Ethereum Sequencing and Preconfirmations Call #13

Transcript

okay uh welcome everyone uh to uh sequencing call uh number 13 the topic here is uh precom uh tip pricing um I think this might become a series uh of calls because it's you know such a big topic and we're started to to go down the the rabbit hole and um really what I wanted to to do here is uh you know summarize some of the the thoughts that I've had in the last few weeks as I've gone down the the rapid hole and uh I don't think what I'm presenting is like 100% you know fully baked and and and bulletproof and I you know I'd love to see the the discussion that ensures from it uh but at least I hope that it is you know directionally correct or at least a

possible Direction uh for for for tackling um preon uh tip pricing so without further ado let me share my slides uh I shared the link in the in the telegram uh chat okay so uh pre-com tip pricing uh the background picture here is a a black hole in the middle of of a universe and we have this jet coming out and hopefully it'll become clear uh what the metaphor is here feel free to ask ask questions as I as I go

along so the the trdr the the summary slide is basically that um I'm trying to achieve the same thing that we have for gateways um sorry the same thing that we have for relays for for for gateways um so in the context of of of relays and and PBS and me boost we have this very simple uh auction where basically where we have the auctioner the the relay uh take the the maximum bid across all the bids that were provided by the by the builders and really this is like super simple from the point of view of the of the auctioner and all the magic all the complicated mathematics and modeling and

simulations and whatnot is happening by by the builders and what I'm hoping we can do is do the same thing for the gateways except that we replace the max by A Min and instead of having block bids that come from Builders we have so-called pre-com bits from from prices and so the yeah the basic idea is that uh we have different prices competing against each other uh to provide the the smallest bit uh possible for for sorry the smallest possible pre-com tip for for the users so the the talk will be in three parts first of all I want to highlight the the problem which in some sense is much more General than pre-confirmation

and this is good motivation because it's something that you know the space is going to have to need to tackle regardless of pre-confirmation um then I'm going to talk about some of the design goals uh that I have for um the uh the the the Gateway auction um and then uh three I'm going to talk about uh actual construction someone's microphone is not muted let me mute folks just a second okay okay let's start with uh the the

problem so in framing the problem I I'm you know suggesting two new terms one is the the window and the other one is the the sacrifice um specifically The Proposal window uh and the commitment sacrifice so um today what we have is this this mental model where there's one slot uh one proposer per slot and basically one commitment per slot which happens at the very end of the slot um and this is the the proposer window it's it's the the amount of time during which the proposer has Monopoly power uh to to to sequence and and order the transactions and we basically have this

this onetoone correspondence between slots windows and and commitments but the the basic observation is that this one to run correspondence actually breaks down in in in various uh situations situation number one is when you have multiple slots in a row so you have one uh operator on ethereum for example that controls two slots in a row and so the window is now uh two slots and and we've broken down this one to one correspondence and if we take an operator like Kiln for example which has about 5% of eum validators and roughly 5% of the time they're going to have uh two two slots in a row and they also going to have

these two commitments and I guess the reason why I I highlighted Kil in particular is because they're they're fairly large but also fairly sophisticated and and willing to do experiments uh to to to that are that are rational and and and so it's possible that we're going to see some some multi slot stuff at least at some point in the future uh another very interesting example is Solana where um you know without with even without the random randomization basically they have this this this mechanism where um every proposer when when selected is given four slots in a row uh and each slot is 400 milliseconds and so the the window

uh you know the minimum size of the window is 1.6 seconds it could be you know multiples of that so it could be 3.2 seconds for example um if one large operator was selected twice in a row but at least it's is 1.6 and you know you have this 4: one ratio between slots uh and and windows but things actually become even more interesting with with Solana because they have this this notion of of ticks So within one slot um they have 32 ticks each tick is I believe 1.25 milliseconds and and they commit to the tix in real time using this cryptographic um function called the a proof of history or a delay function and

the idea here is that you know you need to feed in the So-Cal shreds which are like shreds of blocks into your delay function um and it's difficult because of the laws of physics to go back in time and recompute the delay function on on a different input so effectively you you you're making this these commitments uh within the slot and so not only have we broken the onetoone correspondence between slots and windows but we've also broken down the onetoone correspondence between uh commitments and and slots and you know because Solana in some sense is like so ahead of of of of of uh of of the industry in in in this respect it's possible that they're going

to be faced with this uh pricing problem uh before everyone else and you know the the the bll case here is that Solana does all the work to try and and uh and and address this this this problem now um you know relevant to this talk is is precom so proms also break down the one to1 correspondence between commitments and Slots uh where we have lots of pre- confirmations that happen uh during during the slot and you know pre-confirmation are not unique to to to to base rollups or ethereum they can be applied more generally to to leader based you know consensus mechanisms where there is a look ahead so where you know who the next uh proposer proposer

