The Ethereum Roadmap with Justin Drake

Transcript

you so the L1 changes you want to keep ethereum immutable and just keep it how it is oh no no no I I I love the idea of of reducing this lot times okay okay GM welcome everyone to another episode of the rollup today we have the pleasure of speaking with Justin Drake Justin is one of the leading ethereum researchers working over at the ethereum foundation recently Justin's been battling in the research trenches discussing topics such as scaling the L1 synchronous composability based sequencing the thesis for shared sequencers and the incentive compatibility of rollups and the L1 we discuss all of this and more with Justin

today on the rollup before we get into it first a word from our sponsors enjoy coent is solving the longterm data availability problem where a lot of the data availability providers today in the Moder space when you post data to them they actually only hold that data for about 12 to 16 days and then it's basically they can go into an archive note or often times it's just kind of put away into the abyss and what coill is trying to do is build this long-term data availability solution which can plug into some of these modular data availability providers and provide this longer term Perpetual data solution so what we try to think of this as is when you post transaction data onto Celestial

or Veil or any of these mod da Solutions or even ethereum coal is able to provide a longterm data storage and data availability platform for that data after the time period blockchains are Billboards the databases and you know we need these providers in the space to be able to kind of hold on to the longterm data that is provided from these providers and these blockchain State this is where coent comes in they just had a nice Rebrand check them out on Twitter as well as as the website thank you for being here hope you enjoy the rest of the show Welcome Aboard on our journey into the singularity this is the endgame road map for fra and fraxel the new L2 out of the fra ecosystem that is

totally based on fractal scaling fractal the modular L2 blockchain built by fra with modular and fractal scaling in mind fractal scaling is based on the mandal BR set of fractal images and it explains how l2s become l3s and the entire app chain life cycle goes through its Evolution the FRX ecosystem is made of some of the top builders in the space and has stemmed from its OG days of the frax stable coin it is now in the LST space the LRT space and this fraxel L2 joins the existing Suite of fra fra

products including fra Swap and fra lend you can see fra fairy the bridge to go from any chain arbitrum optimism or main net into the fra main net Bridge your tokens and bring your FRA assets into the fractal ecosystem join today to get the best out of fraxel you are still early join today take part in our Expedition into the singular what's going on guys welcome back to the rollup uh today I expect to be discussing and answering the questions of should we scale the L1 more do we need to decentralize sequencers what is the incentives for sequencing on rollups

uh we are the rollup we're here with Justin Drake Justin we met in Brussels this year I think at a pretty cool caresia event um with some really fun Shadow I don't know if you saw the panel with when when Ed Ed Felton was going at it with Mark from Optimus about fraud proofs but that was pretty that that was pretty spicy yeah that was great um and uh yeah ECC was pretty crazy for me because I I forget the exact numbers but I was doing something like six talks and eight panels and uh that was towards the the very end of of uccc and um it was a bit of a marathon for me yeah saw the uh the panels and presentations that you had at the

beginning of the conference at at the cake sequencing day um I those those are particularly relevant to some of the some of the conversations happening the Ze guys nowadays uh with composability the L1 uh and then and then sequencing as well so I mean why don't we kind of recap like uh the discourse over the last two or three days um there's been a lot going on both Reddit and then responses on Twitter backs back and forth maybe we can just kind of open up the floor as far as like what what are what are you thinking about nowadays in response to kind of the the the nature of the conversation around around rollups the L1 and then and sequencers

as well yeah so um you know eum for a very long time has been wanting to scale um that's been a pain point for for several years and like technically speaking I think we you know have made a lot of progress towards scaling uh We've certainly like dramatically reduced the the base fees and we Al have all the infrastructure necessary to do like full Dan Shing and and grow the data avity portion like even further and in parallel to that we know have amazing uh progress with uh SNS and fraud proof so from a pure kind of throughput perspective TPS perspective we have scaled uh dramatically uh the problem is that um

