Revenue and Pricing Models for Preconfirmations

Transcript

Okay, perfect. Uh, so welcome everyone. Uh, my name is Connor. I'm a protocol researcher at Nethermind. And today I'm going to talk a little bit about the revenue some revenue and pricing models for pre-confirmations. Uh so this is a result of a lot of work that we've been doing over the last two years with never mind. Uh so yeah it's going to be a lot of stuff so strap in. Uh it's going to be quick. Um okay so the talk outline high level introduce some pre-conf protocols and analyze the expected revenue of each. Uh okay so sounds easy but uh we'll see that it's a little bit trickier. So if you've just stumbled in uh quick quick uh recap a pre-conf uh is a basically when you have a proposer who knows they're going to be

a proposer ahead of time instead of waiting to build a block at the last second they can offer a user a promise ahead of time. Uh so that that would be a pre-confirmation. So in this case preconfer slot n plus2 has exclusive rights. So instead of waiting for slot n plus2 to arrive the user ask the pre-confer to give me a pre-confident and this promise from the pre-confer is the preconf. Okay. So uh the classifications that I want to talk about today or at least the classifications that cover most of the pre-mp protocols that exist today uh occlusion only pre-MPs uh so again proposer promises the user that the the transaction will be in the block somewhere but not what it will do where it will be or if it will even execute.

Uh execution premps. So this is a much more powerful pre-confirmation uh where the proposer says okay I will put your transaction at this exact place in the sequence and this would potentially be the out the exact output of your transaction. Uh so within that realm of execution premps uh we just we come up with this these two types of pre-confirmation protocols that we think summarize most of them. Uh so there's independent slub slot auctions which are being run today in Jaido and Salana and uh optimism where you have one proposer who agrees to just instead of waiting till the end of a slot where they have let's say 20 seconds to build they build something every two seconds or 4 seconds. Uh then we have dependent subsot auctions which are less are not

so common today but I think are a real possibility and something which are uh potentially very dangerous for blockchains. Uh I'll get into them more a little bit later on but like these gateway designs if you're familiar with the pre-MPS are probably going to look a lot like the the penetent subplot auctions. Um and and then tao which netine are working with closely we're probably going to expect to see a mixture something between ISSAs and DSSAS. Now what exactly the mixture will be will be revealed in the next couple of months when Tao goes live. And then there's a different type which is builders signing for pre-MPS. So again when we think of pre-mps we think of the proposer who has exclusive rights. Well, it's possible that builders can also offer pre-confirmations, but again, those would be probabilistic. So, not guaranteed that they will they will be

included, but for a good for a good builder, this is pretty reasonable guarantee for a lot of users. Uh, if I win, I will put your transaction in. Okay. So, the the focus of today's talk will be on these three three types. Inclusion only pre-mps and then execution pre-mps, the the independent subsiductions and the dependent subsiductions. Okay. Uh, so yeah, we're going to look for looking at the revenue here today, right? So inclusion pre-mp revenue. Uh TLDDR, we're going to model the tips that are needed for pre- comps. If we can model the tips, then we can model the revenue, right? Um so this is actually much more powerful than coming up with some vague revenue model. If we can predict the exact tips that are needed, that's a very powerful uh thing for a proposer. Okay. So how how we did this, we basically took a lot of MEV

boost data uh and we just said, okay, how are the tips changing as we go further and further back in the block? and and we get we got this very weird shape which we we expected but again just to explain it uh at the top of a block people are paying a lot more for their transactions. So we're getting we're getting this steep uh rise in the in the total reward of a block. Then as we go further out into the block as we get closer to 30 or 50 million gas things start to level off. People are paying less for the transactions there because it's less valuable block space. You're not getting the same guarantees of let's say a first access to to a sec to a DEX. Um so again this little graph is this this line that we plotted through uh these rewards that we are

seeing is going to is going to form the basis for the inclusion pre-MP pricing model that we come up with. Um okay so we took that line the curve that we saw there and we said okay what this curve means is like when you offer inclusion pre-confirmation you're going to put that transaction in the block somewhere. Where is the worst place for most users? at the bottom. Um, okay. Sometimes the users want to be actually at the bottom after everybody else, like a back run. But most of the time, if you want to say, "Okay, I'll put your block in there somewhere." You're probably going to say, "Okay, well, if I'm going to put it in there, I'm going to put at the bottom." So, how we did this, what this graph shows us is how to come up with a tip for a specific inclusion pre-MP. So

