White hat researchers recovered 1,003.62 ETH from a failed 2016 Ethereum ICO, turning flaws in old smart contracts into a reminder that Ethereum’s early technical decisions can live on for nearly a decade.
The researcher, known as 0xFlorent, said he unlocked ETH from the HongCoin contract after the funds were locked up for nine years. Using the Ethereum price of approximately $1,983 on June 1st, the amount recovered is worth approximately $1.99 million.
Recovery relied on the original HongCoin multisig. HonCoin contracts still required action from the admin path for related admin calls.
As such, this episode was closer to contract archeology than traditional exploitation. The same immutable code that saved the refund failure also saved the forgotten route of its avoidance.
The contrast with Hong Kong is striking. Ethereum’s base layer remained static. With authorization passes still valid and coordinated signatures from the original multisigs, 48 original investors became eligible to claim their funds through a refund mechanism that has been defunct for many years.
How the refund route was interrupted
HonCoin is a 2016 Ethereum project that was described as a decentralized venture fund in its public repository. The token sale failed to meet its funding goal, and investors were supposed to be able to get their ETH back through the contract’s refund feature.
The problem lay in the accounting for the contract. In the HonCoin source code, refundMyIcoInvestment() The function checks whether the caller’s token balance is greater than: tokensCreated. If that condition is true, the refund call will fail.
If passed, the function zeroes the caller’s token balance, clears the associated accounting, and tokensCreated Refunds will be made according to the token balance.
Over time, early refunds will result in worldwide tokensCreated counter. That left large holders in a strange position. They still had a balance associated with the original charge, but that balance was likely too large for the remaining counters on the contract.
The refund feature then treated them as invalid and blocked the very users who were supposed to refund them.
Escape paths were also old code. Multisig limited mgmtIssueBountyToken() The administration function adds the provided amount to the recipient’s balance and bountyTokensCreated.
That path belonged to the management side of the contract, so the original multisig needed to participate. Modern Solidity operations default to reverting when an overflow occurs.
Prior to Solidity 0.8.0, arithmetic operations were wrapped in overflow unless the developer added their own checks. Old behaviors formed an escape route.
0xFlorent has identified a way to use the arithmetic operations of management functions to reset the owner’s balance to a low enough value for the refund check to pass. The results were paradoxical. One old bug helped undo substantial damage caused by another old bug.
| stage | key details |
|---|---|
| 2016 Token Sale | Hongcoin raised ETH for a venture fund-style Ethereum project, but later failed to reach its goal. |
| Refund failed | The refund feature denied large holders if the global token counter fell below their balance. |
| old admin path | There was still a multisig restricted function that could change the balance using Solidity arithmetic behavior prior to 0.8. |
| white hat recovery | 0xFlorent has teamed up with the original HonCoin multisig to allow blocked holders to claim their funds. |
| On-chain proof | May 29 transaction shows success refundMyIcoInvestment() A call that generates an internal 96 ETH transfer. |
Multisig enables coordinated recovery
Multisig requirements set the boundaries for HongCoin recovery. The actual recovery depended on the cooperation of researchers and the old control path, as the sensitive path required HongCoin’s original management address to perform the relevant calls.
The adjustments were just as important as the code. This recovery included 41 signed transactions of blocked owners, but another seven small owners were able to get their money back directly without any workarounds.
The ICO was launched on August 29, 2016 and ended on October 28, 2016, but did not reach its fundraising goal.
On-chain records already show refund activity. May 29th on-chain transaction called refundMyIcoInvestment() An internal transfer of 96 ETH was then made from the HongCoin contract to the investor’s address.
The top-level transaction value was 0 ETH because the actual transfer occurred within the contract call.
Those who track funds must distinguish between eligibility and completed distribution. Contract status and multisig implementation have reopened claims channels for funds that have been inaccessible for many years.
Visible on-chain examples illustrate refund activity rather than a complete description of all eligible investor claims.
One should read carefully before generalizing the Hongcoin incident to other old stack funds. The components were unusually specific. There is discernible contract logic, control that remains available in the original control path, a white hat to make adjustments, and enough on-chain value remaining to make the effort worthwhile.
The actual details are ownership and permissions. The old function could change the balance, but only the managed path could call it.
This gives ethical and operational boundaries to retrieval. Outside research found a path forward, the original signatories implemented it, and the claims route was reopened for investors.
The same facts also make it difficult to generalize this case. Many dormant contracts lack active control keys, a clean set of claimants, or a public trail to enable responsible recovery.
This boundary also reduces the temptation to treat the episode as a broad exploitation template. Although technical mechanisms explain why the refund gates were reopened, the end of this story comes from a combination of old codes, survival permits, and public payments.
Similar archeology becomes more dangerous if the contract lacks any of these elements. This is because discovery can reveal weaknesses before available recovery routes are created.
Ethereum holds mistakes and their remedies
Considering the broader history of Ethereum, Hongcoin’s recovery is more than just a curiosity. A 2025 analysis citing Coinbase’s Conor Grogan puts the amount of ETH permanently lost at over 913,111, a conservative estimate that takes into account user and contract-related errors.
This category includes funds sent to write addresses, bugs in contracts, and significant incidents in history.
Some of Ethereum’s most important early moments were also discussions around recovery. In 2016, after the network’s definitive governance crisis, a DAO hard fork moved approximately 12 million ETH from DAO-related contracts to recovery contracts.
In 2017, a self-destruction incident of Parity Technologies’ multisig library resulted in 513,774.16 ETH being blocked in 587 wallets.
These episodes were bigger and more politically heavy than Hong Kong. They still help explain why this small economic recovery is resonating.
Ethereum’s promise of code and state persistence is its security properties and memory system. This saves errors, forgotten assumptions, outdated permissions, and occasional remedies whose future relevance was not visible at the time of deployment.
That long memory now stands alongside a mature security culture. In January, Ethereum veterans announced plans to convert the remaining approximately 75,000 ETH from TheDAO Recovery Fund into an equity fund for Ethereum’s security.
The Hongcoin incident represents the same aftermath of Ethereum’s initial decision, although it is taking place on a much smaller scale.
The next test is recoverability. Do other older contracts contain paths that can be used responsibly? White hat recovery requires more than bugs. It requires legitimate control paths, publicly available on-chain evidence, careful disclosure, and ways to avoid turning contract archeology into an opportunistic attack strategy.
HonCoin shows that some of the trapped funds may remain held within the old logic, waiting for someone to figure out both the flaws and the permission structure around it. This is a hopeful outcome for the 48 investors currently eligible to claim.
This is also a warning to the rest of the ecosystem. Ethereum remembers fraudulent codes, and sometimes even escape hatches.
(Tag Translation) Featured