you know some some other issue kind of came up that we didn't like 100% like uh appreciate or or foresee is that um we the the quality of the block space is kind of lower than what we we'd want basically we have these l2s these rollups that are uh to a very large extent uh siloed and that creates a fragmentation for ethereum um and for some time now um maybe a year maybe a bit more I've been thinking about uh you know how can we fix aerum fragmentation and I have this meme around the United chain of ethereum um how can we you know maximize the network effects of ethereum

of the you know the internet of value if you will um and I you know there's all sorts of things that we can do right at the at the the wallet level you know terms of account abstraction and you know there projects like Ager that try and and provide cross roll La U you know safety like cross bundles basically um and like one other aspect that you know I've been focusing on a lot is um this idea of of shared sequencing um when we have uh shared sequencing we have a an opportunity to do some some some magic uh which is called synchronous composability so this idea that any contract regardless of which rollup

you're you're on can call um another rollup that that kind of shares the sequencer and it's it's a direct call just like you know a would directly call Un swap for example so um if we buy into this feces of money Legos that that compose of each other that so you build these these super structures over time um then composability is very very important and the alternative to synchronous composability is what's called asynchronous composability where you basically have things happen in different blocks like you have some sort of action that starts in one block with one sequencer and then you know it's kind of picked up by another entity another sequencer in another block and it you know it sounds like like such a a

minute little detail you know two transactions two blocks instead of one transaction but um it it turns out to have relatively you know deep consequences from a devx perspective because it means that the way you build applications is completely different to what we have on L1 today and it also has some some ramifications for from a ux perspective it changes you know the the feel of of the applications that you end up building yeah and I think I think there is the asynchronous argument is presenting a very low latency AR uh argument to asynchronous composability that um makes it feel like it's synchronous but it's still not synchronous there's there's nuances and intricacies that still provide ad

advantages to synchronous composability I again saw the um the presentation that you gave at sequencing day around synchronous composability that laid out quite a few user experience benefits as well as as developer experience benefits um and so to just just sound off on a few of those here for the ux you have uh potential uh lower latency uh because everything is in the same slot you have fee certainty you actually know how much the fees are you have better pre-confirmation uh potentially shared liquidity um you have more efficient liquidations if you can make those synchronous contract calls and then you have the ability to do flash loans um

again there are some nuances there but those are all user experience benefits and advantages of synchronous composability and then uh there's devx uh benefits like you know when transactions will revert you have the strongest link liveness uh shelling point you have data compression real-time proving real-time settlement and orchestration that can all provide these developer experience benefits and so I think the argument as as you've laid it out is basically that synchronous composability is important that's the premise um for us to achieve synchronous composability we uh we need a shared environment a shared sequencer basically and then

because we need a shared sequencer to achieve synchronous composability well what is the best shared sequencer it's the ethereum validator set we have the pseudo random validator uh uh election mechanism that is comprised of a thousand or maybe a million plus valid validators and so it's very very decentralized more decentralized than any validator set out there it has tons of Economic Security and so this makes for the best sequencer oh it's also by the way a shared sequencer and it's credibly neutral and it's sensorship resistant because it's ethereum's sequencer validator set so this is kind of the argument for ethereum if for based sequencing and therefore based

rollups and to use the ethereum sequencer and basically the the argument to be based right but then we have the the entire other side where rollups are making money hand over fist and so the there's an incentive misalignment for them to give up some of their revenue I I'd love to get into how that what what makes up that Revenue whether it's base fees priority fees you mentioned in uh some of the recent posts these execution congestion fees so I'd like to dig into that but basically there's this incentive question or dilemma about moving away from centralized sequence and and moving on to decentralized and maybe even based sequencing so that's the groundwork and and could you sort of

dig into like the base fees and the priority fee Dynamic and then sort of demystify these execution congestion fees these me fees where does where does revenue from these like like like base how how are they making their revenue where do those fees come from sure I mean if you don't mind I I want to kind of tie off a couple things that you said um the the first the first one is like you you listed all like this this list of advantages from a ux and a devx perspective and you might think that you know any given like Advantage like shed liquidity or Flash loans might not be you know important enough you know to justify uh know shed sequencing and synchronous composi but there is a long list and I think the the strongest