here this what we've what we've graphed here is um okay if there was 13 million gas already pre-comed and someone wanted a 2 million gas uh transaction to be pre-confed we just basically come up but the curve from the 17 million gas point and the 15 million gas point and that distance on on the on the y- axis is the tip the fair value tip that we've predicted using our model. Now our model was based on like a very small sample size of of data and what we see in in block in in in sort of revenue for blocks is that like it's very seasonal depending on the time of day Asian hours versus American hours the block rewards will change a lot and we expect the distribution the curves to change a lot

too but based on the data set that we used this is what you would do you would basically say okay someone wants a 2 million gas pre-mp I've already given 13 this is the fair value okay so then okay we have two inputs we have the amount of gas that's been pre-mped and we have the amount of gas that someone wants a pre-mp for. Uh so we basically we plot that on on two dimensional space and then we come up with the the curve that we basically use that curve to get the threedimensional shape the surface that we see here and this basically gives us a expected pre-com tip or a fair pre-com tip for any type of pre-MP given any amount of gas already pre-confed. It's a very it's a very simple model but very effective.

And what can we do with this? Okay, a proposer who doesn't know much about block building uh can actually offer pre-mps very straightforwardly. Uh so they can just draw a little blanket over the top of this top of this uh surface and say okay if I charge touay this sh this this shape covers most of that surface that that we that we came up with. What you can do with that shape is say okay I'll I'll offer a touay pre-mp tip. If someone gives me a tou pref tip and there's only been there have been no pre-mps offered already I'm happy to offer a pre-mp up to roughly 20 million gas. All right. Um now again you can draw a smaller a smaller uh surface say okay if I want to be competitive here I want to do one way you can still cover most of that surface so even a very

naive proposer can offer pre-mps pretty safely uh and then once you offer this inclusion pre-MPS you can then offer the you can get engagement boost and get a builder professional builder to build a block on top of all those inclusion pre-MPs that you've already made. So they they can put them at the back, the top, wherever they want, but the high value block uh is still available to that propos to that builder. Um what this means is inclusion pre-MP pricing is easy. So we expect inclusion pre-MPS to to happen to emerge very quickly. Um and this is sort of why we've seen what the inclusion pre-MP protocols that are in the pre-MP protocols that are in production are all inclusion pre-MP protocols. Um but execution pre-MPS are where the real meat is where people really think that the UX will change.

And this is maybe the more complicated part of the presentation. So we're going to analyze the expected revenue of the execution pre-MP designs. Uh we're not going to necessarily be able to handle the tips because measuring the tips is extremely difficult like very conditional on what the the conditions in the markets are, what exists in the me poolool, what type of builder it is, what the risk profile is. So we're not going to look into the tips, but we're going to try and reason about the revenue of those two u protocols that we mentioned earlier on. these ISSAs and these DSSAs. Now, I haven't really properly described these protocols. Um, so I think this this graph is helpful, but I've been in the weeds for so long that I'm not even sure if it's relevant.

But um, what we have here is ISSAS versus the SSSAS. Now, when I think about ISSAS, you have Pete the proposer who has 20 seconds to to uh to build a block. Instead of waiting 20 seconds, he decides to build sublocks every five sec or four seconds. There's five sub slots that we see here. So he offers he runs an auction every 4 seconds and we have these various different winners Bob, Alice, Alice, Carly. Um what they're doing basically is they're trying to build the most valuable block at that time. Um not thinking about like what would be the highest value block for the entire 20 seconds. So you have these sort of like very short periods of of of time where people are racing to capture every opportunity that exists without thinking about letting that opportunity

grow to maybe something even bigger. In a dependent subplot auction, we're trying to capture that concept of letting things grow. Like, okay, I'm not trying to buy the most valuable build the most valuable sublock at this 4-second instant. I'm trying to build the sublock at this 4-se secondond instant that is going to maximize the total revenue at the end of that 20 seconds. Uh, so it's a little bit more complex. How we m how we model that would be Bob building a sub. In this case, P has the original uh rights to to build the entire block at the end at the end of the 20 seconds. Bob wins the first auction, he builds something, but he gets to gain the revenue for the remainder of those subs slots. Um, so in that sense, what Bob will build at that 4 seconds will be the block that

maximizes the value of the entire uh slot. So what he will do there, he will probably not include sex arbitragees, but he will include pre-con like uh very sort of uh non-contentious transactions that are paying tips uh tips that he knows that he'll be able to capture no matter where they are in the block. uh and that won't basically compete with his highv value sex opportunities. Um okay so estimating the revenue of the ISSA is the sort of relatively simple uh protocol a relatively simple uh brother of of this ISSA DSSA pair. Okay so we know the block reward composition for 12 second slots. What does that mean? Okay, so we have a couple of different uh projects that have been trying to uh

