Sybil Attacks on Inclusion List Mechanisms: FOCIL and AUCIL

Transcript

Awesome. [music] Hi everyone. Uh so sorry for the wait. Uh this is my talk on civil attacks on inclusion lists. Uh really nice addendum to the law uh the talk right before this. Um so we're going to be talking about two separate uh inclusion list designs which is fossil and uh so inclusion list mechanisms under PBS there's sensor resistantship tools that allow validators to mandate the block builders to include specific

pending transactions in the uh and if a builder's constructed block ignores these uh transaction rules then uh the network rejects As I said, the two different uh designs we're going to be looking at today are going to be Fossil and OEL. Fossil, really good explanation by Marissa right before me. Uh it's a consensus level uh inclusionless mechanism. There's no explicit pricing, but an alternative uh inclusion list mechanism came up in a paper by Vadwa at all. uh I think it was like almost a year ago where they talked about uh an auction-based design of inclusion lists that actually leads to price formation.

So this is a very simple uh this is taken from a post on ETH research one of the very first posts that came up on inclusion list mechanisms for fossil. Uh basically what happens is that again as I said Marissa did a really good job of like uh going all over this. So I'm going to quickly go through it. What basically happens is during slot n there are 16 selected committee members who gossip their partial inclusion state and the block builder at slot n plus one must pull the transactions from those aggregated lists. Then uh attesters they strictly uh they strictly attest on those new transaction states and they are able to um what's that called? they're able to

reject the transaction states or the transaction batches in case uh they don't form the specific rules uh which are attestation rules folk choice rules right but this is the newer uh other alternative inclusion list mechanism that I want to talk about which is a uh basically proposers monitor transactions in the meool to construct the inclusion lists uh they the lists are then signed and broadcasted across the network to be considered for inclusion inclusion lists are then evaluated and comb combined into uh bids alongside validate accounts and what happens is that the top candidates are selected from the final bidding inclusion list. How that happens it's kind think of it like more of a first price auction where the highest uh bidded inclusion list is included in the

protocol. You might worry about censorship resistance in that case and you might also be wondering about the the benchmark and the differences between uh censorship resistance in Fossil versus OEL which is what I would like to talk about next. So Fossil is inherently civil proof but not bribery proof. I feel like that makes sense but intuitively because Fossil is committee based uh it's completely on the the validator level but OEL on the other hand is not but at the same time Fossil is not bribery proof. Uh a really good paper by Stuka at all came out uh I guess last year where they talked about multi-proposal transaction fee mechanisms. They specifically looked at the bribery bounds of um fossil and similar uh

inclusion list mechanisms. Uh this is the primary formula that comes out of it. Uh when you think about it this way, uh this is basically the difference between uh including the list in the honest conditions minus the bribe it would take to uh take for them to become adversarial. On the other hand, OAL is not civil proof but it is boundedly bribery proof. If you think about it, uh the lema 12 in the OAL paper by Wadol talks about how inclusionless follow that uh if they follow a bribery strategy. If the validators follow a bribery strategy then the total cost incurred by the adversary is at the least uh is scaled by the fraction of adversaries into

aggregate utilities of the producer. So this is the newest work that I did on civil attacks on oel. Uh it was basically a blog post hopefully will become a paper someday. Who knows right? Uh the next thing so basically I set up uh a very simple formula of oill where the utility that you get from being a single proposer is going to be greater than the utility you get from being multiple proposals together. Uh consider an example where transactions X and Y that are included in inclusion list. uh once they are included in the list they pay 10 and six respectively and that 10 and six is shared equally amongst the contributors. The initial allocations say proposer one proposes X, proposer two proposes X as well. Proposer three

uh proposes Y. The rewards are five, five and six. Um alone P2 would not switch to Y. Then its reward falls from 5 to 3 if you think about it. But an attacker controlling both P1 and P2 switches P2 to Y and the total reward rises from 10 to 13. Uh finally I have last two results. You will see that under the civil case the censorship resistance scales uh is the censorship resistance bounds was weakened proportionally to the number of civils. Similarly the bribery bounds are scaled sim u the bribery bounds are scaled as well where it's lesser it takes less to bribe the uh proposers in oel uh as it would have been in the

normal uh non- civil case. I have two open questions that I wanted to deal with which is can we make can we use stake based partitions for fees to improve the correlated equilibrium the greedy algorithm that OEL has proposed or can we improve the scoring auction design which is basically the first price auction that I told you about right in the beginning to uh deal with the civil proofness case uh can we use a statement [music]