Liquidity mining (also called yield farming) is a token-distribution strategy in which a protocol pays out its own governance token to the users who supply liquidity or otherwise use it. The rewards sit on top of ordinary trading fees or lending interest, so for a period the protocol effectively subsidises participation with newly minted tokens. It has been the most widely copied DAO growth mechanism since 2020, and also one of the most criticised: it reliably buys total value locked (TVL) in the short run, and reliably fails to keep it.
Liquidity mining is one of the two ways a DAO gives its token away, and it is useful to hold them side by side. It pays continuously, prospectively, and for a specific behaviour it wants more of; an airdrop pays once, retroactively, for behaviour that has already happened. The two share a failure mode – a large share of the distributed token is sold on receipt in both cases – but they fail on different clocks. An emissions programme keeps buying the same capital every day until it stops paying; an airdrop finds out in a single week whether it bought anything at all.
Liquidity mining as a mass phenomenon dates to June 2020, when Compound began distributing its governance token COMP daily to both suppliers and borrowers across its money markets, in proportion to the interest accruing in each market. Within a day COMP was among the largest DeFi tokens by market capitalisation, and the design was copied across the ecosystem in what became known as "DeFi summer." The insight was simple: instead of selling tokens to investors, a protocol could distribute them to the people actually using it, bootstrapping both liquidity and a holder base at once – a user-centric distribution in place of a sale.
The mechanics are consistent across implementations:
The headline number is APY: the annualised value of emissions divided by capital deposited. The emission side of that ratio is set by the programme and published in advance – Uniswap's only protocol-level programme paid roughly 83,333 UNI per pool per day across four pools, 5 million UNI each over two months – so the APY moves with the reward token's price rather than with anything the depositor did. A rising token lifts it without the protocol deciding anything, which is exactly what pulls in capital that has no interest in the protocol beyond the yield. High token velocity (reward tokens earned and immediately dumped) is the direct on-chain signature of this dynamic.
Mercenary capital is liquidity that chases the highest available yield with no loyalty to any one protocol. It arrives when a program launches and leaves the moment emissions taper or a richer farm appears elsewhere. The consequence is that TVL, during an incentive program, measures the size of the incentive far more than the health of the protocol: a chart of deposits often tracks the emissions schedule almost exactly, then collapses when rewards end.
The canonical demonstration ran over ten weeks of 2020, and both sides of it were liquidity mining. SushiSwap launched on 28 August 2020 as a fork of Uniswap's contracts, paid SUSHI emissions to anyone who staked their Uniswap LP tokens, and on 9 September migrated that liquidity into its own pools – the original vampire attack, and about $840 million of it. Uniswap answered a week later with the other instrument: the UNI launch of 16 September 2020 retroactively airdropped 400 UNI to each of some 251,000 historical addresses and opened a two-month emissions programme of its own across four ETH pairs. The liquidity came back.
What happened when that programme ended isolates the mechanism, because nothing else changed and the date had been published two months in advance. Rewards stopped on 17 November 2020; Uniswap's total value locked fell 38% within twenty-four hours, and 43% from the $3.07 billion all-time high it had set three days earlier.
The costs compound. Emissions dilute existing holders to pay for liquidity that does not stay; the sold reward tokens depress price; and the protocol is left with a temporary TVL number and no durable moat.
The first major refinement was to make emissions earned and directed rather than sprayed. Curve's design is the archetype: liquidity providers still receive CRV emissions, but the split across pools is decided by gauge-weight votes cast by holders who have locked CRV into vote-escrowed veCRV. Locking longer grants more weight, converting some mercenary capital into longer-committed capital and letting governance point rewards at the pools that matter. It also spawned a secondary market – "bribes" and vote markets where protocols pay veCRV holders to direct emissions their way – which is a story in its own right (the "Curve wars", catalogued under vote markets and governance bribery).
The second response was to stop renting altogether. Rather than pay emissions to third-party LPs forever, a DAO can acquire the liquidity for its own balance sheet – protocol-owned liquidity (POL). OlympusDAO popularised the approach with bonding: the protocol sells its token at a discount in exchange for LP tokens, building a treasury of liquidity it controls and no longer has to keep bribing to retain. A liquidity bootstrapping pool solves the adjacent problem of the initial fair launch, though the reference implementation of that format has itself been switched off: Balancer disabled its v2 LBP factory on-chain in May 2026, and the v3 successors had produced 21 non-mock pools across Ethereum mainnet, Base and Arbitrum as of 19 August 2026. Each of these trades an ongoing emissions bill for a one-time treasury cost, and each treats liquidity as an asset to hold rather than a subsidy to pay.
The third response is to stop paying and start collecting: a fee switch reverses the direction of the flow, taking a slice of the fees liquidity providers earn and routing it to the protocol. Uniswap has now run both, and the emissions half was the short one: two months in 2020, never restarted at the protocol level, and then a five-year gap before UNIfication executed on 28 December 2025, turning the fee on and pointing the proceeds at UNI burns.
Multicoin Capital's design-space analysis frames the choices well: who gets paid (makers, takers, or service providers like liquidators), how much (total allocation and how it is measured), and when (immediate, vested, or clawback-able). A program that pays a fixed emission with no lock-up should expect short-term arbitrageurs to crowd out organic users; lock-ups, vesting, and usefulness-weighted rewards push in the other direction. The recurring failures are the mirror image: overpaying for headline TVL, rewarding volume rather than usefulness, and emission rules loose enough to be gamed. Liquidity mining is a distribution tool whose outcome depends on these parameters and on whether the protocol has a reason for the capital to stay once the subsidy stops.
A Caper runs no liquidity-mining program and emits no reward token. Instead of renting liquidity with emissions, each caper's token trades against an always-on bonding curve whose reserve provides continuous liquidity at the curve price – there is no third-party LP to bribe and no emission schedule to taper. Governance weight is not farmed either: the soulbound vote tokens that price an exit are minted by voting alone, one per ranked ballot cast, and not by trading at all – so every holder who has turned up to a decision keeps a credible exit right to redeem their share of the treasury. Liquidity mining pays capital to stay until the reward stops; a caper gives its holders a reason to stay, or a clean way to go, without a subsidy underneath.