Pendle Finance is a decentralized protocol that tokenizes future yield: it wraps a yield-bearing asset and splits it into a Principal Token (PT), redeemable 1:1 at maturity, and a Yield Token (YT), which captures the variable yield until then. Traded against each other on a purpose-built AMM, PT and YT let users lock in a fixed yield, speculate on a floating one, or hedge — the on-chain analogue of an interest-rate market. What makes Pendle a governance case study rather than just a DeFi venue is how it steered its own token incentives: for four years it ran one of the most-copied vote-escrow gauge systems in DeFi, and in January 2026 it tore that system out — making it a live data point in the broader retreat from the ve-lock model.

vePENDLE: gauge voting, the Curve way

From 2022, Pendle governance ran on vePENDLE — vote-escrowed PENDLE, the same veCRV pattern Curve pioneered. Holders locked PENDLE for up to two years; voting weight scaled with both the amount locked and the time remaining, decaying as the lock ran down. That weight did one central job: direct emissions. Every epoch — snapshotted each Thursday at 00:00 UTC — vePENDLE holders voted to channel PENDLE incentives toward specific pools, and in return earned 80% of the swap fees from the pools they backed. An incentive cap tied each pool's maximum incentive to its actual swap-fee performance, so votes could not endlessly farm emissions into a pool nobody traded.

This is the exact token-weighted, lock-to-vote design that produced the "Curve wars" and the Convex liquid-locker layer built to game it. Pendle inherited both its power and its problem.

January 2026: retiring the lock (vePENDLE → sPENDLE)

On 20 January 2026 Pendle announced sPENDLE, a liquid staking token that replaces vePENDLE as the protocol's primary governance and reward token. The team's stated reason is a governance-participation failure: only about 20% of PENDLE supply was ever actively engaged under vePENDLE, because the multi-year lock was too complex and too illiquid for most holders. sPENDLE keeps skin in the game without the trap — a 14-day withdrawal period (or instant redemption for a 5% fee) instead of a two-year lock, and rewards funded by up to 80% of protocol revenue used for PENDLE buybacks. Existing lockers were not stranded: they received boosted sPENDLE up to 4× based on remaining lock duration, decaying linearly to 1× over the life of the old lock. sPENDLE staking went live 20 January; new vePENDLE locks were paused 29 January.

The deeper change is under the hood. Manual gauge voting — the heart of the ve-model — was replaced with an algorithmic emissions model that routes incentives by data-driven demand signals rather than by weekly vote, cutting overall emissions by roughly 30%. Governance shifts from humans bribing votes to steer rewards toward a formula allocating rewards, with tokenholders governing the formula.

Where it sits: the post-ve turn

Pendle's move is not isolated. Within months of each other, the two other most prominent ve-forks in this directory made the same call: Balancer voted (BIP-919/920/921) to deprecate veBAL for raw one-token-one-vote, and Pendle retired vePENDLE for a liquid staking token plus algorithmic emissions. The vote-escrow model that dominated 2022–2024 DeFi governance — Curve, Balancer, Pendle, Convex, and the bribe markets on top — is being unwound by the very protocols that popularized it, on the recurring finding that lock-to-vote concentrates power in a rentable few while leaving the median holder disengaged. Pendle's own number — 80% of supply sitting out — is a clean statement of the problem. The open question these redesigns raise is whether making governance more liquid improves participation or simply makes vote weight easier to rent by the epoch. Not every ve-fork is unwinding, though: on Base, Aerodrome kept the ve(3,3) model, became the chain's dominant DEX, and in 2026 is consolidating Velodrome under it rather than retiring the lock.

How Caper approaches this

Pendle spent four years engineering elaborate escrow to make governance weight expensive to acquire, then concluded the escrow itself was the problem and swapped it for a liquid, tradeable stake. Caper starts from the other side: it never issues a lockable or tradeable governance token at all. In a caper a member's vote weight is (held × votes) / (vote_supply × circulation) — the amount of the caper's token held, multiplied by a soulbound, non-transferable vote record that is minted one-per-vote and can never be sent, sold, or locked to another wallet (verified in contracts/src/caper_dao.rs: the vote token is DIVISIBILITY_NONE and minted straight into the voter's account, and compute_vote_weight is that product). Because the record is earned by participating rather than bought or escrowed, there is no lock to optimize, no bribe market to build on top, and no liquid governance token to rent — the failure modes Pendle, Curve, and Convex spent years fighting simply have no surface to attach to.

The same weight also settles the exit: a member's pro-rata share of the treasury on exit uses the identical (held × votes) / (vote_supply × circulation) helper, so the number that governs and the number that cashes out are the same — governance and skin-in-the-game are one quantity, not two tokens to reconcile. (This is a design contrast, not a claim that Caper does yield tokenization.)