is so that's basically almost all blockchains with a few exceptions like Bitcoin for example where because of the proof of work mechanism you don't know who who the next leader will be okay so that was the the the first um you know concept of of a window but the other one is the the notion of a sacrifice so when you make a commitment which is intra window so not at the end of the window but within the window you know you have an incentive to make that commitment so you're receiving some sort of tip and the the the question from the point of view of of the proposer here is to ask themselves you know does the tip here compensate for what I call the

sacrifice which is the the opportunity cost here of constraining yourself um here so you know if you make a commitment here by the time you need to propose your block and make the final commitment your hands are tied to a certain extent so you're constrained and that could be you know negative there could be you a negative externality for for you and so you kind of need the tip to compensate for this for this for this uh constraint and like two examples of of sacrifice uh are you know related to to congestion and contention so let's say that um you made a commitment at this

point in time within the window to include a specific transaction at x gay per gas um and you know you think it it was a a good decision because you know more likely than not the block is not even going to be full and so you know it you might as well just include transactions but it turns out that you know it was you know one of these blocks where there was a lot of congestion that happened in the in the next few seconds within the window and so now you're at the end of the slot and you realize that had you not made this commitment you would have had the option to include a transaction which is paying uh a higher uh G gas price and so here in some sense you're you you're you're

losing money relative to the counterfactual of not having done the the commitment and there's a very similar thing here with with execution where you know there's some sort of uh price in in in in of of e for example in US Dollars you can go buy or sell assets you know on on Unis Swap and then you know you you for example do a trade here and you realize that it was the the wrong trade to do because the price moved uh against against you okay any any questions on this on this first section before I I move on okay great so um part two uh um goals um

so there's there's various people involved in the in the in the auction um and let's let's look at them uh let's look at the guarantees that we can provide them uh one by one so first let's look at the the the user and the wallet on on the one hand so um one of the big guarantees that we want is for the the auctioner which is the the Gateway in this case to to not front run uh the the the the users whenever a a pre-confirmation request uh comes in and we want both uh to prevent the onchain front running um but also the the the offchain front running so we uh for

example if uh if the price onchain and and offchain of an asset is the same um and there's a a a pre-confirmation request that comes in that changes the price on chain we don't want um you know the the gateway to go do the Arbitrage basically the the off offchain instead we want the the proceeds of the Arbitrage to to to go back uh to to to the user the other thing that we want is some sort of competitive pricing um right we don't want to be in a position where the the Gateway can just you know single-handedly you know do price gouging or or or you know just have arbitrary pricing instead we want the

price to be to be fair and transparent and and set set by an auction um and so there's this this idea of of a price auction that comes in and then the other thing that we want is uh basically users to to be able to users or wallets to be able to capture the the background so nowadays it's mostly wallets that are capturing the the background for example in the case of of Rabbi wallet or or or metamask um but uh you know it's possible that at some point some fraction of that or all of it will will will go back uh to to to the user because ultimately they they are the ones that are creating the flow and then um we also want this this

guarantee of of optimal latency right we don't want a pre-con request to to come in and and for for the gateway to just sit on it and execute it whenever uh you know it's it it's it's it's it's best for them and so we don't want any artificial delays and and from the point of view of the user I want to have this this guarantee that it's pH kill so as soon as the Gateway receives uh the the the pre-con request either it gets executed or or it doesn't get executed uh but but the decision has to be done now and then and and and the user needs to be informed uh immediately and the basic Direction kind of the the basic hand wavy uh you know

direction that I'm suggesting here is just just use suav here as as infrastructure to to try and satisfy some of these goals so for example the no front running uh can be done uh using encryption here you have kind of an auction a competitive auction that's happening within within suav and then for the latency here you can you know potentially rely on on the notion of time uh within within suav or you could have some sort of policing that is done by the by by the wallet now on the other side uh of of the auction you have the the the gateways and the proposes and and this is maybe like the the number one design requirement here that I'm trying to

achieve which is that we want the gateway to not be financially sophisticated so it's it's okay for them to be relay like in the sense that they are sophisticated from a devops perspective and they're sophisticated from a reputational pers perspective um but they they don't need to have you know lots of capital or lots of fancy algorithms and mathematical models uh to to price things and so really what we want is the gateway to to delegate the pricing to a class of actors which which I call prices another kind of um guarantee that I'd like is this notion called strict profitability so every time a Gateway gives without a pre-confirmation to your

user um they need to have some sort of guarantee that this this is actually a rational move um they're making some sort of like Epsilon uh for every pre-confirmation that they're that they're doing and this is you know there's there's kind of two flavors of of this profitability you could have like profitability in expectation so you know if you're if you're a very large operator for example then you know everything will smooth out and in expectation you'll make profits from from your your prec confirmations but there's something even stronger which I guess you could say it's in realization where like for every single precom you're actually um making money and then we also want this this this notion of like no abuse so we don't

want users to be able to um you know probe the auction to to leak information uh from from from from the auction um and you don't want also to to give a free option to to to the users um so in some sense the well in a real sense the the Gateway has the the the last look it's it it's not the uh the user that that has the last look and again like all of this stuff is is directionally something that uh that flash BS has been thinking about with with with SUV now um one notion that that I guess we can that we have with

