[AIP-27][Discussion]Sender Aware Transaction Shuffling

[thepomeranian]

AIP-27 - Sender Aware Transaction Shuffling

Summary

Sender-aware shuffling enables the shuffling of transactions within a single block so that transactions from the same sender are apart from each other as much as possible. This is done to reduce the number of conflicts and re-execution during parallel execution. Our end-to-end performance benchmark show that sender-aware shuffling can improve the TPS by 25%.

Link to AIP: https://github.com/aptos-foundation/AIPs/blob/main/aips/aip-27.md

[thlfranklin]

Asking from a research stand point: are there any analyses on how it changes the behavior / game theory and impact on MEV, for example?

[sitalkedia]

@thlfranklin - Since the shuffling is deterministic and tries to preserve order of the transactions if they are not conflicting, it doesn't change any behavior around MEV - it is still possible for validators to submit transactions in a certain order to get the expected output ordering.

[alnoki]

@thepomeranian @sitalkedia

Does the sender have any input onto the shuffling process?

E.g. it sounds like this is a quasi MEV vector that allows a sender to stipulate shuffling in the absence of a validator override

[sitalkedia]

@alnoki - The sender can't directly influence the shuffling process as it doesn't control what other senders will be included in the block. Validators (block proposers) on the other hand can of course propose a block to support MEV - this is true with or without shuffling.

[alnoki]

The sender can't directly influence the shuffling process as it doesn't control what other senders will be included in the block

Got it, so I understand then that the term "sender aware" just means that shuffling is informed by the sender

Noting in this case for high-profile events like NFT drops, that senders trying to send in a bunch of bids will have unfavorable ordering, while actors who intentionally distribute bids across multiple accounts (e.g. one bot multiple senders) will be advantaged.

So this approach massages the block in order to make Block-STM have fewer collisions, but also introduces a new method for savvy actors to outcompete less savvy actors