measure like how what type of transactions are making up a block and Sorella come up with this like breakdown of what they believe the transactions are doing. So they have sex decks sandwiches remainder which are unclassified um they have JIT sandwiching they have atomic arbitrage. Uh so with each of these different segments we're going to basically try and think what will happen if instead of having a 12-second slot we'll have a six-cond slot or a 3second slot. How will we expect the revenue of of each of these things to grow or change as we as we change the slot time? So again, as we're as I mentioned that second point, then we're going to find curves for each component. So we're basically going to measure each we're we're going to come up with a predictive model for each of those components um as we change this

slot time. We're going to merge those curves after the fact and then we're going to say okay with this merge curve it should approximate the revenue from an ISSA uh sort of protocol. Okay, so I've just taken the sex arbitrage example because it's relatively interesting and we're using results from people that are much smarter than me uh or or my my co-author which says basically that sex me grows um super linearly in time like the longer the slot like you expect your your profit to actually increase superlinearly t to the^ of 3 over2. What that means if you reduce the slot time is that you expect that similar like similar curve to happen which is actually dropping off like faster than linear h which is bad for a proposer who's trying to capture that revenue.

Not only that if you have let's say three three sub slots or six sub slots every time someone captures that revenue they're going to pay a base fee and proposers don't capture base fee. So every time someone pays a base fee, if they instead of paying one base fee at the end, they pay three or four or five base fees during a slot, that means that there's three or four or five times less fee being paid to the proposer. Uh which is like a significant drop off and basically we're getting this very very steep curve which um appro like basically goes to I don't know it's very close to zero for proposer. So in that sense they really don't want to run an ISSA unless they're getting compensated for it. Um, so we had all those components in that first graph. When we merge them all together, we get

something that looks like a 60% drop off in proposal revenue from doing pre-MPs through independent subsidu. Again, independent subset auctions are like one of the most popular like examples of pre-con protocols that exist. So there is a big amount of money being lost or money to be made by uh going to like me boost or going to dependent subset auctions. Um, so like why do these things happen? pre-conf revenue and other examples, other other reasons why I'll get to at the very end hopefully if we have time. Um, so again, if you want to do pre-mps, you're giving up a lot of money. You have to take a big leap of faith that you're going to get compensated through pre-MP tips. Um, which, uh, again is is is something that we haven't seen just yet. It's going fromsas to DSSAs. Um, so

I've sort of hinted at it already, but again, like my last attempt to try and visualize it for you. Every time you run an ISSA auction, you're not thinking about the end. You're not thinking about the the most valuable block that you can build at the end of the slot. You're thinking about the most valuable block that you can build right now. So, you have these blue like like hard blue lines at the bottom. Like you have this like reasonably high value block sublock, but at the top like the maximum value that you can achieve at the end of the sub at the end of this whole slot is dropping off. Every time you build one of these high value blocks uh sublocks, you're reducing the total value that you could have achieved if you had just waited till the end. Whereas compare that to DSSAs where they are building

slightly less valuable subblocks but they're keeping this value of the entire slot at the max or close to like basically the max like that's the standard that we're using. That's the thing we're comparing to. Uh so like relative to DSSAS ISSAS are reducing the value of the slot. Um so I these blocks what the value means it's like okay if you had the opportunity to build that sublock how much would you pay for it? paying less in the DSSA for the sublock itself, but the whole slot is actually more valuable and and and sort of this like you have this steep steep growth at the end of of the of the of the DSSA which is basically at the very end of the slot you're doing all the sex decks you're doing all the contentious arbitrage you're doing all the contentious me transactions and capturing everything at the end. Okay,

so a very brief lema on why this works. Basically, whenever you the the strategy that you take as a builder in this in each of these DSSA auctions is to build a sublock that maximize revenue. I've said it a few times, but this is just putting into a very informal uh element. Okay, so putting all of this together, right? So, we have we have the revenue of um the revenue of the of of each of these pre-com protocols. Okay, that's only one of the the components that these pre-MP protocol developers, these roll-up users, the rollup uh developers themselves are thinking about when like they're trying to get people to do pre-mps, right? So, MEB boost as the standard pre-mp the proposal revenue is

flat like zero with MEB boost plus inclusion pre-MPS because it's easy to do inclusion pre-mps. We expect the profits to be slightly higher for proposers. For ISSAS, we've already seen it's a 60% drop off without tips. So that's that's bad. Uh DSS is good because they're again they're choosing the strategy that's maximizing the value of the block. So it has to be better, right? It has to be the best. And we actually prove it's the best in in the in the corresponding article uh which we we did with Lido. If you search pre-MP for revenue Lido H, you'll find all of this this this information all this information in much more detail. Uh in terms of user experience, MEB boost bad because no pre-MPs, right? But if you add inclusion pre-MPS better. If you add execution pre-MPS best uh because you're getting again almost instant

pre-confirmations or something very close to it. Uh I'll come back to the DSSA column because users users who send the pre-MPS are only one part of the the puzzle, right? There's also the people who are let's say providing liquidity on on DEX's or people who are like I don't know the app developers. It might not be so good for them. These DSSAs pricing ease. Okay. So now we're starting to see that DSSAs aren't necessarily the best. Uh because pricing ease, it's actually very very complicated. It's very centralizing. Um spam prevention non-existent unless you're very very sophisticated. H whereas ISSA you can see okay ISSAS and DSAs. They both offer execution pre-MPS.