pre-confirmation is this notion of of of a Max tip so the the user or the wallet is the the pre-con buyer and they're willing to say you know I want this this this transaction preconiran-food.com the notion of the minimum tip uh that that needs to be paid uh by by the user for the pre-confirmation to go through and this is this Market set and so whenever the max tip is greater than the mint tip

here then there is you know an actual trade s an actual pre-confirmation that that that that that that that happens and the the reason why I mentioned this is because um you know one of the the questions that that that people ask is like are pre-confirmation for for arbitrages are there for sophisticated entities and at least in terms of the the the the consumers of prec confirmations uh let's assume that you know we have an arbor here that that wants to to buy a pre-confirmation from from a Gateway the way that they would set the maximum tip is that that they would set it to the sacrifice minus the the minus Epsilon for the arbor so this

would be the Arbitrage opportunity uh that they want to capture um and and on the the other side the Gateway will set the minimum tip that they're willing to accept to be the sacrifice plus Epsilon right because because of strict profitability they want to be able to to make a profit for for for this particular pre-confirmation and so you know Arbors will never agree with gateways on on on tip pricing and so really what what pre-confirmation are all about is is for users like users are willing to pay a a ux premium so they're basically willing to pay sacrifice plus um Epsilon you know ux premium or something um and and this this ux premium needs to be greater than

whatever the the the Epsilon the the Gateway is is uh wants wants to extract okay um now this is a bit of a of a a side goal this is my my light Mass light in the goals is that it's it's actually possible that with pre-confirmation we have like we naturally have better ofas so the the way that um some of the ofas work today with with with the wallet um is that um you know for example like metamask will will sell you know access to at least as I understand them

metamask will sell access to individ individual transactions uh in an auction and let's say that I I have the highest bid for this particular uh transaction like now it is my job to actually you know get it on chain and and extract some sort of backround uh from this transaction and there there's all sorts of uncertainty so I don't know for example what is the pre-state that I'm building on uh I I don't know you know because I'm I'm basically a Searcher that's including a bundle but I don't know where this bundle will necessarily get get included I don't know if even if I will get included uh and executed within the within the next slot or in the next two slots so on next three slots um and you know similarly there

isn't this notion of of timeliness but with uh with pre- confirmations you can basically bundle the the auction for uh for the the tip pricing the preon tip pricing with the OFA and the OFA auction becomes better because you have like guaranteed State you know exactly what you're building on guaranteed execution and guaranteed Timeless and so you know what we could potentially hope for is that uh users and Wallets on the other side of the OFA get get a better deal and you know most most importantly is that they get bigger rebates because because there's like less uncertainty here in the auction um

you're in a position where the the the various Searchers can bid more aggressively and that's going to lead to to bigger rebates for for the wallet today but hopefully in the future uh users also get a cut of of of of the rebat and then a nice thing is that it's it's a single unified auction um so right now you know every wallet has their own like fragmented auction so there's the the metamask auction you know there's the rabby wallet auction whatever it is um and and with with pre-confirmation you basically have this one logical uh point which is the the Gateway where a unified auction happens so it it fixes fragmentation uh in some sense of of the of the O phas and a really uh nice thing

as well is that um it's it's uh an OFA which by default is is is turned on it's not something that the the it's not infrastructure that the the wallets need to go build so my understanding for example is that metamask you know uh acquired SMG you know they they paid you potentially millions of dollars or whatever it is they spent many many weeks building you know the their own proprietary OFA it's it's a huge lift for the wallets to do all this work um it would be nice if there was just infrastructure that all the wallets you know on an equal footing could just plug into and and and and provide all the benefits of of these backround rebates uh to to themselves or or to or or to

their users okay I've done a lot of speaking any any questions uh so far hey Justin can we go over the goals again particularly there's a no Financial sophistication and another goal and maybe we can dive into why those are goals like the strict profitability one for example okay so I think no Financial sophistication the idea here is that the the the barrier to entry to becoming a a a Gateway is is is is is relatively know

so this is um you know this is more of a uh of a you know in order to minimize centralization and I guess you know this might all of these goals might be like slightly less important once we have APS and we have like very clean separation between um between ATT testing and proposing so but uh you know in at least in the short term I think this is this is important uh to to have and then strict profit ability here basically is about giving um you know proposers the confidence that opting into pre confirmations is a rational move right because like by

default like giving a pre-confirmation is constraining yourself it's like it's in some sense it's reducing the the the the block value naively and so you you need to be you know compensated with this tip and the question is like is this tip uh greater than than what you're losing um now if if you have um this this this this Market kind of work you know for for a period of time and it's being bootstrapped and maybe you don't need the this in realization versus in expectation maybe like people will be fine just being profitable in expectation um but yeah I guess it depends on the risk tolerance as well as as of of the proposer like um does the proposer want

uh the exact same rewards plus Epsilon um that that they're currently having or are they willing to have more more volatility um and and and instead only be profitable in in expectation well the the per precon one is I think you know is a difficult optimization right because you're you're trying to optimize something that you don't know so it's going to lead to a lot of over payment if that's a goal it may lead to over payment yes um but you know we we'll we'll we'll go to that in the we'll talk about that in in the in the construction uh part of

