Airdrop farming means using a protocol now in the hope of being paid later, by a formula you cannot see. On 23 March 2023, an address that had bridged funds to Arbitrum One, used it in two different months and made more than four transactions, all before the chain’s Nitro upgrade, could claim 1,250 ARB. That was the minimum tier of 3 points. The score came from a snapshot taken more than six weeks earlier, on 6 February, and nobody had been told the rules while it was being built. Someone who did the same with 50 wallets stood to collect 50 times as much, unless the Sybil rules caught them first.
This lesson explains the machine from both sides. It shows what protocols measure and how they filter, then does the arithmetic farmers tend to skip: fees, hours, capital tied up, tax on receipt. It names no project to farm, ranks nothing, and is not investment or tax advice.
What airdrop farming is, and why protocols pay for it
An airdrop is a free distribution of tokens to a list of addresses. Two different things share that word. A marketing giveaway hands small amounts to anyone who follows an account. A retroactive airdrop pays past users of a product, and it is what farmers chase.
Uniswap set the template. Its post of 16 September 2020 gave 400 UNI to each of 251,534 addresses that had ever called its v1 or v2 contracts, using a snapshot that ended on 1 September, and the historical users and liquidity providers together took 15% of the supply. Around 12,000 of those addresses had only ever sent transactions that failed. They were paid anyway, a sign of how little filtering seemed necessary at the time.
Protocols have reasons to give away what they could sell. A token spread among users looks less like a founders’ asset, and voting power needs holders. As the previous lesson on governance tokens, fee switches and vesting explained, holders are only useful if they stay, so an airdrop is partly a bet that a user will become an owner. Airdrop farming is what happens when people take the wage and leave the job.
Farming is the rational response to that bet. If past usage pays, then usage before a token exists becomes a speculative activity of its own, with a cost today and an unknown reward, or none, later on.
The mechanism: snapshots, criteria and points
The snapshot and the retroactive formula
A snapshot is a block chosen as the cut-off. Everything before it counts, and what you do afterwards does not. Arbitrum’s was block 58642080, on 6 February 2023, and its documentation lists 14 qualifying actions on Arbitrum One and four on Arbitrum Nova. They fall into four families: bridging in, transacting across two, six or nine distinct months, passing thresholds of more than 4, 10, 25 or 100 transactions or contracts, and reaching dollar thresholds of value moved or bridged, from $10,000 to $250,000. The docs add: “Point scores were capped at 15.”
Look at what that list rewards. Time mattered more than a burst of activity, since a month is hard to fake cheaply in hindsight. Volume mattered, so capital could earn more than diligence alone. And the rules were published only after the snapshot, which meant no one could aim at them in advance.
Points earned before Nitro reached Arbitrum One on 31 August 2022 were worth twice those earned later. Scores then mapped to tiers, shown here for points scored before Nitro. Points earned afterwards counted for half, so the floor was 625 ARB, and the ceiling stayed at 10,250.
| Points (pre-Nitro) | ARB entitlement |
|---|---|
| Fewer than 3 | Not eligible |
| 3 | 1,250 |
| 4 | 1,750 |
| 5 | 2,250 |
| 6 | 3,250 |
| 7 | 3,750 |
| 8 | 4,250 |
| 9 | 6,250 |
| 10 | 6,750 |
| 11 | 7,250 |
| 12 or more | 10,250 |
Note the jumps: the step from 8 to 9 points added 2,000 ARB, while the step from 9 to 10 added 500. Nobody could plan for that in 2022, which is the point of a retroactive formula.
Points seasons: a scoreboard with no contract
Points were the industry’s answer to retroactive secrecy. Instead of a hidden formula applied once, a program publishes a running score, usually in seasons, and leaves open what the score will buy. Hyperliquid documentation describes one of the best-known versions: 1,000,000 points a week for six months from 1 November 2023 to 1 May 2024, 700,000 a week in a second phase from 29 May to November 2024, a multiplier for activity between 1 and 28 May, and one point for an affiliate for every four earned by the people they referred.
The same page carries a sentence every farmer should read twice: “Hyperliquid reserves the right to modify previous point distributions under its sole discretion.” A point here is an entry on a company’s scoreboard. Hyperliquid did pay out, but the docs page on points does not mention an airdrop at all. The reader who treats points as a deferred salary is holding a promise that nobody made.
