Is the minimal CREATE2 factory EIP-7997 worth including in a fork?
Protocol design · 8 participants · 14 quotes · 2026-10-15 ~ 2026-10-21
pcaversaccio criticized EIP-8037 and EIP-8038 for repricing all three elements of onchain privacy without analyzing the impact. fradamt of the EF, a co-author of EIP-8037, countered that the repricing is only a neutral correction based on resources. nero_eth found from measurements that gas usage rises, but that the larger gas supply could actually lower absolute costs.
Background
EIP-8037 (State Creation Gas Cost Increase) and EIP-8038 (State-Access Gas Cost Update) are Glamsterdam core EIPs proposed by Maria Silva, Ansgar Dietrichs, Francesco D'Amato and others. They raise the gas cost of state creation and access operations in order to limit state growth when the block gas limit is raised. EIP-8037 introduces a new variable, cost per state byte (CPSB), targeting 120 GiB of annual growth at a reference 150M gas, and EIP-8038 raises the cold account access cost and separates write fees (ACCOUNT_WRITE and STORAGE_WRITE). The authors cite as motivation that costs had drifted from resource reality for years. Both EIPs were at the Last Call stage and included in Glamsterdam.
Why it matters
If repricing changes the relative price of using privacy, the usage patterns of apps such as Tornado, Railgun and Privacy Pools could change, which is in tension with this year's roadmap that raised privacy to a priority. Because Glamsterdam was postponed indefinitely at ACDE #248 for security reasons, the repricing spec itself may be reopened, so it is worth watching whether a privacy impact analysis is reflected in the review.
Resolution
At ACDE #248 (2026-10-22), ChainSecurity found that the EIP-8037 and EIP-8038 repricing could let contracts be exploited through transactions that were previously invalid. The Hoodi Glamsterdam fork was cancelled indefinitely, and the mainnet schedule was put on hold until a full impact analysis is done. The security agenda item pcaversaccio posted the day before was part of that discussion.
Key points
- How does raising the cost of state creation and access change each of the three elements of privacy: new identifiers, commitment and nullifier state writes, and unlinkable gas payment?
- Is the repricing a neutral correction that aligns costs with resources, or does it create harmful externalities for privacy apps?
- Does raising the gas limit (200M) offset the increased operation costs so that the absolute cost of using privacy tools actually stays flat or falls?
For — Repriced all three elements without analyzing the privacy impact
Toni Wahrstätter ⟠, ً, sudo rm -rf --no-preserve-root /
- Onchain privacy needs three things: a fresh identity for each transaction, state writes for each use (commitments and nullifiers), and gas payment that does not relink identities. EIP-8037 and EIP-8038 reprice all three, and neither analyzed the impact on privacy apps.
- He did not stop at raising this concern on Twitter, and the next day put it directly on the ACDE agenda, citing security considerations other than privacy.
Against — A neutral correction of costs that had drifted from resource reality
sudo rm -rf --no-preserve-root /, Paweł Bylica, Francesco
- Pricing should be neutral and based on available resources and the demand for them, and the status quo, in which gas costs had drifted far from resource availability, was broken.
- EIP-8037 and EIP-8038 did not make state operations expensive in order to favor other operations. They corrected costs that had gone untouched for years and no longer matched reality.
- You have to choose among heavier nodes, higher gas costs and an unsafe protocol, and since state does not get cheaper on its own without new technology, you cannot improve every tradeoff at once.
Neutral — Gas usage rises, but the supply increase can offset the fees
Toni Wahrstätter ⟠, ً
- Replaying 53 real mainnet Tornado, Railgun and Privacy Pools transactions on Sepolia with EIP-8037 and EIP-8038 raised deposits by 1.2 to 1.85x and withdrawals to fresh addresses by 1.4 to 1.8x. A fresh address costs about 184k more gas than a reused address (25k today).
- That is gas usage only, and the basefee reflects demand relative to capacity, so if gas supply grows about 3.5x, the absolute cost of privacy protocols could even fall.
- The gas costs of these operations clearly rise a lot, but raising the gas limit to 200M increases supply by a similar multiple, so I expect fee costs to stay similar to today.