the of the talk okay like how how but yeah you're right like if I were to say that okay which one could be relaxed the definitely this one in the middle is the one that that that that could be relaxed as a as a requirement um first one make a preference over like devop sophistication over Financial sophistication which is unclear to me why but um I'm happy to move forward as is well devop sophistication is devops is like a commodity and anyone can do can do devops uh but you know Financial sophistication is it's much much harder because why you know potent well potentially it means that uh you know

you need to be like a very high volume like Trader on binance in order to get like these these lower fees or it might mean that you know you need to very close to jump to have access to the microwave Towers or whatever it is like like it means that that you need to have like an army of phds or you know just that that's the meme that that it's very very difficult like the ultrasound Relay For example you know in some sense devops experts and is like like a rack tag of like three people you know uh and on on a really really tight budget um but uh you know Financial you know sophistication you know think think winter mute like think think jump or you know Jan Street or something

or so yeah there's a very big uh difference in barrier to entry you just started sorry go you justed to touch on this but but uh I was just going to ask two quick questions which is one who do we expect the Gateway operators to be from the current set are they the relays and then two related to that what are the economics for them so this pre per pre-confirmation profitability is interesting but does it go to the proposer does it get split is there a standard there because we've been down this road before um right so who who would it be well if we look at the kind of the skill set devop and reputation that you know aligns very much with the relays and we actually have three different relays uh

the Estus relay the Titan relay ultrasound relay they have all kind of signaled that they'd be interested in in spinning up a Gateway and then you're right like in terms of like the the Epsilon um you know for for the Gateway uh where is it uh yeah that that would be split you know between the proposer and the and and the Gateway in the same way that um so yeah here um the the Epsilon for the Gateway here is actually split uh between the the proposer and the Gateway and what I expect will happen is that the Gateway will take none of it you know at least at the beginning um just like relays you know uh didn't take

anything um and it's it's actually a very similar thing to to the relay sorry to the wallet SL user like right now 100% gets taken by the wallet but I don't think that's a long-term equilibrium eventually once there's competition between wallets and whatnot uh there I what I expect will happen is that there will be rebates given uh to users to incentivize users to join this particular wallet um and eventually in the long term almost everything will be given uh to to to the users on this point I would strongly encourage us to consider uh rational economic incentives for the uh Gateway operators because we'll have a network

that's dependent on awesome Dev Ops we have people who benefit economically but we don't incent the people who are doing the hard work of that I'm not saying how what the incentives should be but I think one of the big weaknesses of the current relay Market is the lack of this um and so I would to see us here repeat that same pattern and now we have all sorts of out of band incentives that get weird quick so anyway it's just much too yeah I mean it depends on your time Horizon like if you're like shortterm in terms of like your time Horizon is only a few years then then then yes it is a problem if your time Horizon is you know you know the the end game in some sense then then it's not a problem because of

APS right with APS there aren't two entities proposing Gateway there's just one entity uh which you could call the Gateway I guess that just wins the execution ticket or wins the execution auction and there's no proposer Insight again that would then suggest that we're back to financial sophistication because the only people who can play this game are those who have the financing necessary to wait until the end game right and so my view of the financial sophistication is you don't want to have Deep Pockets in order you don't want to require Deep Pockets to play right so hey the only way to do this is a $100 million bank because not everyone's going to have a $100 million bank right similarly do we want to create an open market for gateways or

any ragtag group of three guys can do it or do we want to have an ins shine so anyways we've been down this road before so I'll stop but just move for thought about incentives here yeah you're right um it has the same issues as as as relays uh for in terms of being sustainable I also have a quick question about this Arbor slide that show yes so yeah exactly so there's this sacrifice now both for the AR and the Gateway right and it seems that this this conclusion assumes it's the same kind of but uh so I mean if if the Gateway I would assume does not do the Arbitrage himself kind of right that the aror

could potentially make more profit than the gate they would would lose from this opportunity or like yeah so a great question so even though like I'm trying to argue here that the the arbor is not a buyer like the arbor will actually be on the seller side of things and here I showed Gateway but actually is Gateway plus prices and so you basically have all the prices combined uh that are competing against each other and so any individual Arbor like you know maybe won't be able to compete against the set of all Arbors uh here on on on on the right hand on the right hand side uh but you're right like if it if it so happens that there's one

arbit that's just like extremely good uh relative to to the others then then yeah sure they that they uh they they can make use of uh of uh of of pre-confirmation but like they like I I don't know if the pre-confirmation buys them everything anything um maybe it does like maybe like they can make all of their profits just by uh just by being on the other side here or by um basically participating in the final block building uh which you know I guess is the polishing stage where you you you take all the all the all the pre-confirmation and then you you do the the final back run over the over the

