An airdrop distributes a token for free to a set of addresses chosen by a rule. In practice the rule is retroactive: a snapshot is taken of past on-chain behaviour, a formula converts that behaviour into an allocation, and anyone who matches can claim. The mechanism does three jobs at once, which is both why it is ubiquitous and why it is hard to judge. It puts a token in many hands, which is what “decentralised distribution” means to a regulator and to a market maker alike. It rewards people who used a protocol before it had a token, which is the only compensation early users ever get. And when the token carries votes, it hands out an electorate – it is a governance event, not just a marketing one.

That third job is the one the DAO industry keeps underweighting. An airdrop of a governance token is a decision about who will hold voting power for the life of the organisation, made by a team, before launch, on the basis of a proxy for contribution that everyone can read and most people can game. This page traces how the mechanism was designed, what the measurements say about whether it works, and what it does to governance afterwards.

The Uniswap template

Uniswap’s UNI announcement of 16 September 2020 set the shape almost every later airdrop copied. One billion UNI were minted at genesis, released over four years, split “60.00% to Uniswap community members”, 21.266% to team and future employees, 18.044% to investors and 0.69% to advisors – the last three tranches on four-year vesting. Fifteen percent of supply, 150,000,000 UNI, was distributed retroactively against a snapshot dated 1 September 2020.

The retroactive tranche is where the template lives. Four hundred UNI were claimable by each address that had ever called the Uniswap v1 or v2 contracts – 251,534 historical user addresses – with a pro-rata tranche of 49,166,400 UNI for the 49,192 historical liquidity providers, and 1,000 UNI each for the 220 holders and redeemers of the protocol’s SOCKS token. A flat grant for having shown up, a proportional grant for having supplied capital, and a small collector’s tranche.

Three features of that design became defaults. The snapshot predates the announcement, so behaviour after the announcement cannot qualify. Eligibility is binary at the margin – one transaction, ever, is enough. And the allocation is flat across a very wide band, so the marginal reward for the 500th transaction is zero. Each of those choices is defensible on its own and each of them is a farming instruction once the rule is known in advance.

Designing the criteria: Optimism’s six rules

Optimism’s first airdrop is the clearest published example of an airdrop designed as a selection problem rather than a payout. Its own documentation records 248,699 eligible addresses against a snapshot of 25 March 2022, allocated by six distinct criteria sets – two for OP Mainnet usage and four for behaviour on Ethereum L1 – which stack, so an address matching several receives the sum.

The four L1 criteria are the interesting part, because they are explicit statements about what the Foundation wanted to buy. DAO Voters: 84,015 addresses at 271.83 OP each, qualifying by having voted on or authored at least one on-chain proposal or two Snapshot proposals, filtered to “active DAOs” with at least five proposals and five votes each, and – against Snapshot spam – restricted to voters in the top 99.9% of voting power in each DAO. Multisig Signers: 19,542 addresses. Gitcoin Donors: 23,925. Users Priced Out of Ethereum: 74,272. On top of those sat overlap bonuses that paid steeply for breadth – 2,707 addresses matching four categories, 627 matching five, and 45 matching six, the last receiving 27,534.98 OP.

Read as a governance act, this is an electorate deliberately recruited from people who had already voted somewhere else. Whether that worked is an empirical question the same document cannot answer, and the honest reading of the later turnout record on most L2 governance – see voter apathy – is that prior voting history bought less durable participation than the design assumed.

Points, and the arms race with farmers

Arbitrum replaced criteria sets with a points score. Its eligibility and distribution specification allocated 11.62% of initial supply – 1.162 billion ARB – through a scheme that awarded points for bridging, for transacting across distinct time periods of two, six and nine months, for interaction counts at 4, 10, 25 and 100 contracts, and for aggregate transaction and bridging value crossing $10,000, $50,000 and $250,000. Three points was the eligibility floor and 15 the cap; three points paid 1,250 ARB and twelve or more paid 10,250. Activity before the Nitro upgrade counted double.

The anti-farming rules are stated just as plainly, and they are subtractive rather than exclusionary: one point removed if every transaction in the wallet fell inside a single 48-hour window, and one point removed if the wallet held under 0.005 ETH and had touched fewer than two smart contracts. A wallet could lose two points and still qualify. The third rule was harder: outright disqualification for any address flagged during the Hop protocol Sybil bounty, an open GitHub repository where anyone could file a report clustering addresses by funding source, timing and behavioural signature.

