[CLOSED] Express Relay Prototype

Ideas Bank
22

This is an extremely exciting proposal.

Express Relay would be a revolution for DeFi liquidations - redistributing value back to protocols and providing fuel to supercharge the entire DeFi ecosystem!

5 Likes
23

Hey, tumilet here from Spartan. Really excited about Pyth’s OEV solution!

For context, we have been looking at OEV for a while and for anyone that wants an ELI5 explanation we wrote a post with the @decentralisedco guys a few months ago, for anyone interested reach out! (it doesn’t allow me to share links here).

In regards to Express Relay, I was wondering:

  1. Has Douro Labs done work on market sizing the OEV opportunity (from our internal calculations 90%-ish appears to come from Aave and Comp, which are protocols that are very risk-aware and difficult to see changing from Chainlink as a price feed provider in order to integrate a new OEV solution)?

  2. Does the Express Relay Prototype work only for lending and borrowing protocols, or also for perps integrations, if so what’s the opportunity there?

  3. Would be interesting to know what 8 DeFi protocols have expressed interest?

  4. Will Express Relay charge a service fee as a % of the OEV that gets recaptured, or what is the monetization opportunity for the PYTH DAO?

Thanks, and look forward for the discussion!

4 Likes
24

Congrats on this prototype!
I had several questions looking at this design.

In your picture it seems that searchers are just interacting with this express relay and not the chain itself anymore (as there is no arrow going from them to the chain).

This brings several questions up:

  1. How are searchers interacting with the express relay? Do they submit bundles with bids and the highest bid gets executed?
  2. Who executes these “won” bids? Is it the express relay?
  3. If it is the express relay (operator), who is operating this express relay?
  4. What happens to the protocols relying on this express relay for prices & liquidations if the express relay is down?

Thanks in advance!

3 Likes
25

This is Stanley from Caladan (caladan.xyz), we’re a VC and MM from Singapore. We’re active on DeFi and are keen to support new innovations and solutions in the space.

Our team has been actively participating in the active relay testnet on op-sepolia, and everything is working smoothly so far.

Caladan is in huge support of this solution, as we greatly welcome tools that align protocol incentives and cut out the value leakage to miners. Auctioning off liquidation rights to searchers is a terrific way to retain more value for protocols and their users. In particular, preventing sandwich attacks and resolving liquidations in a single transaction within the same block is potentially a game-changer, especially given how so much value extraction comes from block miners on Solana.

Looking forward to seeing more community feedback and we look forward to the mainnet launch once the prototype is approved!

4 Likes
26

Really looking forward to this feature, congrats on this prototype! and totally support this proposal.

1 Like
27

These are good questions. Sharing my thoughts below:

  1. A key design goal of Express Relay is that it is oracle-agnostic, so integration with Express Relay is possible across all apps (those that use different oracles, those that don’t use any oracle at all). Much of the market size in terms of TVL is concentrated in a few big protocols, but that doesn’t cleanly translate to liquidatable position size or liquidation volume. Aiming for the big protocols would be great of course, but 1) it’s also good to get integrations with some of the fast-growing protocols, many of which I expect to continue growing due to their innovations in protocol design and UX (such as integrating with Express Relay); 2) the biggest protocols require a more proven track record to demonstrate that the integration is reliable, and I think Express Relay will achieve that after launch once it’s helping to recapture value on mainnet.

  2. Express Relay is not lending-specific. It can also be used to auction ordering on transactions on perps protocols as well, for liquidations on perps but also potentially for other types of transactions (e.g. trading).

  3. Protocols that have expressed interest and provided valuable feedback on the prototype include ZeroLend, Ionic, and Keom.

  4. This seems like the most straightforward way to monetize for the DAO. If you or anyone else has any other suggestions, please do share.

5 Likes
28

Also good questions, here are my thoughts:

  1. Indeed, searchers submit their transactions to the Express Relay server. The server hosts an auction that is designed to choose the subset of transactions that maximize the revenue that flows back to the protocol, and it then forwards those transactions on to an RPC node.

