The Strange Economics of Airdrops: Paying People to Look Like Users

Imagine a new restaurant opening in town with an unusual promise:

"Every time you eat here this month, we will record your visit. Later, our most loyal early customers may receive a share of the restaurant's future rewards."

The next day, every table is full.

Some people came because they genuinely liked the food. Some discovered that ordering the cheapest side dish every afternoon might improve their chances. One person brought twenty friends. Another came back again and again under different names.

At the end of the month, the owner looks at the numbers and sees extraordinary growth.

The visits really happened.

The register really rang.

But the data now answers a much harder question than it seems:

Did people come because they wanted the product, or because being seen using it had become valuable?

This is the strange economics of the crypto airdrop.

The Cold-Start Shortcut

Every new blockchain, decentralized exchange, bridge, or DeFi protocol starts with a problem: empty systems are hard to grow.

Liquidity attracts traders. Traders attract more liquidity. Developers prefer ecosystems with users. Users prefer ecosystems with applications. But somebody has to arrive first.

Airdrops offer one way to break that loop. Instead of reserving all token ownership for founders and early investors, a protocol can distribute governance or economic tokens to people who used the system early .

The idea is appealing: reward the people who helped create activity, spread ownership, and give early participants a reason to stay involved.

But the protocol immediately runs into a problem.

It cannot reward "genuine users" directly.

It has to reward something it can actually see.

The Metric Becomes the Game

A blockchain can see addresses, transactions, deposits, swaps, bridge activity, active days, governance votes, and liquidity positions.

It cannot see motivation.

So a project chooses measurable behaviors as proxies for valuable participation. Perhaps using the protocol on several different days matters. Perhaps bridging assets matters. Perhaps supplying liquidity for a month matters.

Then money enters the picture.

Once users suspect those behaviors may determine a future token allocation, they stop being passive measurements. They become clues to a scoring system.

This is the logic behind Goodhart's law: when a measure becomes a target, people begin optimizing the measure itself .

The question changes from:

"Do I need this product?"

to:

"What is the cheapest sequence of actions that makes my wallet look like the kind of user this protocol wants to reward?"

A person might bridge a small amount back and forth, make token swaps they did not otherwise need, spread activity over several days, or interact with extra applications simply because those actions might count later.

None of those transactions is fake.

The blockchain records them perfectly.

The problem is that the same transaction can now mean two different things: genuine demand, strategic positioning for a reward, or some mixture of both.

One Person, Many Wallets

Then crypto adds another complication.

Creating a new blockchain address is computationally cheap. One person can control one wallet, ten wallets, or thousands.

That does not mean making each wallet look eligible is free. Farming can require gas, bridge fees, capital, liquidity exposure, time, scripts, and operational infrastructure.

But once the expected value of an airdrop becomes large enough, manufacturing another plausible wallet profile can become an economic calculation.

This is where two ideas that are often mixed together should be separated:

• Airdrop farming: changing your behavior to maximize the expected token reward.
• Sybil farming: controlling multiple wallets or identities in order to obtain multiple allocations intended for distinct participants .

A person can farm an airdrop with one wallet without pretending to be one hundred different users.

A Sybil operation goes further. It asks whether one economic actor can manufacture the appearance of many independent participants cheaply enough to make the extra allocations worthwhile.

That turns user acquisition into a contest between mechanism designers and people trying to reverse-engineer the mechanism.

Then the Protocol Starts Doing Forensics

LayerZero's 2024 token distribution is a useful example of how strange this can become .

The interoperability protocol said that nearly six million unique wallet addresses had interacted with it before the distribution.

That sounds like an enormous user base.

But addresses are not people.

So LayerZero faced a question that sounds almost absurd for a permissionless crypto system:

Which pseudonymous wallets should count as legitimate participants?

The protocol effectively said: We want to reward users.

Farmers could reasonably answer: Define user.

LayerZero then built a multi-stage anti-Sybil process. Suspected Sybil operators were given a window to self-report and retain a reduced allocation. Later stages used internal analysis and community reports to identify coordinated wallet clusters.

At that point, token distribution had become a form of blockchain forensics.