whole block uh but yeah I agree with your initial statement that if one of the The Searchers is like exceptional here then then yes they they can be on the left hand side but the thing is that um it's maybe not rational I guess for the arbor to be on the right hand side purely because of the the last look thing so a a request you know takes time even if it's like say 1 millisecond and so there's going to be like one millisecond at least of like ad adverse selection where in some sense like the right hand side here will only pick you know your your OB if it's if the if the price moved in against you

on on the left hand side um whereas if if it if if it moved in in in in in your favor then they'll say no sorry the your max tip is is is not high enough and so really as an OB you want to be on the right hand side here because you have the you have the last look but it's kind of also assumes that you allow the the Aras to be this prices kind of it depends also how do you select how does the Gateway select these people aha great question so this is where like suf comes in where you want this to be like maximally open and permissionless um and in order to do that you need privacy because otherwise um you know you otherwise you need trust

and you can't trust all the Arbors um but yeah we'll we'll talk about that a little bit in in the in the construction section so relays today you know the the trust Foundation is just pure reputation I guess um and uh and and and and that I guess that's that's fine you know it scales to a certain extent so if if you look at um like metamask like my understanding is that metamask only sends their flow to like two or three Builders um because those are the the only ones that have you know enough reputation U to to be trusted to not front run either onchain or offchain um and and and not do other bad stuff with the flow which is you know in plain

text is not encrypted just I just I wonder around some of these Dynamics if you have like cash settled uh like box based features and that could impact some of the Dynamics of who would be a buyer um or a seller outside the constraints of just thinking about transactions on chain I don't know like the implications but just one thing I was kind of thinking through like does this impact some of the assumptions we're making or we can't design for something you know there's obviously not cash settled box Bas features but there could be at some point in time it's like you may want to be on one side of this for other reasons unrelated to a transaction in the current spot gas price but for some other reason because

you so a box based feature or went long one uh that's that's in a cash sett on Market um so just to make sure I understand so basically what you're saying here is that um you know I have this model where there's a window and there's like an intra window commitment and what you're saying is that the commitment doesn't even have to be within the window it could be like ahead of the the window where you're selling a block space future and yeah it's the same problem um you want to make sure that the tip that you're receiving for making this commitment is greater than the sacrifice at the time that you you make your final commitment here um and like maybe this is another example of of this more General problem

that we need to solve I mean another example where you know the correspondence between windows and Slots like starts breaking down is um is uh denal of service attacks let's say that you know I control one slot and then I don't control the next slot but I control the slot after that now I'm incentivized to go denal of service attack the guy in the middle and now I've you know T I've created my my window uh just by dsing the guy in the middle um when you said arbs um are you thinking

also about six stick Arbitrage or just di sticks in uh sex de yeah as well yes yes in sex de can I see so the people will know the sacrifice in that case too well this is the thing like the this value here is kind of an an emergent thing from the auction like the the Gateway is unsophisticated it doesn't have you know direct connections with the with the with the sexes in some sense it's for the market to tell to tell the gatway what this the sacrifice is uh but yeah like this this is a this

is a you know difficult because actually I I kind of cheated a little bit here um it's you know expected Arbitrage as opposed to the the actual Arbitrage like this is the expected arbitra sacrifice sorry at the at the end of of of the slot and you know being able to price the future is is is difficult being able to price it in low latency is even more is even is even more difficult and so like what the construction um you know part is is going to be talking about is basically um having the the prices take all of the the price risk related to this expected uh

sacrifice and then the idea is that pricing the sacrifice post facto where you don't have this low laty constraint and you have like full information about the the full uh auction Dynamics during the slot is is quite a bit easier than than pricing this sacrifice in in in real time with with partial information okay let let let me continue uh and then uh and then we can uh take more questions so part three uh

construction so the the high level kind of structure uh that I that I'm suggesting here is that um you know we have the the user the the Gateway the user sending a request and we have this block which has been partially built so each strip here is like one shred or one tick or whatever you unit or granularity you want to use and it has been partially built so the the the blue stuff is built the top of the block is built but not the not the rest of the block and this this request is given to prices um and the the there's there's competition amongst the prices to to provide the the smallest possible tip and um if if the the Min tip that

they're willing to accept as a as a group is um is is less than the max that the user is willing to pay uh then this transaction here in green you know gets added to the to the top of the block and the user is notified that the that there's a request that there's a a pre-confirmation for that transaction and one of the things that you know I'm suggesting we do is that we use the prices to to shift the the risk so the user instead of paying the the Gateway the the the tip the the the tip actually goes to the to the to the pricer but um you know a little bit like

an insurance mechanism you know for the benefit of receiving the this tip which is um you know the tip that the pricer was willing to accept they need to compensate the the sacrifice uh at the end of the slot so just to to put this in in formulas here like the the the the pricer kind of set that tip to be the expected sacrifice at the end of the slot plus the Epsilon for the Gateway so that what makes uh the Gateway happy and then plus an Epsilon for for the PR itself right the pricer needs to be a a profitable Endeavor and then what they compensate um is at the end of the slot