intuition that I have for why synchrony is valuable is actually an empirical one where you look at you know successful block CHS in production today you know you have like these blockchain systems that are extremely asynchronous you know like like Cosmos like polka dot like near definity like the ton from Telegram and what do all these things have in common they're not very successful with developers or with users right um you know they they've tried to to to do something that is you know scalable or you know with Shing or whatever but that that has come with costs uh and then you look on the other side of the spectrum you have etherium and Solana you know that in some sense have you know gone down this this uh the synchronous route

and and they've been very successful um you know Bitcoin is also you know has a synchronous virtual machine even though it doesn't really make use of it um and so just empirically I think I think uh that that's where a lot of the the intuition and conviction comes from that synchrony is the is the way forward um another thing you mentioned is you know um you know the need for for a shared sequencer and the key word you said is credible neutrality and the reason is that you know in theory we could use any sequencer as a shared sequencer um but I think the most important aspect of a Shar sequencer is credible neutrality because you want to have projects and their competitors feel comfortable opting into neutral uh

neutral territory and so if arbitrum you know develops their own sequencer is op optimism super chain going to opt in probably not and vice versa um and so I think you know in the in the short term we're going to have ecosystem wide shared sequences within the super chain for example within the arbitrum ecosystem but I'm kind of thinking one step ahead if you will um and trying to to come with a a compelling ethereum wide story um now there's various different projects that are working on this ethereum wi story um there's for example it's it's so is maybe the the best example they have this this notion

of of a shed sequencer for for for ethereum um I guess the the the downside of these you know alternative uh sequencers is that as you said they they they compromise um credible neutrality in some way or another and the the three things that are you know are I guess easy to point out is that they introduce a new brand uh they introduce a new token uh and they introduce a new security assumption and so by reusing the L1 validator Set uh we're we're maximally um you know credibly neutral uh but you know as we probably discussed there's all sorts of of uh you know things that need to be worked out for base sequencing to to to actually be be compelling and one of the big ones which

is your actual question is around incentive uh compatibility um and the the question boils down to if I am an an island and I'm playing alone as a as a silo do do I earn more than if I were to connect to a a a a Shar sequencer and in order to to kind of answer this question we need to look at what are the incentives um and there's basically two sets of of of income for rollups uh one is uh congestion and this is specifically virtual machine congestion so any given virtual machine will have a gas limit um for for example you know

base has a gas limit of I believe 120 million gas uh and that's for 2 second slot and they also have you know a corresponding G gas Target and if there's just more demand than the virtual machine can can deliver uh then you have these congestion fees that that start kicking in and the the good news here is that um it if you opt in a um a shared sequencer that you don't control then you still keep your congestion fees as as the virtual machine the virtual machine uh congestion fees in some sense are agnostic of the ordering they just care about frer total frut and total Demand on on the virtual machine and then there's this other aspect uh which

I I I call contention fees but is more usually known as as me um which um you know naively speaking if you were to just change the sequencer you would you are going to lose and so the the the question at least you know naively speaking if you don't do fancy stuff which we can talk about later the question is um is the loss of me kind of compensated by the composability gains and basically my argument is that if you are to increase composability if you could were to connect if you will to this internet of value in the the most uh you know strong way possible then you would

increase the the value of your your applications the value of your state and you'd have basically more traffic coming in and you'd have more congestion fees um and you know that would basically compensate for the loss of me now what I've tried to do is use the L1 as a proxy for you know the you know the ratio I guess between uh congestion fees and and and contention fees and uh the basic observation is that it's it's roughly an 8020 uh ratio so 80% of the income at least on layer one comes from this congestion the 1559 base fees um it's about 3,200 e per day you know

since uh since we've had e559 and then we have MV which is about 800 e per day um since uh since uh since the merge um and and you know I've I've had this thesis that the ratio is actually going to become even more extreme so it's going to go from 8020 to something like 99 1% um and the the reason is that there's there's two sources there's two main sources of me um there's uh sandwiching uh which is a form of of of front running and back running uh together um and that should completely disappear and