Analysts could look for patterns such as many wallets funded from a common source, highly synchronized activity, repeated transaction sequences, similar amounts, common counterparties, or clusters that moved through the same protocols in unusually similar ways .

But this creates another feedback loop.

Once farmers learn which patterns look suspicious, they change them. Timing becomes less regular. Funding paths become more diverse. Transaction sizes vary. Activity is spread across longer periods.

The detector changes the farmer.

The farmer changes the detector.

And neither side ever gets access to the one thing it actually wants: a perfect label saying one human, one wallet, genuine intent.

After the Airdrop, the Experiment Gets Interesting

The cleanest moment often comes after the token is distributed.

Before the airdrop, users may still believe that more activity could improve their allocation. Afterward, that particular incentive weakens or disappears.

What happens next can reveal how much of the earlier activity had reasons to survive without the same expected reward.

But even here, the story is not as simple as "airdrop equals ghost town."

When Activity Falls Back

Starknet and zkSync both provide examples of networks where activity rose sharply around their token-distribution periods and then fell substantially afterward .

That does not mean every address that disappeared was fake. Some were real people who had a perfectly rational reason to use the network while a potential reward was available and less reason afterward.

The important point is more modest:

pre-airdrop activity can contain a large incentive-sensitive component.

Once the expected reward changes, that component becomes easier to see.

When the Network Keeps Growing

Arbitrum complicates the simple collapse story.

Nansen's analysis two months after the March 2023 ARB airdrop found that overall transaction activity and active addresses remained above earlier historical averages, even though many individual airdrop recipients reduced their activity or sold tokens .

That does not prove the airdrop caused durable growth, nor does it prove that every remaining wallet represented organic product demand.

But it does show something important:

an airdrop recipient losing interest is not the same thing as an entire network losing its users.

In Arbitrum's case, the post-airdrop activity was at least consistent with a network that had substantial underlying usage beyond the distribution itself.

What Did the Metric Actually Measure?

This is where crypto analytics becomes dangerous if the labels are too casual.

Active addresses are not active people. One person may control many addresses.

Transaction count does not necessarily measure meaningful use. Scripts can generate real but economically trivial interactions.

Volume can be useful, but repeated or circular activity may distort it depending on the protocol.

Total Value Locked is a different kind of signal again. Wallet multiplication does not create capital from nothing, although token incentives can attract temporary or highly mobile capital that leaves when rewards change.

And then there is retention.

It is not perfect, but it asks a better question:

Does the wallet keep using the product after the eligibility criteria are fixed, the token is distributed, or the subsidy changes?

That moves us closer to durable utility than a pre-airdrop spike ever could.

Maybe the Goal Is Not to Eliminate the Farmer

The obvious response to farming is to build better filters.

But there is another possibility.

What if the protocol designs the reward so that even a strategic participant creates something useful while chasing it?

Instead of paying for ten meaningless transactions, reward liquidity that remains in the system for a long time. Instead of rewarding raw wallet count, use reward curves that reduce the benefit of splitting one position across many addresses. Instead of paying for easily copied actions, reward outcomes that are expensive to manufacture without contributing something real.

That shifts the design question from:

"How do we catch every farmer?"

to:

"Can we make farming behavior useful?"

A participant may still be there for the money. But if the incentive makes that participant supply durable liquidity, test infrastructure, create useful market depth, or perform some other positive externality, the distinction between "farmer" and "valuable participant" becomes less clean.

The behavior can be strategic and still help the network.

Were They Users, or Were They Optimizing the Test?

Every transaction in an airdrop campaign can be completely real.

The signatures are valid. Gas was paid. State changed. The blockchain did exactly what it was supposed to do.

What the ledger cannot tell you is why the transaction happened.

That is what makes airdrops such a strange experiment.

A protocol tries to identify valuable users by measuring behavior. Then it attaches money to the measurement. Users notice. Their behavior changes. The protocol builds filters. Users adapt again.

By the time the token finally arrives, the metric may still be accurate in the narrow sense — those transactions really happened — while becoming much harder to interpret as evidence of genuine adoption.

Once a metric carries money, it stops merely measuring behavior.

It starts shaping it.