EigenLayer used a different unit. Its points tracked restaked ether over time, and the Eigen Foundation’s stakedrop then mapped them linearly to tokens, which critics said favoured large restakers.
The conditions that arrive at claim time
Being on the list does not always mean being paid. EigenLayer’s claim was closed to users in 30 countries, among them the United States, Canada, China and Russia, and the foundation said it took steps against VPN use. At Hyperliquid, Forklog reports, users who had pre-registered for the Genesis Event received HYPE straight into their spot accounts. Claim windows also close: ARB could be claimed only until 24 September 2023. A farmer who lives in an excluded country, or who misses a deadline, carries all of the cost and none of the reward. Check the identity checks and the geographic terms before you begin, not after.
Three airdrops, three outcomes: Arbitrum, Hyperliquid and EigenLayer
Arbitrum, March 2023: the rules revealed after the fact
The foundation allocated 1.162 billion ARB to users, 11.62% of the 10 billion initial supply, and 113 million to DAOs in the ecosystem. Cointelegraph put the number of eligible wallets at over 625,000. DefiLlama shows ARB at $1.34 at 00:02 UTC on 24 March 2023, which put the 1,250 ARB tier near $1,675 on the first day. At the 31 October 2026 price of $0.1615 (09:51 UTC, DefiLlama) those same tokens were worth about $202, a fall of 88%.
Hyperliquid, November 2024: 31% of the supply, none for investors
Hyperliquid’s genesis event on 29 November 2024 sent 31% of the 1 billion HYPE cap, or 310 million tokens, to eligible users, fully unlocked. The project said there were “no allocations for private investors, centralized exchanges, or market makers”, according to The Block. Launch-price reports differ, and DefiLlama has the token at $4.27 at 12:00 UTC that day. On 31 October 2026 it stood at $87.12 (09:51 UTC), about 20 times that level.
The headline hides how uneven it was. ChainCatcher, relaying ASXN’s dashboard on 2 December 2024, gave about 94,000 wallets, a median near 64 HYPE and a 99th percentile near 58,000. Forklog, citing ASXN on 23 December, wrote of more than 94,000 users and a median worth about $2,250 with HYPE near $35, which also comes to about 64 tokens. It added that 56.6% of participants received up to 100 tokens. The two reports differ on the average (2,916 against 2,882 HYPE), and neither average multiplied by 94,000 gives 310 million, so we quote no exact figure for it. What holds in both is the shape: half of recipients got about 64 HYPE or less, near 2% of the average, or about $275 at the 29 November price above. Our guide to using Hyperliquid covers the exchange itself; what counts here is that points were scored for a year on a page that promised no token.
EigenLayer, 2024: a stakedrop with its token frozen
As a reminder, restaking lets staked ether secure other services for extra rewards. The Eigen Foundation announcement set an initial supply of 1,673,646,668 EIGEN, with 15% for the stakedrop as a whole and 5% for Season 1, based on a snapshot on 15 March 2024 at block 19437000. Claims opened on 10 May 2024 for 120 days, and the token could not be transferred at first. Investors held 29.5% and early contributors 25.5%, with one year locked and then 4% a month for two years.
The backlash was fast enough to change the terms. On 2 May 2024 the foundation added 100 EIGEN for each user who had interacted with EigenLayer before 29 April, bringing the Season 1 minimum to 110 EIGEN after simple Sybil filtering. The foundation said it had heard from users who paid high gas costs to take part. Transfers were due to open on 30 September 2024. On 1 October at 11:58 UTC, DefiLlama shows EIGEN at $3.81; on 31 October 2026 it was $0.187 (09:42 UTC), down about 95%. The protocol survives, under the name EigenCloud on DefiLlama, with about $5.77 billion locked as of 31 October 2026.