actually we're seeing it disappear like as we speak in some sense um there's this very recent trend where the big wallets for example metamask are no longer publicly gossiping you know their users transactions and so user transactions are no longer exposed to to sandwiching and instead they go through ofas and whatnot um and you know with with encrypted men pools and with things like like suav and you know the proliferation of these of these you know private me me pools I think we're going to see this this component the sandwiching just go go to zero and then there's another component of me which is uh sexex Arbitrage the Arbitrage between centralized exchanges

and and and decentralized exchanges and here um it's it's a little bit more of a of a subtle story but basically there's um some various projects out there that are trying to build better decentralized exchanges specifically decentralized exchanges that instead of leaking the me to the sequencer they recapture the meev and then they rebate it back to the LPS the liquidity providers and that ultimately means better prices better liquidity for the users and that you know ultimately probably means that this is going to be the the winning place where where liquidity and price formation happens so um if if if this were to to come to bear

and there's you know there's there's various projects like Sorella that are working working on these uh better dexes one of the issues here is that we need to be patient because um sella is kind of um hinges upon Unis swap V4 Hooks and Unis swap V4 hooks depend on the next eum hard Fork so um you know there's there's a little bit of of a leap of faith you have to do in order to kind of um believe that that me is going to zero but at least that's that that's my thesis and so if if the ratio is indeed like 99% 1% then you need to ask yourself is the gain from composability by having more traffic you know will that compensate for the 1% and I think yeah obviously yes like you're going to

have uh way way more traffic by being uh synchronously composable with the interet of value and I call this the me Gambit right like you give away the 1% of M you give away a porn and then you you win the game um now um it turns out that the the picture on layer one is very different to the picture uh on on the l2s um and during the AMA I kind of made the statement which turned out to be completely wrong uh that basically Bas uh has uh uh has has no me that is just fully all congestion um I want it turns out that

base was in this situation um before July 2023 but then in July 2023 they did an upgrade uh where basically they changed the way that the sequencer worked instead of being first come first serve it was um basically a priority gas auction um and and in a priority gas auction what happens is that the uh the me from from Arbitrage goes to the goes to the uh to to to the sequencer there essentially sequ sequencer fees and if you look at uh the the data um about 80% of the income for base is is sequencer fees

um and so like as as things are are today go ahead real quick when you when you say sequencer fees you mean these these contention fees or and correct me if I'm wrong but people would call those the priority fees yeah so people use the term sequence of fees like a bit liberally and you know what they what a lot of people say is that you know sequence of fees would be for example the $200 million that coinbase is making through base like the Agate fees uh that they that they're that that they're making in terms of of income but actually the sequenc of fees is specifically the subset of those fees which um pertain to sequencing which is the me as you say

the the other part which is the virtual machine congestion fees that that that is an easy thing to to capture got it okay and and so uh these are these are the fees that people are paying to have some sort of priority in the top of the block or maybe in the bottom of the block and reorder based on front running back running sandwiching yeah um and so me the way that it works because it's actually you know relatively interesting is that um base has a two second slot times and then Within These two seconds they take all the transactions that they received and then they order them by uh By Priority fee and then that that is the the order in which they get

sequenced and so what happens is that you have searchers arbitrages that are going to compete to have the lowest uh you know transaction index basically position within the block position zero would be at the very top there's no one Etc uh and so what you see um there's this very interesting graph that that that was shed where like most of the uh priority fees come for like these you know position like two one two three four and five something like that uh and then it tapers out and then the vast majority of users they pay you know almost everything in in in congestion fees and then they pay very little in in

priority fee so if you're doing a transfer or an ERC 20 transaction or some sort of nft transaction then more likely than not uh you'll pay a tiny priority fee and what's happening is that um as it you know the data suggests that on base the you know the defi Arbitrage is just so overwhelming relative to the congestion fees and part of the reason for that is that actually coinbase has like repeatedly just like systematically killed uh virtual machine congestion fees because every time you know there's a bit too much congestion congestion fees rise and then they increase the gas limit and then it goes back down and they have this whole road