That cross-reference is the most instructive detail in the whole episode. One protocol’s crowdsourced Sybil list became a second protocol’s disqualification input, which means the industry’s de facto identity layer for airdrops was a bug-bounty issue tracker. It also means a farmer’s cost of being caught compounds across drops, which is the only anti-Sybil pressure in the system that scales. The underlying problem does not go away: as long as eligibility is a published function of on-chain behaviour, generating qualifying behaviour is a business, and it is cheaper than the reward whenever the reward is generous.

What the measurements show

The first broad empirical study of the mechanism, “Airdrops: Giving Money Away Is Harder Than It Seems” by Johnnatan Messias, Aviv Yaish and Benjamin Livshits (arXiv:2312.02752, submitted December 2023, last revised July 2025), examined nine major airdrops across Ethereum and Layer-2 ecosystems. Its central finding is that “a substantial share of tokens – up to 66% in some cases – are rapidly sold, often in recipients’ first post-claim transaction.”

The authors attribute the pattern to farmers who optimise against the published eligibility criteria “to extract value without contributing meaningfully to the ecosystem”, and their Arbitrum case study shows the activity spike around the snapshot failing to convert into sustained use. They name the recurring design pitfalls directly: Sybil vulnerability, poor incentive alignment, and governance-token misuse.

The last of those is the one this wiki cares most about. Selling on claim is a price event and a defensible one – a recipient who never wanted the token converts it. But a governance token sold on claim does not disappear; it moves to whoever will pay for it, which is a transfer of voting power to buyers selected by nothing except willingness to bid. An airdrop that is 66% sold has, in governance terms, run a rather inefficient token sale with extra steps. This is the sharpest practical argument for the position that distribution and enfranchisement should be separate decisions – see token-weighted voting and DAO tokenomics.

Unclaimed tokens: expire, or push

Every airdrop leaves supply unclaimed, and the two large L2 drops resolved it in opposite directions on nearly the same date. Arbitrum ran a hard claim window: claims opened at Ethereum block 16890400 on 23 March 2023 and closed at block 18208000 on 24 September 2023, with unclaimed ARB returning to the DAO treasury. Optimism did the reverse – on 15 September 2023 it sent the unclaimed OP directly to the addresses that were eligible to claim it, ending the airdrop by delivery rather than by expiry.

The choice is a governance choice, not an operational one. Expiry concentrates the distribution on the attentive and returns the rest to a treasury the existing holders control. Pushing delivers to everyone the rule selected, including the inattentive, and creates voting power in wallets whose owners never took an action to accept it. Neither is obviously right; what matters is that the decision is usually made after the fact, by whoever controls the distributor contract, without a vote.

How Caper approaches this

Caper has no airdrop, and the reason is structural rather than stylistic. Every governance token enters circulation through a purchase against the caper’s bonding curve – the contract’s buy_token path is the only route by which supply is sold, priced off circulation at the time of the buy. There is no snapshot, no eligibility function, and therefore no rule to farm; the thing an airdrop tries to approximate with proxies is replaced by the plainest possible signal, which is that someone paid.

Voting weight is also not a straight function of the bag. A member’s weight is computed as (held_amount × members_votes) / (vote_supply × tokens_in_circulation) – the product of their governance holdings and their vote tokens, over the corresponding totals. Vote tokens are soulbound: they are minted directly into the voter’s account by the caper’s own state component, which is the only frame permitted to deposit them, and they are burned when the member leaves. They cannot be bought from a seller, which means the half of the weight that an airdrop would have handed out is not transferable at all.

The same expression is what a departing member redeems against. On exit a member returns both their vote tokens and their governance tokens and receives that share of the treasury – so the exit route for someone who no longer wants in is a redemption at their own weight, rather than a sale of voting power into the market. That is the specific failure mode the 66% figure above describes, addressed at the point where it occurs.

References and further reading

• Introducing UNI – Uniswap’s launch announcement, 16 September 2020: the genesis split, the 1 September snapshot, and the per-address retroactive allocations
• Optimism Airdrop 1 – the Optimism community hub’s own record of the six criteria sets, the per-category address counts and allocations, and the September 2023 push of unclaimed OP
• $ARB airdrop eligibility and distribution specifications – the Arbitrum Foundation’s points table, anti-Sybil deductions, and the block heights bounding the claim window
• hop-protocol/hop-airdrop – the open Sybil-report repository whose flagged addresses were reused as a disqualification input by later drops
• Airdrops: Giving Money Away Is Harder Than It Seems – Johnnatan Messias, Aviv Yaish and Benjamin Livshits, arXiv:2312.02752; nine airdrops measured, up to 66% of tokens sold rapidly

Related pages: Liquidity mining and mercenary capital · Mechanism design · Sybil resistance in DAOs · Retroactive public-goods funding · Voting and delegation