| Arbitrum | Hyperliquid | EigenLayer (Season 1) | |
|---|---|---|---|
| Distribution | Claim from 23 March 2023 | 29 November 2024 | Claim from 10 May 2024 |
| Share of supply | 11.62% to users, 1.13% to DAOs | 31% | 5% (15% across the stakedrop) |
| What was measured | Bridging, months active, activity, value | Weekly points for activity, plus referrals | Points for restaked ether over time |
| Recipients | Over 625,000 eligible wallets (Cointelegraph) | About 94,000 (ASXN data, via press) | Not stated in the sources opened |
| Main friction | Sybil deductions, Hop blacklist | Points adjustable at sole discretion | 30 countries barred, no transfers at first |
| Early price (DefiLlama) | $1.34, 24 Mar 2023 | $4.27, 29 Nov 2024 | $3.81, 1 Oct 2024 |
| Price on 31 Oct 2026 | $0.1615 | $87.12 | $0.187 |
The table carries the lesson nobody can use in advance. The same mechanism produced a loss of about 95% for a holder of one token and a twentyfold rise for a holder of another. Past airdrops tell you nothing about the next token’s price.
Sybil filtering: how a protocol tells a person from a farm
A Sybil is one actor posing as many. The term matters here because an airdrop pays per address, or by tiers, so splitting activity across wallets can multiply the reward. Defences come in three kinds, usually combined: rules inside the formula, graph analysis and paid informants. Each makes farming costlier, and each can catch honest users.
Rules inside the formula
The simplest filters are deductions. Arbitrum subtracted a point if all of a wallet’s transactions fell within 48 hours, and another if it held under 0.005 ETH and had touched at most one contract. Cheap to apply, and cheap to avoid once known.
Graph clustering: Arbitrum’s public code
Arbitrum also published its detection method on GitHub, an unusual step. It first removes known services such as bridges and exchanges, using data from Nansen, Hop and Offchain Labs. It then builds two graphs: one of transactions with value, one of funder and sweep transfers, meaning the first ether into an account and the last ether out. Connected groups of addresses become clusters, the large ones are split with the Louvain community-detection algorithm, and clusters are flagged on patterns such as funds moving inside a group of more than 20 addresses, a common funding source, or similar activity.
In plain terms, ten wallets funded from one exchange withdrawal and swept back to one address draw a shape on a graph, and that shape can be found by software long after the farmer thought the job was done. The repository gives no total of excluded addresses, and the docs we opened do not either.
Paying the crowd: Hop and LayerZero
Informants came first at Hop, the year before Arbitrum’s airdrop. On 6 May 2022 it said that 10,253 of 43,058 initially eligible addresses were Sybil attackers, on rules that included shared funding addresses, correlated transactions and near-simultaneous actions. Community hunters had to submit at least 20 addresses for a share of the tokens saved, under a one-year lock. A DAO proposal of June 2023 put the savings at almost 3.5 million HOP and paid 25% of them, about 869,566 HOP, to 35 recipients. Arbitrum later disqualified addresses identified in that bounty, so one project’s list followed a wallet into another project’s airdrop.
LayerZero went further and offered amnesty. Its snapshot was finalised at 23:59:59 UTC on 1 May 2024. Farmers could self-report by 17 May, according to its blog, and keep 15% of their intended allocation; the blog told them, as quoted by The Block, “If you think you are a sybil, you are most likely a sybil.” Co-founder Bryan Pellegrino said up to 100,000 addresses had done so by 14 May. The team then narrowed a first list of over two million to 803,093 flagged addresses, which would get 15% with the remaining 85% going to eligible users. Bounty hunters had from 18 to 31 May to report at least 20 addresses, for 10% of each Sybil’s intended allocation.
A two-million list cut to 803,093 says something on its own: the first pass was wide enough that the team changed its criteria to reduce false matches. If you are wrongly caught, there is rarely a court to hear you, only a form or a forum thread.
What still slips through
No filter is complete. In a report published on 22 March 2023, researcher X-explore claimed that 148,595 Sybil addresses had been given about 253 million ARB, or 21.8% of the airdrop, by missing clusters of fewer than 20 addresses, routes through bridges and exchanges, and post-snapshot behaviour. That comes from one team’s model, not an official audit.
The best public evidence on returns comes from a McGill University paper that studied 150 groups reported in Hop’s bounty, 3,551 addresses in all. The authors estimated what those groups would have earned if not caught: 104 groups (69.3%) would have ended with a positive expected net profit, most groups stayed under $10,000 in total, and most averaged under $200 an address. Those are modelled gains, and the addresses were in fact disqualified.
