Launch attacks and Sybil-neutral design

Wonder

1. Why does being first in a block determine who gets the cheapest tokens?
2. How can one person own ten thousand wallets without anyone knowing until after the fact?
3. Is it possible to design a launch where arriving early gives no advantage?

Explore

Part ILaunch attacks: sniping, bundling and Sybil wallets

1. At a token’s birth, price is determined by position on a bonding curve. The first buyer pays the lowest possible price; each subsequent purchase pushes the reserve higher and the unit cost up. This creates a race: whoever lands in the first block after creation captures the maximum discount. Speed is not just an advantage here; it is the entire economic value of the entry point.
2. Snipers monitor the network for new token creations. The moment a mint transaction lands, they execute a buy order in the very next slot. On high-throughput chains like Solana, this race is measured in milliseconds. The sniper buys thousands of tokens at the curve’s opening price and often sells them within minutes to organic buyers who see the token trending.
3. Bundling formalizes this race. A creator can group the token’s creation and buys from many fresh wallets into a ‘bundle’ of up to five transactions, each able to carry several buys, which execute together or not at all. This lets a single actor take much of the early supply, ensuring that no outside sniper can interject between the creation and their own coordinated buying.
4. The cost of creating new addresses on most blockchains is negligible. A script can generate thousands of unique public-private key pairs in seconds, each appearing as a distinct holder on-chain. This enables Sybil attacks: one actor splits their holdings across many wallets to disguise concentration or to farm per-wallet rewards like airdrops and governance votes that assume each address represents a unique human.
5. Detection is retrospective. Tools like Bubblemaps analyze the funding graph of these addresses, tracing flows back to a common source wallet. If a hundred ‘independent’ holders all received their initial funds from the same exchange account or the creator’s wallet, they are revealed as one entity. However, this analysis happens after the price impact has occurred and the profits have been realized.
6. The design challenge for future launches is structural: can a mechanism be built where first-come pricing is removed entirely? If every order in a block gets the same price, being first within the block earns nothing (a later lesson). If rewards scale with balance rather than with the number of addresses, splitting into Sybil wallets gains nothing.

Part IIFees, symmetry and Sybil-neutral rewards

7. A fee on every token transfer, with no exemptions for specific addresses or relabelled wallets, imposes a constant cost on each leg of any trade. Wash trading, where an entity buys and sells to itself to simulate volume, incurs this fee twice per round trip. Because the cost is real capital lost on every execution, the economic incentive to generate fake volume diminishes sharply as the fee rate increases.
8. Just-in-time (JIT) strategies jump in for a single moment to capture a payout: on AMMs, adding liquidity just before a large trade to collect its fee and removing it right after, taking fees passive providers would have earned; with reward tokens, buying just before a distribution and selling right after. A fee charged on both entry and exit makes such round trips cost more than the brief payout unless that payout is unusually large; cheap entry and free exit invite them.
9. J. R. Douceur’s 2002 paper on Sybil attacks established that systems relying on identity counts are vulnerable to an adversary creating many fake identities. In tokenomics, rewards distributed equally per wallet invite this farming: one user can create ten wallets to claim ten times the reward for the same total capital. This decouples reward from actual economic stake, allowing a single entity to dominate yield without proportional investment.
10. Linear, pro-rata distribution solves this by tying rewards strictly to balance size. If a reward pot is distributed in proportion to holdings, splitting one balance across multiple wallets yields exactly the same total payout as holding it in one. The adversary gains nothing from Sybil attacks because the sum of rewards remains constant regardless of how many addresses hold the tokens. Identity count becomes irrelevant; only capital matters.
11. Consider a reward pot of 10,000 units distributed across a total supply of 100,000 tokens, so each token earns 0.1 units. One wallet holding 1,000 tokens receives 100 units; ten wallets holding 100 tokens each receive 10 units apiece, again 100 in total. The distribution is invariant to address fragmentation. This linearity ensures that Sybil attacks are economically neutral, forcing attackers to compete on actual capital deployment rather than identity multiplication.
12. The trade-off is volume suppression. Fees reduce the net return for legitimate traders and can kill a market if set too high. Solana’s Token-2022 standard supports native transfer fees, while EVM chains often rely on protocol-level taxes or smart contract logic. SafeMoon, whose executives were charged with fraud in 2023, shows how a tax can serve insiders when a team controls its proceeds. Designers must balance deterrence against market viability.

Key facts

Exercise nine, six, three, one

Fact check

Dates, figures and claims checked against the published sources listed below.

• Jito Labs documentation: bundles
• Bubblemaps documentation
• J. R. Douceur, 'The Sybil Attack' (IPTPS 2002)
• Solana Program Library, Token-2022 documentation: Transfer Fees
• US DOJ and SEC charges against SafeMoon executives (November 2023)