or I guess you know sometime after the end of the slot uh the actual realize sacrifice plus plus the Epsilon for the Gateway and so the the Gateway you know they they have this this strict Prof profitability where they always make an Epsilon relative to what they could have made had they not made the commitment um so yeah this is where the strict profitability comes from and so as a as a pricer I'm taking on the price risk so for example if uh if there's you know unexpected congestion within within the slot then I'm going to have to pay the Gateway much more than I received as a

tip if the price moved against me uh then again I'm going to have to pay uh pay the the the the the Gateway more than I received as a tip but vice versa um you know if the the default situation happens which is that there isn't like a full Block in that slot then I I collect the whole tip essentially for myself and I don't have to pay the the the Gateway anything other than than the Epsilon for the for the Gateway and so like if we if we put ourselves in in the shoes of the pricer the the pricer at this point in time like during the window is receiving you know as a tip the expected sacrifice but then you know at the end of the window

they have to to pay the realiz sacrifice and you know go going back to this question of like who um you know which prices is the the Gateway going to to to work with you know that depends on on on the trust foundations um and like you know there's this usual like trust progression I guess uh which which is first rely on reputation and then rely on on tees and then and then rely on you know fancy cryptography and I think it's okay to kind of fake it till you make it start um just like relays have started with with reputation and then once uh you know flashboard has uh perfected that te

technology we can move to to enclaves and then eventually once the cryptographers the applied cryptographers uh have something good enough we can use threshold cryptography and like in in these two regimes here on the on on the right hand side um we can have we can have any pricer kind of trustless and openly uh participate in in the in the in the in the price auction and like the the the mental model that that I have here going back to the opening slide is that um you know you have this this this black box of of information you have the the the the black hole in the middle here uh where which is the the this unified auction

where transactions come in uh strategies from prices come in you know price feeds come in and nothing comes out except for this um for this jet here uh where which which are the prec confirmations and necessarily have to be made public so we have this this policy of like maximum encryption or like minimum plain text um that that is delivered to to to the to the public and yeah that is the the end of the the presentation um maybe just a very quick question here um is like by swap do you mean like like a certain way of using te

or like swap as e out of the box no I mean like the the general notion of SW like what it literally means like single unified auction for for Value expression um basically this this single place where the the the there's an auction happening like within within the Gateway um and uh like the trust foundations are either uh either one of these things so like traditionally when someone says SUV like they mean the TV based version of suav but you could have an equivalent suav which is just just based on trust so for example you know like the ultrasound team like could just build suav uh but without tees and you like me bro kind of suav um and then you

could also you know build this super fancy version of suav based on fral cryptography and and this is actually what flash Bots uh you know is working towards in the very very long term what do you think about the Gran of these pre confirmation then do you think they will price per pron for pre transaction or is it be something going happen because if it's per transaction seems like it can be two transaction

coming in the later one might invalidate the first one but it might be much more profitable to include that one that kind of situation R happen right rational to like bundle right so I guess you know it would be rational for the gateway to be able to uh to to to sit on on a on a transaction request uh and and execute it at at some point in the future once you have more information for example about about other other transactions and and this is something that you know I think we explicitly want to to prevent

um and uh you know this is something that that can be done with with suap for example where you can have this this encrypted uh pre-confirmation request that comes in and it will only get decrypted if you know an auction is run on it an encrypted auction and uh you know the the the minimum bid that the prices are willing to accept is lower than the maximum bit so it actually transforms into a real pre-confirmation so let's say you know that the the pre-confirmation request is paying too low of a of a precom tip well the the the encrypted transaction never gets decrypted um so you know it's not like uh the Gateway now has the option to uh

to uh to abuse it in some later point in in the future see I'm kind of reled but like how do you think the incentives will play out I is will the Gateway actually be in device to run this whole thing as like who's going to enforce this whole system is from like the L1 protocol or like the preon protocol perspective you just have or will the L1 the pre-com protocol itself like force this kind of pricing mechanism happening so this is fully offchain

outside of of of the L1 um and you know in terms of the entity that I expect will run it is is is the the the relays and in some sense they already have this kind of sunk cost right like they have the reputation like for Dam it it's free to to leverage their reputation um and you know they already have like a lot of the the devops infrastructure to you know to run a relay so they might as well like Dual Purpose that um yes but like from the proposer perspective they want to like delegate to whatever Gateway that gives the most profit and giving the most profit might not align like what here what being

described [Music] here right some amount of badging might be just much more profitable and a big we that does that might like the whole thing right so like difference in like profitability I guess from the gateways assuming that this design wins out would would would be the same as for the for the relays so it would be you know latency based for example and this is what we're seeing today like there's two relays that that have almost almost all the blocks one just because they they have low latency and they are the ones that are providing the the most

rewards to to to their proposes um so and and and then you know going back um to to Matt's question uh around like you know how can gateways monetize here well they can do the same thing that the Reay can do which is like the very top Gateway that's you know more competitive than the other gateways can collect the their their little Epsilon um because in some sense that they have they have sovereignty o o over this this this Edge is something that they they they can they can capture um so uh you know assuming assuming that they do capture the edge then uh then the the first and the second Gateway

