Encrypt The Mempool #002
Transcript
Ask about this call
Answered from this call's summary and transcript only.
AI-generated — verify against the recording.
Call summary
Highlights
- Proposals:
- ·soispoke published encrypted frame transactions proposal with same-slot interleaving — 00:03:59
- ·Semantic block chunking proposed: regular txns, then encrypted, then optional backrun — 00:32:30
- ·Primev proposed builder optionality for same-slot inclusion with 80ms decryption time — 00:47:26
- Eip Status:
- ·Lucid design assigned EIP-8184 — 00:02:36
- Design Questions:
- ·Per-transaction vs per-block backrunning efficiency debated; trade-offs discussed — 00:09:41
- ·Interleaved plain text/encrypted transactions vs separate buckets explored — 00:14:36
- ·Same-slot vs two-slot designs orthogonal to interleaving decision — 00:26:00
- ·Frame transactions compatibility with two-slot Lucid design unclear — 00:42:02
- ·Fork-choice timing concerns raised for encrypted mempool designs in 12s slots — 00:50:25
Action Items
- •Luis — Share Yannick's AMM simulation model and repo for evaluation — 00:08:34
- •Anthony/Working group — Explore visualization frontend for demonstrating sandwich prevention — 00:09:01
- •Murat (Primev) — Create write-up on same-slot builder optionality proposal — 00:50:40
Meeting chat42
- soispoke
gmgm
- Anders Elowsson
HI
- Anthony
👋
- Stefan Starflinger
Pk is working on a builder for epbs depending on the timing we could consider building on top of that
- Stefan Starflinger
Replying to "Pk is working on a b…" https://github.com/ethpandaops/buildoor
- soispoke
Practically there are definitely ways to interleave plaint text and encrypted txns (this is part of the encrypted frame txn design I published) https://ethresear.ch/t/encrypted-frame-transactions/24440
- soispoke
Replying to "Practically there ar..." This doesn’t change what Anders is saying about needing to use onchain searching
- Justin Florentine (Besu)
yes, but is it meaningful if their order was committed already prior to reveal?
- soispoke
Well it avoids competition for doing the 1 backrun at the end of all encrypted txns
- soispoke
What are the costs and benefits in your opinion?
- soispoke
(Interleaved between separate buckets)
- Potuz
if they could interleave txs they could signal the other ones on how to branch isn't it? you'd always want to be committed on one of the encrypted one and then reserve the right to execute it or not or pick branch of execution by interleaving the trigger.
- soispoke
Replying to "if they could interl..." The idea would be to commit to both plain and encrypted txns at the exact same time (if even if they are interleaved)
- Potuz
Replying to "if they could interl..." as long as the full ordering is committed that could work.
- soispoke
Reacted to "as long as the full ..." with 👍
- Gottfried Herold
I mean you could have interleaved txs in soispoke's design also for Lucid, but with the difference that essentially the builder only commits to the ordering (but WHICH encrypted transactions have to go through is still determined by a FOCIL-like mechanism)
- soispoke
Replying to "I mean you could hav..." Yes that’s the difference, which to be fair is a big one (because it changes the way ePBS commits to contents/ordering) but I think it’s worth considering
- Potuz
is it me or Anders is the only one I can't hear?
- soispoke
Replying to "is it me or Anders i..." I can hear
- Justin Florentine (Besu)
Replying to "if they could interl..." ePBS committing to something a little looser than a stateroot seems like a powerful unlock
- soispoke
Replying to "if they could interl..." I agree
- Potuz
Replying to "if they could interl..." how is the commitment going to be? the bid sends the ordering?
- soispoke
Replying to "if they could interl..." But just not something to take lightly
- soispoke
Replying to "if they could interl..." @Potuz yes
- soispoke
Replying to "if they could interl..." In the figure above you just have tx_ordering_root in the bid
- soispoke
Yeah I agree, frame txns cleany separate validation and execution which is great for 1 slot but weird to think about for 2 slots
- Anders Elowsson
Can’t we just have a sidecar to a ticket inside a frame tx
- Potuz
Already FOCIL kinda kills shorter slots anyway
- soispoke
Replying to "Already FOCIL kinda ..." No I don’t think it does!
- soispoke
Replying to "Already FOCIL kinda ..." Probably not the right breakout room but happy to chat more about this
- soispoke
Replying to "Already FOCIL kinda ..." The tldr is: just do a separate BAL deadline
- soispoke
Replying to "Already FOCIL kinda ..." That reveals it earlier
- Potuz
Replying to "Already FOCIL kinda ..." I think you guys aren't equally afraid as I am with split views. Already ePBS is borderline crazy timing :)
- soispoke
Replying to "Already FOCIL kinda ..." Haha ok then let’s frame it as ePBS kills shorter slots 🙂
- Potuz
Replying to "Already FOCIL kinda ..." All the designs that are being thrown for encrypted txs seem quite impossible to me from the forkchoice perspective
- Potuz
Replying to "Already FOCIL kinda ..." with 12 seconds even
- soispoke
Replying to "Already FOCIL kinda ..." Ok yeah tbh I don’t have a good view on how much time it would take
- soispoke
Replying to "Already FOCIL kinda ..." To release keys etc
- Justin Florentine (Besu)
Replying to "Already FOCIL kinda ..." ssf bro, 12 seconds are fine with instant finality
- soispoke
Replying to "Already FOCIL kinda ..." 3SF is more the vibe these days (but still very nice)
- Potuz
Replying to "Already FOCIL kinda ..." gotta jump meetings, a pleasure seing y'all
- soispoke
I have to go too, thanks for the discussion!