Although ISSAS are worse for the proposers, they're much better for everybody else. Um, so potentially they might be about me a more realistic uh pre-conf protocol, execution pre-com protocol. Uh, but how do we stop proposers from saying, "Okay, I'm not running an ISSA, I'm running a DSSA." Uh, we have a few ways that we can do that. Um, okay. So, as I mentioned, user experience for DSSAs, it might not be so good, potentially even terrible. Uh, okay. So, pre-mps are profitable. Um, going back to this, we see we saw that the the prop the revenue from MEB boost plus inclusion pre-MPS or DSSAS is positive. It's better than ME boost.

We expect these things to happen. Okay. So, I wasn't sure how much time I would have for this. So instead of going into a lot of detail, here is just a summary of some of the things that we can do to actually protect against uh not protect against build our protocols in light of the fact that people will probably start offering pre-mps. Um and that if we don't keep track of them, though, if we don't keep our tabs on them, these protocols could slip into one or two people doing all the pre-MPS across all the protocols and offering users potentially very very bad experience in the long term.

Okay, so the different I presented this um at a present at a at a conference a couple of months ago. The only thing that's changed since then uh is this topline which is an overseer to enforce subjective behaviors. And this is actually a really powerful tool um that I come up with with the with my co-author in the back of the room here uh Lynn. And basically with an overseer the if we control who can who can actually offer pre-MPs someone like let's say if we talk about Tao who we're working with closely if they have um a governance controlled watchtowwer that doesn't do pre-MPs they're not very sophisticated but they're just sitting there looking at the user looking at the LPS looking at the memp pool looking at the blocks and saying okay pre-MP one is is doing an ISSA clearly because they're

not trying to maximize the value each each sublock but pre-MP 2 is actually clearly let doing very bad things for for like the dexes because they're leaving all of the arbitrage till the very end. They're not executing user orders that are trying to interact with the dex. Um meaning that the only user transactions that are getting in are actually inclusion pre-mps which is like not the purpose of the pre-mp protocols that we came up with in the first place. an overseer controlled by governance can come in and say, "Well, pre-MP number two, that's not going to fly. You're out of the system." Uh, black called blacklisting. And and with this like very basic watch tower, they can prevent a a pre-coner from doing something bad like like a DSSA. Uh, which again I think is is something that's probably going to be bad for users, bad for LPs,

and bad for blockchains in general. Apart from that, we have incl I think pushing for inclusion pre-mps now, execution preamps later. Pre-mps, the execution pre-mps are going to happen. H this was a bit of an old slide. Um, open source interoperable designs, no vendor lockin, making sure that pre-MPS like pre-mpers can switch in and out. Uh, permission list as much as possible, driven by research, keep me in the job. Uh, inclusion lists, making sure even if the pre-MP protocols go bad, we can still get transactions in there. Um, tested proposal separation. So, proposers offering pre-MPs is a very centralizing force. having like sophisticated proposers is one of the potentially one of the best ways that we can like protect against the bad effects of proposer centralization and um yeah what others are there I'm not sure but I

encourage you all to to reach out to think about yourselves and and think about the next solutions uh so yeah research ongoing get in touch and that's my Twitter if you want to reach out um always DMs are always open thanks very [Applause] I think we have time for one question. You have a Hi. Uh just wondering what's the role of Nethermind.

Um well I guess the rolling out the mind we were given free reign to uh to look into whatever we want maybe like when I started at least and one of the first articles we wrote was in pre-conf articles maybe half of which more than half of which have been funded by Ethereum Foundation lighter research grants. Uh so again we have a research team we're just trying to like push the push the narrative keep things as neutral as possible and because there's so many non-neutral projects potentially coming along the pipeline we think that it's a really like important job to take there. We're also working with tao um we never mind have uh surge running blockchains as as a blockchains as a solution service like we're also looking into many different types of pre-mp

protocols. So we're using our research like edge to like push an engineering edge to push like thing pre-mps in the right direction hopefully. Uh so yeah we're we're fingers a lot of pies in the pre-mp space. Uh yeah. Yep. Any other question maybe? Okay. Thank you very much. [Applause]