would would would provide roughly the same um you know returns to to to the proposers uh but you're right like just like there's like profit switching like as as a proposer I'm connected to to to multiple relays um and uh and I will I will profit switch in real time uh by by by picking the the relay that gives me the best bid uh there there probably will be profit switching across gateways that you trust but it won't be in real time as in within within a slot it will be across slots so you know you're you connect it to one Gateway and then you realize that some other Gateway is is is is better for you because you know it's geographically closer or whatever it is

then you can go switch uh switch to that to that other Gateway but again like really the the fact that proposers have to delegate to gateways is like this ugly thing just because we don't have APS like in the long term uh all of these complications uh around delegation go away Ellie you have your hand up yeah so with thee the pricers and the gateways it seems like the pricers also need pretty low latency as well as the

gateways so it seems to me like Why doesn't the pricer just become the Gateway and then they don't have to give this EP Epsilon to the Gateway right so the the prices are collocated um to the to to the Gateway just like the the builders are collocated to to to the relay um and like U you can ask yourself okay why you the same question for the for the relays like why why AR the builders like winning out you know as as as relays like uh you know top Builder is ultrasound top relay is ultrasound relay you know where that's not a builder and same for blocks route you know like these these top two two relays are not Builders and I guess the reason is that

um like it's it's a different role it's different skill set you know it's there there there there's a big component of reputation for example where you need to uh to to get the proposers to connect to you um I mean yeah reputation I guess is is more important for gateways than it is for for for relays and the reason has to do with the The pre-con liveness Faults so unfortunately with pre-con liveness faults you don't know how to attribute it you know is it the the proposer that went offline or the Gateway that went offline either could have could have happened and so you're you're basically trusting the gateway to either not have not go offline or if they do go offline

to compensate the the the proposer and so from that perspective there's um it it's it's not like a a pure Financial thing just by looking at the at the bids but you also need to take into account of the the risk of of liveness failures from from a specific Gateway and then you know going back to this this choice of Gateway uh question I guess that there's also a big aspect of like um you know this the propos are having some some element of of preference you know for example when it comes to to to to censorship so um if if if if they want to to delegate to non- sensoring relays then

then that's something that they can express in in in in their delegation um but yeah the good news is that the the three relays that have signal interest in becoming uh gateways have uh are all non non- censoring uh relay uh Matthew thanks Justin so we we had talked about this before um but I'm trying to understand like the the difference in experience for the user um and and the wallet right so when the user is you know interacting with the um with the pre-con Gateway in in this

scenario where you have all the pricers here they are are paying some tip ahead of time um that is like roughly equivalent to the like the value of the worst execution that they would receive if they didn't do it the confirmation right um and so what we're doing is we're kind of moving the um somewhat like obfuscated constitut the user who sends like a transaction to the public pool now they would get you know front runner or sandwiched um and moving that to like ahead of time right so the user is paying a um a preon tip before the um like the Valu is actually extracted um so how do you think this affects like the user

experience um and like is this going to be a problem for for adoption I guess is my question right great question so um I think front running will be like not a thing uh by by the end of this year like the the the public men pool is is is dying and the reason why it's dying is because uh of of uh front running protection so like the all the big wallets you know metamask rabby but I expect all the other ones um will will send their flow to these private ofas uh to to protect against against front Runing now the what I expect will happen is actually the the in some sense the

opposite of what you were saying which is that um you know what what about the the the back run like today users are not capturing the backround just because the wallets are are capturing everything for for themselves but this this what I'm suggesting here is like accelerationist towards competition in the OAS and it's it's basically providing some amount of it's lowering the barrier to entry for wallets to participate in OAS um and um and it's Pro providing some amount of of of transparency into into the whole process um and so what I expect will happen is that uh users will be like hey hold on like wallets are making how much per

year you know and uh and they're going to start asking for for a cut and like the smaller wallets that the that don't have such a large share will start having a button which is like do the transaction and pay me the back run and the back run could be you know a few hundred or whatever it is um and then when this's this this acceler R and aspect and competition across the wallet then you know as I said I expect the end game is that uh the user is actually getting the the the background so in in some sense here like this is the best deal possible for the user they're getting protected from the front run they're getting the back run rebate and they're getting the instant pre-confirmation like what more could

could you ask for well yeah I I guess there's two different types of pre confirmation I guess we in here the first is like um an actual like payment forward flow where the the user is receiving value and I guess you could say that the pre confirmation tip is like essentially negative right um or they're only paying like they they effectively paying um a negative value right or you could imagine where the user is trying to make a um a transaction that would reduce the Arbitrage available right and so they're paying a tip um and in this case then the uh the value that they are paying would be be positive right and it would be paid up front instead of uh what it isday which is like obvious gated and kind of paid um in like the slipage

execution price that they end up receiving on this one I see what you're saying um so basically you're saying that okay when when the the the precom tip is negative that's all well and good uh but when it's positive like they need to be sophisticated enough to realize that they would have paid that tip of you know regardless of the pre-confirmation um or or they might have right because they they're paying the expected value um here whereas if they did it today then they might not actually have to pay anything if the price moves the right way right um I mean this this this has to do with I guess education of of users um like

