BuildNot yet confirmed elsewhere1 publisher2 min readPublished
52 addresses sending 0.0000001 ETH produced half the failed transactions in a 40-block Ethereum sample
Fixed 21,000-gas ETH sends to EIP-7702 accounts made up 201 of 401 failed transactions in a dev.to block sample. Delegated code runs on receipt, so those sends could not succeed at that limit. The author's breakdown puts all 201 on one cluster of senders, so the half says little about the network at large.
The Engineer · Build desk
Key figures: 201 of 401 failures were plain ETH sends at exactly 21,000 gas. All 50 recipients had delegation code. The sending addresses retried at least 5 and up to 8 times within eight minutes.
| Figure | Value | Note | Claim |
|---|---|---|---|
| Plain ETH sends at exactly 21,000 gas | 201 of 401 failures | 1 | |
| Recipients, all delegated | 50 accounts | 2 | |
| Retries per sending address, at least | 5 retries | up to 8, within eight minutes | 16 |
What happened
- All 50 recipient accounts carried 0xef0100 delegation code, pointing at several delegates including MetaMask's delegator and Uniswap's Calibur account.
- The 201 sends came from 52 addresses that retried 5 to 8 times within eight minutes.
- Base was counted separately: 800 distinct senders produced 2,745 failures, and 59% of them were lost races that succeed when replayed a block earlier.
- The author replayed each failure with eth_call against the parent block's state on free public RPCs, then decoded the revert data.
Why it matters
- exposure Payout bots, scripts and older wallet code that hard-code 21,000 for simple ETH sends fail against every delegated recipient, and a retry at the same limit fails the same way.
- decision Sender code has to pick one of two fixes: estimate gas on every ETH send, or check the recipient for 0xef0100 code and raise the limit when it appears.
- contradiction The post's own counts sit awkwardly: 201 failures across 52 addresses is about 3.9 per address, below the 5 to 8 retries per address it reports.
When ETH goes to an account carrying 0xef0100 code, the delegate's code runs on receipt, and the post says 21,000 gas is then no longer enough [3]. All 201 sends in the sample set a limit of exactly 21,000 and used all of it [1].
The half has a narrow source. Every one of the 201 sends moved 0.0000001 ETH [11]. The author wrote: "That looks like automated dusting, not people losing money." [4] The senders were persistent, if not adaptive [16].
The 401 appears to be the Ethereum count. The 110 failures that succeeded on replay come to 27.4% of 401 [14], in line with the 27% the post states [6]. So the half is an Ethereum figure: 201 of 401 is 50.1% [13]. Base was sampled over 150 blocks and counted separately [12]. With its failures added, the 201 are 6.4% of 3,146 [15], assuming the Base count in the post is complete. The post does not give a delegation count for Base.
For the 201 to transfer to another window of blocks, a comparable cluster would have to be sending to delegated accounts in that window. A window without one would show a much smaller share. The author built the Pitcrew decoder and disclosed it, and linked the sampling script and raw results so the counts can be rerun [9]. The author also calls replaying against the parent block an approximation [8].
We would estimate gas. Estimation returns a limit to use. Detecting 0xef0100 tells the code to raise the limit, but leaves the new value to a guess, and the recipients in this sample point at more than one delegate [2]. That view holds for payout code that sends to arbitrary recipients, since any EOA can delegate [3].
What to watch
- A rerun of the linked script on a block window without the 52-address cluster, to see whether delegated recipients still appear among failed 21,000-gas sends.
- A delegated-recipient count for Base from the same replay method.
- The author's linked raw results, for whether the 201 failures and the 5-to-8 retry count reconcile.
Clarity's read
What the record supports and how the coverage leans. The claims behind it follow.
Reality
- Evidence55
- Adoption
- Insufficient
- Hype gap+8
- Incentives35
- Confidence55
Claim ledger
Ranked by verification strength, evidence, and original report placement.
- [1]
201 of the 401 failures were plain ETH sends (no calldata) with a gas limit of exactly 21,000, and every one used all of it.
- [2]
All 50 recipients had EIP-7702 delegation code (0xef0100...), pointing at, among others, MetaMask's delegator and Uniswap's Calibur account.
- [3]
Since Pectra, an EOA can delegate to contract code. When you send it ETH, that code runs, so 21,000 gas is no longer enough. Anything that hard-codes 21,000 for simple ETH transfers will fail against these accounts every time, and retrying with the same limit changes nothing.
- [4]
"That looks like automated dusting, not people losing money."
- [5]
Fix: estimate gas for ETH sends, or detect 0xef0100 code on the recipient and raise the limit.
- [6]
110 failures (27%) succeeded when replayed against the previous block's state.
- [7]
For each failed receipt, the author fetched the transaction and ran eth_call with the same from/to/data/value/gas against the parent block, using free public RPCs only.
- [8]
Replaying against the parent block is an approximation.
- [9]
The author disclosed building the Pitcrew decoder used in the analysis, and linked the sampling script and raw results so the numbers can be rerun.
- [10]
On Base, 800 distinct senders produced 2,745 failures, and 59% were lost races that succeed when replayed a block earlier.
- [11]
Every one of the 201 sends was 0.0000001 ETH.
- [12]
The post sampled 40 Ethereum blocks and 150 Base blocks.
- [14]
110 replay-successes are 27.4% of 401, matching the stated 27% and indicating the 401 is the Ethereum failure count.
- [15]
If 401 is the Ethereum count and 2,745 the full Base count, the 201 delegated-recipient sends are 6.4% of the 3,146 failures across both samples.
- [16]
The 201 sends came from 52 addresses retrying 5 to 8 times in eight minutes.
- [17]
201 failures across 52 addresses is about 3.9 per address, below the 5 to 8 retries reported.
Sources
1 independent publisher whose own reporting we read for this story.
- dev.toI decoded every failed transaction in 40 Ethereum blocks and 150 Base blocks. Here is why they fail.
1 article · October 13, 2026
Topics and entities
Follow any of these and your For You feed starts watching them — no settings page required.