAO governance models and What is a DAO?.
The community pools its money in a contract that also holds the exit, so any member can burn their membership tokens and receive their share of the treasury without a vote or anyone's permission. Three designs run this in production, and they differ in when a member can leave and how much of the treasury the contract counts as theirs.
| Design | How members pay in | When a member can leave | What the exit pays |
|---|---|---|---|
| Moloch (v2, and Baal, the v3 that DAOhaus deploys) | A tribute of tokens, accepted by a member vote in exchange for shares or loot | Any time in Baal; in Moloch v2, any time except while a proposal the member voted yes on is still unprocessed | The member's proportion of every token in the treasury |
| Juicebox | Paying the project, which issues its token at a rate the owner sets | Whenever the project's current ruleset allows cash outs | A share of the surplus, the funds above what the project has set aside to pay out; a cash out tax rate above zero withholds part of it, which favours holders who stay |
| Caper | Buying the caper's token on its bonding curve, which funds the treasury | Any time, once the member has voted | A fair share of the treasury; Leaving a caper has the steps |
In the Moloch model, membership is denominated in shares (voting + economic rights) and loot (economic rights only, no vote). Every passed proposal enters a grace period before it can be processed. During that window any member can rage-quit: the contract burns their shares and loot and transfers them a pro-rata portion of every token in the guild bank, in one atomic transaction.
(your shares ÷ total shares) × treasury, spread across the full asset mix rather than cashed out at a single price.A treasury-controlling DAO concentrates real money behind a majority vote, which creates two failure modes that exit rights directly address:
The trade-off is treasury instability: a wave of exits can drain a DAO precisely when it is under stress, and a persistent gap between a token's market price and its "redeemable" treasury value invites arbitrage – the pattern critics call turning governance into a cash-out button.
That trade-off is real, but the Aragon DAO case shows the arbitrage does not wait for an exit right to exist. Aragon had no rage-quit: ANT holders had a promise of treasury control and no mechanism to enforce it. The discount between ANT's market price and the treasury behind it drew buyers anyway, and with no formal way out the only route to the money ran through a majority vote – so the claim arrived as a hostile takeover rather than a redemption, and the association blocked it. A DAO with rage-quit cannot be raided for its treasury, because there is no locked discount to arbitrage in the first place. The absence of an exit right does not remove the pressure; it converts it into a governance fight.
Between Moloch and the mega-treasuries, Nouns DAO showed exit rights working at scale: its 2023 fork mechanism let a dissenting 20% of holders split off with their proportional share of the treasury, which they did in two separate forks. The debate then moved from framework contracts to billion-dollar treasuries in 2026, when GnosisDAO took two votes on the same idea seven weeks apart and got opposite answers.
GIP-150 was defeated. Posted on 21 April 2026, it offered any GNO holder a one-time, opt-in exchange of tokens for a pro-rata share of the treasury at net asset value. Illiquid off-chain investments – roughly $25M the DAO had not previously disclosed – and Gnosis Ltd's enterprise value were to be handled by a synthetic claim token, gLTD-CLAIM, paying a fixed share of future distributions above a return-of-capital hurdle. The case rested on the treasury dashboard delivered under GIP-148: $22.5M had been deployed to Gnosis Ltd across three quarters while the discount to NAV widened rather than closed. It lost on Snapshot: 167,692 GNO for, 432,545 against, on 95 votes.
GIP-151 passed by dropping the contested half. Posted on 12 June 2026 by the same author, but this time "developed jointly with the Gnosis founding team", it kept the one-time pro-rata redemption and removed the synthetic claim token and any attempt to value the operating business – stated plainly as a way to avoid "any dispute over the valuation of illiquid or strategic holdings". Redeemers take liquid assets in kind plus a discounted share of capital called to date by GnosisVC; holders who stay keep their full claim on Gnosis Ltd. The eligible base was fixed at 1,364,058 GNO, the 3,000,000 total supply less the DAO's own holdings and Gnosis Ltd's, and redeemed GNO returns to the DAO and leaves circulation permanently. It carried 157,748 GNO for against 2,499, clearing the 75,000 quorum at about 215% on 49 votes.
The pair says more than either vote alone. The proposal that failed asked the DAO to price everything it owned; the one that succeeded asked it to price only what was liquid, and the opposition collapsed from 432,545 GNO to 2,499. But the winning vote also drew roughly a quarter of the weight and half the voters of the one it replaced. Retrofitting an exit right onto an established DAO turns out to be less a fight about whether members may leave than about what they are entitled to carry out with them – and the narrower the answer, the easier it passes.
Moloch's rage-quit governs millions; Lido's Dual Governance applies the same idea to the largest staking pool on Ethereum, and it is the clearest evidence available of what a retrofitted exit right does once it is live. LDO holders govern, but stETH holders carry the risk, so since 2025 every Lido governance action on Ethereum passes through a timelock that stETH holders can extend by escrowing stETH: 1% of supply buys a delay that slides from 5 to 45 days, and 10% freezes governance until the dissenters have withdrawn their ETH. It is rage-quit rebuilt for a two-token DAO – LDO proposes, stETH leaves. (Dual Governance 101)
Two features of that design are worth carrying into any exit-right discussion, and both are legible only in the deployed contracts rather than the explainers. Read on 7 August 2026 at Ethereum block 25,705,002:
Normal since 8 August 2025 – vetoSignallingActivatedAt is still 0 and getRageQuitEscrow() still returns the zero address, which the contract documents as meaning no rage quit has occurred. The signalling escrow holds 0.0000114 stETH against a first seal of roughly 94,329 stETH, and all twelve proposals put through the timelock have executed. (deployed contracts)That is the general lesson, not a Lido-specific one. An exit right bolted onto a large token-weighted DAO is a coordination instrument, not an individual one: no single holder can trigger it, so it only ever fires if a dispersed base organises. Compare GnosisDAO above, where the same right had to be put to a vote of the very majority it constrains, and Moloch, where any member can leave alone on any block. The size of the threshold is what decides which of those three a DAO has actually built.
Caper builds a standing exit right into every caper by default: no grace-period window to catch, no bolt-on vote required. A member redeems out by burning their soulbound vote tokens together with the caper tokens they are exiting, and the caper's treasury pays out a share priced by their canonical weight – the governance tokens held combined with the vote tokens earned – rather than the pro-rata slice of raw balance that rage-quit pays in every other example on this page. The treasury is funded through the bonding curve, so an exit settles against a transparent on-chain balance rather than a negotiated one – there is no net asset value to dispute, which is the whole substance of the GnosisDAO pair above.
The exit is open to members who have voted. Each ranked ballot a member casts mints them one soulbound vote token, and since the redeploy of 11 September 2026 nothing else a member can do mints one, so buying a caper's tokens does not by itself open the exit (Voting). The exit also checks that the vote tokens handed in come from the leaving member's own balance; Leaving a caper walks through the steps.
gLTD-CLAIM synthetic claim token.