How to farm an airdrop without fooling yourself
Collecting one reward many times through many wallets is what the filters above exist to catch. What follows is airdrop farming in its single-wallet version: using a product you would try anyway, with the costs written down. Here are the steps in order. Our pictures do not show interfaces, so each step is in words.
- Set a budget you can lose in full, and a limit on hours. Write both down before you connect anything.
- Create a separate wallet for this work and fund it only with that budget. Main savings stay elsewhere, ideally on a hardware wallet, and the recovery phrase of neither is ever typed into a website. Our guide to seed phrase security covers the storage.
- Start on a testnet when one exists. Ethereum.org recommends Sepolia for applications and Hoodi for staking tests, and test ether comes free from faucets. A testnet run usually earns nothing by itself, unless a program says so, but it teaches the interface when a mistake costs nothing.
- Read the program’s own documentation: what is counted, the dates, the excluded countries, and whether a token has been announced at all. No announcement means you are guessing.
- Move to mainnet with a small amount. Check the fee your wallet quotes before each signature, since rollups add a data charge that the raw gas price does not show.
- Approve only the amount each action needs, and revoke approvals you no longer use.
- Use the product as a real user would, over several weeks. A burst of activity in one afternoon was exactly what Arbitrum penalised.
- Keep a log from the first day: date, transaction hash, fee, value of anything received. You will need it for tax, and it shows what the farm has cost so far.
- Stop when the cost passes what you budgeted. A deadline for quitting is worth more than a hunch about the next snapshot.
What it costs: a worked estimate for one wallet
The price of airdrop farming starts with gas, because it is the cost people quote and often the smallest one. At 09:51 UTC on 31 October 2026, Ethereum’s RPC reported a gas price of 0.047 gwei with ether at $2,242.89 (DefiLlama). A simple transfer uses 21,000 gas, so it cost about $0.002. A contract call of 200,000 gas, a round figure we assume for the example, cost about $0.02. On a congested day at 10 gwei, the same call would cost about $4.49, more than 200 times as much. That number moves by the minute, and it is why the Eigen Foundation could cite high gas costs in 2024.
The table below is an example, not a forecast. Every input is our own assumption, labelled as such; change them to fit your case.
| Item (example, six months, one wallet) | Assumption | Cost |
|---|---|---|
| Network and bridge fees | 60 transactions at an average of $0.50 | $30 |
| Trading losses (spreads, slippage) | 0.3% on $10,000 of volume | $30 |
| Opportunity cost of capital | $1,000 tied up at a 4% alternative rate | $20 |
| Your time | 20 hours at $25 an hour | $500 |
| Total before tax | $580 |
Now the possible payoffs. We use the Arbitrum minimum tier as a yardstick, because its two prices are on record, and a US reader in an example 24% bracket who sells at receipt.
| Outcome (example) | Gross reward | Tax at 24% | Net after $580 of cost |
|---|---|---|---|
| No token | $0 | $0 | -$580 |
| 1,250 ARB at the 31 October 2026 price | $202 | $48 | -$426 |
| 1,250 ARB at the 24 March 2023 price | $1,675 | $402 | $693 |
Two things stand out. Time is 86% of the cost, so someone who values their hours at nothing sees a very different farm from someone who does not. And if the only outcomes were zero or the best row, the farmer would need about a 46% chance of the reward to break even, or about 6% if the hours are free. Nobody knows the real odds, because no one publishes a failure rate for points programs. The three cases in this lesson are the famous ones, and programs that ended with a small token or none are harder to find. That is survivorship bias, and it flatters every estimate you will read.
Risks the worksheet cannot price
Some losses do not fit in a table because they are rare and large. A contract you deposit into can be exploited, and a position held for a points multiplier is exposed for as long as the program runs. Restaked and liquid-staking positions add another layer of contracts. Unlimited token approvals outlive the session in which you gave them.
Price risk is the one the case studies show best. ARB and EIGEN holders lost most of their first-day value by October 2026, while HYPE holders gained. A token you cannot transfer for months, as with EIGEN, cannot be sold during those months at any price.
