Frame Transactions vs. SchemedTransactions for Post-Quantum Ethereum

EIPs
15

A few issues I have with this proposal

  • Sender address is determined stateless, which means existing accounts cannot utilize the new crypto schemes. That is an almost certain guarantee that there will be very little adoption.
  • There is no support for authorization_list. That means if a user sets up an account with an R1 key or a Falcon key, they cannot later take advantage of AA benefits. We would need to extend the 7702 tx type or add a new 7702++ tx type to support the schemes idea.
  • No blob_hashes are supported, which means only K1 ECDSA keys can submit blobs. It doesn’t seem like the worst trade off, but it feel fairly incomplete.
  • The origin == caller issue was already broken with 7702, so it isn’t a new issue for the frame tx.

To me, the first point totally disqualifies the proposal since it is a very high friction way for account to make use of PQ, even if it is slightly cheaper.

–

I think we should once and for all solve the AA problems. Frame tx wasn’t pitched as the easiest solution, but it is a complete solution and forward compatible transaction type. We are addressing many outstanding complaints from the community beyond just PQ, such as

  • native batching format via frames
  • per-frame receipt for proper resolution of events per call and per userOp
  • default account to avoid delegation risk

A prix-fixe menu of crypto algos tx just seems too short-sighted. Yes the problems it solves, it does well. But it is too narrowly focused. And does not address many of the issues that do need to be addressed:

  • 4337 accounts cannot use FOCIL. We have the responsibility to ensure that 7702 delegated accounts native smart accounts have same inclusion guarantees as regular EOAs.
  • native batching remains undefined
  • gas abstraction continues to rely on centralized services instead of public mempool

Yes the quantum threat is looming, but I don’t believe it is so imminent that we should rush something out to paper over the problem. Frame tx is small generalization over 7701, which itself is an direct transalation of 4337 into the protocol. This work has been ongoing for over 5 years now and I think we have a good idea how this should land in the protocol.

1 Like