Curve DAO

The Curve DAO governs Curve Finance, a decentralized exchange specialised in low-slippage swaps between like-priced assets (stablecoins, staked-ETH pairs) and, since 2023, an issuer of the crvUSD stablecoin. Curve matters to DAO governance out of all proportion to its size because of how it is governed: it introduced vote-escrowed tokenomics (veCRV), the design that most later DeFi protocols copied, and it became the arena for the "Curve Wars" – the clearest real-world demonstration of how token-based voting power becomes a market to be bought. (Curve DAO docs)

veCRV – the vote-escrow model

Curve does not let you vote with liquid CRV. To gain governance power you must lock CRV in the VotingEscrow contract for a fixed term of up to four years, receiving non-transferable veCRV in return. Voting power scales with lock length and decays linearly as the unlock date approaches, so influence must be continually renewed. Locking buys three things at once: governance votes, a share of protocol trading fees, and a boost of up to 2.5× on the CRV rewards your own liquidity positions earn.

The point of the lock is to couple control to commitment: a veCRV holder cannot vote and immediately sell, so voting weight is tied to a multi-year stake rather than to whoever happens to hold tokens at a snapshot. This is the ve model's answer to the plutocracy and vote-renting problems of plain token voting, and it was influential enough that "ve-tokenomics" became a category of its own across DeFi (see DAO tokenomics) – later protocols copied it directly – Yearn's veYFI and Frax's veFXS among them – and its most prominent fork, Balancer's veBAL, ran the same 80/20-lock playbook before deprecating it entirely in 2026. (curve-dao-contracts)

Gauge weights and the Curve Wars

veCRV holders do more than pass proposals – each week they vote on gauge weights, which decide how Curve's ongoing CRV emissions are split across liquidity pools. Directing emissions to a pool deepens its liquidity, so any protocol that wants cheap, deep liquidity for its own token has a direct incentive to control veCRV. That incentive produced the "Curve Wars."

Rather than each protocol locking CRV itself, Convex Finance emerged as an aggregation layer: deposit CRV with Convex, it locks the maximum term on your behalf and hands back a liquid receipt, and Convex votes the accumulated veCRV as a bloc. Within months Convex controlled more than half of all veCRV, becoming the effective power broker of Curve governance. It still is, and the share is drifting up.

At Ethereum block 25,943,617 (10 September 2026) the Convex voter proxy held 418,549,575 veCRV of a 776,158,764 total – 53.93%, up from 53.68% a month earlier – while holding only 49.88% of the 849,681,279 CRV actually locked in the escrow. The gap is decay: every other lock runs down toward its unlock date, and Convex's is re-locked to the four-year maximum. That maximum is a rolling date rather than a fixed one, which is why it should not be quoted as a standing fact – this page previously gave it as 25 July 2030, and the same on-chain read now returns 22 August 2030.

All four figures are eth_call reads of the VotingEscrow and CRV contracts and can be reproduced at any block. The full on-chain read, including the CVX mint reaching its final cliff, is on the Convex Finance page. On top of that grew an open bribe market (Votium and others) where projects pay veCRV/vlCVX voters directly to steer emissions – turning voting power into a rentable yield.

Major protocols ran their own strategies here too – Frax amassed one of the largest vlCVX positions to steer Curve emissions toward its own pools. The template also travelled: the Solidly ve(3,3) fork of this gauge-and-bribe design became the dominant DEX on the OP-stack chains, most prominently Aerodrome on Base. The Curve Wars are the canonical case study for the lesson on the token-weighted voting page: even a lock-based system can have its votes bought, the leak just moves to a new place. (Curve governance forum)

The amplification coefficient those votes sit alongside is not only an economic parameter. Curve's ramp_A moves it over a rate-limited window because the invariant a trade is priced against is solved by iteration rather than by formula – see Numerical pricing in AMMs and bonding curves.

crvUSD, and the 2023–2024 stress tests

In 2023 Curve launched crvUSD, an over-collateralized stablecoin with a novel "soft liquidation" mechanism (LLAMMA) that partially converts collateral instead of liquidating it all at once, plus the LlamaLend lending markets built on the same engine. The same period exposed Curve's risks in public:

• July 2023 – the Vyper exploit. A reentrancy-lock bug in specific versions of the Vyper compiler (0.2.15, 0.2.16, 0.3.0) – not in Curve's own logic – let attackers drain several Curve pools for a cumulative $60M+. A large share was ultimately returned by white-hats and the attackers, but it was a systemic scare for the whole Vyper-based ecosystem. (Halborn post-mortem)
• June 2024 – the founder-liquidation cascade. Founder Michael Egorov had borrowed roughly $95M in stablecoins against about $141M of CRV spread across five lending venues. A ~24% CRV drop on 13 June triggered cascading liquidations that wiped out most of that collateral and briefly left ~$10M of bad debt in Curve's own LlamaLend market. Egorov repaid the bad debt within days, so no permanent loss was passed to depositors – but the episode is a standing lesson about founder over-leverage against a governance token. (CoinDesk)

How Caper approaches this

Curve's veCRV set out to solve a real problem – transient token holders voting with no lasting stake – by forcing a lock. It worked, and then the lock itself became a tradable asset: Convex tokenised it and a bribe market rented it out, so control drifted back toward whoever could pay. A caper pursues the same goal – tying influence to genuine commitment – without a lock to tokenise. Its voting weight combines the stake a member holds with the participation they have actually shown, so weight is earned rather than rented, and a one-off borrowed position buys little say.

There is also no gauge system and no emissions to direct: a caper's tokens come from a bonding curve, not a weekly inflation vote, so there is simply no emission stream for a "war" to capture. And because that same earned weight is also each member's claim on the treasury at exit – that same weight, not the pro-rata slice by balance a rage-quit pays, control stays coupled to real economic exposure – the property the lock was reaching for – with nothing to unbundle and sell. This is a design contrast, not a claim of superiority; the mechanics are on the linked pages and verified against the contract.

References

• Curve DAO: Voting Escrow (veCRV) – the canonical vote-escrow reference (primary).
• curvefi/curve-dao-contracts – the DAO and gauge/escrow contracts (primary).
• Curve governance forum – proposals and gauge-weight discussion (primary).
• Halborn, Explained: The Vyper Bug Hack (July 2023) – Curve reentrancy post-mortem.
• CoinDesk, CRV slides as founder loans face liquidation (2024).