The fake version is the danger most specific to this niche. A site promising an airdrop checker or a claim page is the classic bait, and our guide to fake airdrops and wallet-draining claim pages covers the whole scam side. One rule from it is enough here: a real airdrop never needs your seed phrase, and a claim that asks you to approve all your tokens is a drain until proven otherwise. Reach claim pages only from the project’s official site, never from a direct message.
Finally, a self-report is an admission. LayerZero’s scheme kept 15% for those who confessed, but a wallet labelled as Sybil on a public list stays labelled, and Hop’s list showed that such data travels between projects.
Taxes on airdrops: what is settled and what is not
This is the area where farmers are most likely to be wrong in good faith. Nothing below is tax advice, and the airdrop rules in every country named are thinner than people assume.
United States. The IRS has no ruling of its own on token airdrops. Rev. Rul. 2019-24 addresses an airdrop following a hard fork: ordinary income at fair market value, with the basis equal to the amount included in income. Its test is control: “B has dominion and control of Crypto S at the time of the airdrop, when it is recorded on the distributed ledger.” The ruling does not cover token airdrops directly, so using it for them is an analogy: the working assumption of this lesson, not a holding. Rev. Rul. 2023-14 concerns staking rewards and does not mention airdrops, but it ties income to the ability to sell, exchange or otherwise dispose of the asset. Whether a non-transferable token like EIGEN at its claim falls on the same logic is a question for a professional. The IRS digital assets page says to answer yes on the Form 1040 question if you received an airdrop related to a hard fork. Our guide to crypto taxes in the United States covers the rest.
United Kingdom. The HMRC manual is clearer than most. Tokens received without doing anything in return, and outside a trade, may escape Income Tax; tokens received in return for a service, or in expectation of one, are taxable as miscellaneous income or trade receipts. Capital Gains Tax can apply on disposal either way. Airdrop farming, which means doing things in the hope of reward, sits awkwardly on that line. See our guide to UK crypto tax.
France. Settled: private gains on sales fall under a flat 31.4% levy (12.8% income tax plus 18.6% social contributions, page modified on 17 July 2026), unless the taxpayer opts for the progressive scale, reported on form 2086. Accounts held abroad go on form 3916-bis. That official page does not mention airdrops. Not settled: what the receipt itself is. Waltio, a tax software vendor, says a surprise airdrop is probably untaxed until sale, while one earned through actions could count as non-commercial income on receipt, and it cites no official text. We found none either. Our guide to French crypto tax gives the context.
Italy. Settled: the Agenzia delle Entrate brochure on the 2025 budget law confirms that the substitute tax rose from 26% to 33% on gains and other income realised from 1 January 2026, with no exemption threshold. Our guide to Italian crypto tax sets out the exception for euro e-money tokens. Not settled: the word airdrop does not appear in circular 30/E of 27 October 2023, the agency’s main text on crypto-assets, which we searched. Commercial guides disagree on how the receipt should be reported.
| Country | Settled | Open |
|---|---|---|
| United States | Hard-fork airdrop is income when you have control (2019-24) | Direct application to token airdrops; tokens that cannot be transferred yet |
| United Kingdom | Income Tax if received for a service; CGT on disposal | Where a farmer’s activity falls on that line |
| France | 31.4% flat levy on private gains, as of 17 July 2026 | Tax nature of the airdrop on receipt |
| Italy | 33% on amounts realised from 1 January 2026 | Tax nature of the airdrop on receipt |
The practical result is the log from step 8: value at receipt, in dollars or euros, at the time you got control of the token. Without it, you cannot follow any of these treatments, and our guide to calculating crypto capital gains starts from the same records. The cost column of the worksheet should also carry a tax line that you can defend.
Before you start: a short test for any farm
A program can pass or fail these questions without anyone needing to rank it.
- Would you use this product without any reward?
- Has a token been announced, by the project, in writing, on its own domain?
- Are the rules, dates and excluded countries published, and is your country allowed?
- Do the terms let the team change or cancel points at will?
- What can you lose in the contracts you must use, in dollars?
- How many hours will it take, and what are they worth to you?
- Can you record the value at receipt for your tax return?
Answers you cannot find count as a no. The next lesson of this path moves to delta-neutral strategies, where a position is built to earn funding or yield rather than a future token.