  2. The “won” bids are forwarded by the Express Relay to RPC nodes for landing on chain. The Express Relay cannot alter the transactions but rather just forwards them on-chain.

  3. There is a party referred to as the “relayer” (or “operator”) that handles running the off-chain components of Express Relay (the auction, the off-chain server). The DAO could choose a community member to operate the relayer function.

  4. There are ways to protect against Express Relay downtime. One way is to make the priority for Express Relay in performing liquidations on a protocol conditional—the protocol developer could make it prioritized with a time constraint (e.g. only for the first minute after a position becomes liquidatable) or with a collateralization/TLV ratio constraint (e.g. only when collateralization ratio is above 108%). Otherwise, the protocol could make liquidation permissionless, to protect against Express Relay downtime and prioritize protocol health. The protocol won’t get any of the revenue it could have gotten through Express Relay in that case, but that’s totally up to the protocol to decide what constraints and fallbacks to set.

6 Likes
29

Very clear, thanks for the response! I agree on the go-to-market strategy of targeting the “longer-tail” of lending dapps which will probably keep capturing share away.

Look forward for mainnet launch!

3 Likes
30

This is so innovative. Sometimes I forget how deep some people like to go and what the real limits of projects are.

Fascinated to see how this plays out.

3 Likes
31

One would hope so!
I like experimenting with stuff, so it could be interesting to try this.
I find it better to use things to understand them anyway.

1 Like
32

Anything to share about Ionic incentives for PYTH holders? Should we be watching closely somewhere? Twitter perhaps

33

Watch Ionic Twitter! We are definitely looking to expand our partnership with Pyth and are constantly looking for ways to benefit both communities!

2 Likes
34

Great Post! A couple of questions

  1. Does the Relay Express host open, English auctions or private sealed bid auctions? May I know the reasoning behind the choice?

  2. What happens if the liquidator who wins the auction does not use its right to execute the liquidation and causes a delay in the liquidation process? Will the oracle have to release its price into the public mempool to avoid further delays on the liquidation?

  3. How long does an auction last?

35

Good questions.

  1. The prototype design features sealed bid auctions. This is not a hard requirement, and open bid auctions could be considered in the future. In the initial stages, a sealed bid auction is less of a shift from the status quo and allows searchers to maintain privacy until their transaction lands on-chain.
  2. A searcher submits their intent to transact in the form of fully formed transaction data which is relayed directly on-chain if they win. So there is no dependence on the searcher to actually submit their transaction on-chain after they win the auction, that relaying happens immediately. If for some reason their transaction fails on-chain (e.g. due to some state changes) that block, the auction could be reheld the very next block.
  3. In the prototype design, the auction lasts around the blocktime of the chain. This allows liquidations to happen quickly and not suffer from any delays during volatile time periods.
3 Likes
36

So, in effect you’re saying that this completely centralized and it’s up to the protocol to implement it in a way where it isn’t but risk “losing” all the benefits in the process?

37

Quite the opposite. The solution offers protocols a free option to earn rewards, and there’s nothing forcing protocols to integrate. They can and should choose to introduce safety fallbacks, in order to prioritize safety in case of the system going down. They don’t “lose benefits”; instead by integrating they realize only a net benefit over the status quo.

With regard to centralization, relayers handle the off-chain components of the stack for efficiency reasons. A relayer is appointed by the DAO and can be revoked if it performs poorly. Importantly, they cannot manufacture user transactions. The alternative of building out an auction on-chain or otherwise decentralizing the off-chain components of the stack are future considerations but right now are not the priority—protocols have put much more weight on value recapture as soon as possible relative to this element.

1 Like
38

So the auction duration on Arbitrum for example, would be 250 seconds or so? and 12 seconds on Ethereum?

39

Are there any fall-backs in place if the Express Relay is down? or if there are any delays in the auction?

40

With regard to auction duration, the design goal has generally been the blocktime, yes.

With regard to fallbacks, see the earlier response: Express Relay Prototype - #28 by Apollo

1 Like
41

Mainnet release is now up for vote:

2 Likes