@cairoeth released a PoC of preconfirmations leveraging restaking by using Relic’s Prover. It’s a pretty clever design so here’s a quick summary and some thoughts.

This implementation leverages EigenLayer’s AVS to power the slashing mechanism. Ethereum validators who wish to participate must opt-in via EigenLayer and run the AVS software. It allows to offer 100ms tx preconfirmation on top of 12s L1 finality

They forked MEV-Share, which serves as a trusted matchmaker between users and block proposers. Users submit preconfirmation requests along with hints about their transactions to the preconf-share node, which then gossips the requests to validators that are the next block proposers

Block proposers may be more receptive to requests that include logs or calldata information, even if they have lower tips, as it opens the door for value extraction or OFAC limitations. Conversely, requests with little information can be accompanied by a high tip to balance the lack of information and have a higher inclusion rate. 

Block proposers must also consider other preconfirmations they have made for future blocks to make sure that new preconfirmations don’t collide with transactions obtained from the global public mempool, so one thing preconfers should be careful of is not preconfirm too many txs containing too much gas, because if they do so they’ll be slashed

For the economic guarantees to be considered secure, the cost of corruption must exceed the profit from corruption. Under the current Ethereum design, stakers can have a maximum of 50% (16 ETH) of their total staked ETH slashed.

However, is important to note that in the current implementation, block proposers act as both block builders and proposers, so one area to further explore is how to integrate with MEV-Boost to allow proposers to delegate block building to specialized parties.

Some of the implementation details that we explored in the based rollup report were applied to the PoC, such as suggesting proposers stake additional ETH and that the slashing funds could serve as insurance funds to compensate affected parties, 

If you’ve read our based rollup report, you’ll notice that the mental model is very similar. As we mentioned in the report, SUAVE executors or Anoma solvers seem to be ideally equipped to provide preconfirmation services. This is because 1. they are highly specialized builders running expensive infrastructure and 2. they are already suited to offer cross-domain atomic bundles 

Although SUAVE is still under development, MEV-Share can be considered the initial step towards a fully integrated SUAVE. It provides users with programmable privacy and preference expression capabilities. Users can share any desired hints with the trusted relay during transactions, while also hiding specific data such as calldata, emitted logs, and to_address.