one one kind of like related thing is this idea of uh like an informational precon where basically you set the the max tip to negative Infinity so it will it will never execute um as as a precom but at least you'll get a response saying you know you were denied you know you should have set your tip to to X and so now you know from an informational standpoint you you you you know what your uh what your your expected uh uh tip is and uh that could also be used for for the wallets to to to do the the initial bootstrapping and and onboarding where um you know they they give the user

visibility like this this is the estimated tip that you're going to pay if you want if you want it instantly pre-confirmation and realize that uh uh that you know sometimes it goes in their favor sometimes it doesn't go in their favor but this is the exact same Dynamic to to to waiting on a slot by slot basis as opposed to tiby tick basis

like I can do the trade now or I could do the in one slot in the future or two slots in the future and I I know there's some element of uncertainty and gambling and the in some sense the only thing that we're doing is is is is changing going from slots to ticks and that but you're right like from Material difference in who you're paying right because you also have to pay the Epsilon to the the pricer in the Gateway so and you're not only paying the expected value which you would today right for for slot you're also paying the Epsilon of the price or in the right that that that's a fair point so we need to to give visibility um to the user on you know the cost of the the actual cost of the pre-confirmation

which are the epsilons um and it's it you know it's it's hard for me to to know how big they'll be but like my hope is that there you know on the order of one cent or sub Cent that kind of of range okay yeah so I guess maybe to reate um users are paying Epsilon to receive the instant pre confirmation quality right that's the three yes but they're still like the rest of the the tip that they are paying is um is still the expected value that they would lose from um submitting the transaction rly exactly okay I can see how that would be a problem but I think if we have better better monitoring and better education

that could be U resolved as long as we don't you know send this every uh every millisecond or else we could have a potential dos Vector on our hands negative Infinity tips don't sound like a great idea to me uh Matt two areas that I think were probably worth worthy of few further research down the line is understanding and user sensitivity to pricing because one of the things I am hearing anecdotally is users are not very sensitive to gas price and that users generally just take whatever the wallet gives them and doesn't really think twice about it and

so we're having a lot of conversation about is this overpriced or not and it's perfectly possible that users just don't care or they care a whole lot less than we assume and return for predictable settlement so just this is something that I think is we don't really know but is worth thinking more deeply about and in fact overpricing may be not a problem and actually may be a feature not a bug so just a thought and then separately is sort of this whole notion of pushing down on the balloon I do believe and this is just my intuition is one of the negative externalities of all of this will be increase in volatility in the base fee and so the good news is we'll get some certainty on inclusion the bad news is prices will swing all over the

place that's just an intuition and so these are two areas that I think are worthy as we go forward here of further investigation and contemplation Matt let me just real quick let me just um ask a little more about what you just said so you said um we've noticed that users are not super price sensitive and I guess um if we accept that then then fair enough right but I think what we can also understand is that many users are not time sensitive right so maybe as a user I'm willing to wait an extra five minutes if I get better execution on my trade um and in pre-confirmation case maybe I'm okay with waiting six seconds right to to get um to avoid having to pay Epsilon

um and so I guess the the price sensitivity is important but also like the the timel sensitivity is is maybe What's um especially like world to break on or I think to your point that there is maybe a bifurcation there's a certain class of users that are price sensitive and there or you know you know what I mean there're sort of different things and interestingly we have examples on the time side where there's a a solution out there called gashaw that basically allows you to basically delay transactions in return for paying less price and so we can learn from what gas hack is doing so in terms of um the the volatility of

the base fee like my intuition is that we're going you know regardless of pre-confirmation we're going to have this this this flow of of transactions come in and like almost all the time the blocks are not going to be going to be going to be full um one interesting thing is that is the the volatility of the the tips so the the tips do get extremely volatile you know potentially when when the blocks do get full because you know we we revert to this first price auction and in some sense uh pre-confirmation have this uh you know potential to smooth out the the the the the tips um and and and the reason is

that you know let's say that the the tip when there's no congestion is one way and the tip when there is congestion is is 10U well what the prices could do is say okay well you know the fair the fair kind of expected um you know tip is let's say 2way and so what what you you have the option to do as a as a Gateway is just sell your inclusion pre confirmations at at two way per gas like you know all the time when you um and so instead of like suffering the spikes you just you just always get the the the tuway okay

just again these are uh matters that are I'm just saying are worthy of research not what's going to happen my crystal balls as hazy as the next guys but I think these are things to consider I mean it it feels like pre-confirmation like you know dramatically like you know open up a new a new design space and it's a little bit like the early days of MF boost right like we're trying our best to try and predict what the the the you know the second order effects will be but like me Boo has have surprised has surprised us you know every six months there's some sort of crazy crazy stuff that

happens okay we're um above time for the public part of the call so I'll um I'll end the public qu the call now and uh open it up to General discussion