map to just keep increasing the the gas limit and so they they're just constantly killing the the the the base fees and so this is part of the reason why there's a a completely different ratio is actually inverted uh between the L1 and the L2 yeah and I think that's the The Dilemma of the and the the root cause of the of the parasitic argument as well as the uh argument with regards to to not using ethereum for da for for for other kind of validium and optimums that people are kind of getting at um but when it comes to directly to Bas given that this Dynamic is how it is with regards to the priority fees does this in your View kind of change the

incentive structure on base or in a given l2's circumstances to move forward with a decentralized sequencing kind of process or like how does this change the incentive Dynamic for base to move away from a centralized sequence right so there's actually um like two decision points that that that are kind of slightly different one question is should the base sequencer decentralized and that's you know interest interesting question we can look at but an even kind of stronger question is should they opt in to a decentralized sequencer which is not theirs basically should they opt into a shared sequencer which is you know credibly neutral um now the decentralization aspect I think uh is

actually fairly easy from the perspective of of Bas and the reason is that the way that you typically decentralize that you introduce a new token um and then you have like validators and then the validators will will receive the me and you know indirectly you know the value of your token would would be tied to the flows from from me so if there's like you know a billion dollars of m per day let's say uh then you know the base token would be worth a ton because you you want to stake so that you can go capture capture the uh the the the me um so I think the the decentralization is actually fairly easy uh another as you know another reason why I think the decentralization will be

a kind of a no-brainer when when the the tech is mature enough is that number one assuming that these better dexes you know do flourish then as you know as we discussed before like the ratio between congestion and contention will be very much in favor of of the centralization and then there's an even more subtle argument which is this this this fancy uh research and and development space called execution tickets where basically the chain can sell execution slots uh and what will happen is that the the the price that that these tickets would sell on the open market would reflect you know the the revenue that you could get which is DME

and and basically they could capture the the me as well the the the thing that I'm slightly worried about uh execution tickets is that well first of all there they add a little bit of complexity um secondly it's possible that they have some negative externalities in terms of of centralization like if you're the the best block builder then you can afford to you buy out the vast majority of tickets and then this kind of self- reinforcing um and and then another kind of worry that I have about execution tickets is that um they they might compromise the composability a little bit right because if if coinbase has their own execution tickets and then some other projects has there some other

type of execution tickets then you know the Stars won't always align and so you you you're reducing the the opportunity for synchronous composability and this is something that you know espresso for example is trying to to to address but that comes with its own set of trade off because it introduces its own it's its own token so ultimately going back and forth with Ellie from espresso and Cilia on Twitter today from from from uh Tao and their Bas rollup and we were just kind of discussing the trade-offs and kind of what the what the what the better use cases are for base sequencing versus basically opting into these uh kind of uh decentralized or shared sequencing models and um the I think that the common thought that a lot of

the the rollup operators are going to be faced with is basically if they can sell their sequencing rights for a sum of Revenue which is somewhat um and this is going to be a subjective value or a subjective uh question for each one is somewhat reasonably close to what they could get with centralizing and the composability benefits as well as perhaps thinking about like the regulatory benefits uh in in a future state are there's a seesaw here where if it makes sense then we'll see that kind of go through um but perhaps until until that experiment is Tried by a big rollup operator perhaps until there's more legal scrutiny which we all hope that that that that you know

doesn't happen um then the incentive to kind of do it uh is going to be like you're saying it's a patient process it's a g it's a it's a thing that that's going to take some time um but that is kind of the trade-off that that that Ellie was describing and um I think Cecilia from gwith also agreed although I'll tell you what Justin it's tough to get the base rollup peeps off of their high horse because the base rollup Peeps are very set on it being the best and kind of like not really accepting other arguments because it just seems to make the most sense from all the properties of livess composability and then being able to kind of also return Revenue to the rup operator as as as well which I'm

sure we're going to touch on more yeah I mean once you get base pilled it's it's it's difficult to unsee it and uh the interesting thing is that it's kind of nonobvious right just like for example call data it was not obvious that it could be used uh by rollups you know it was normally you know you use call data to to submit you know inputs to to to a transaction uh but here you could just arbitrarily post data uh that's Rel to to to roll up and it's it's it's the same thing with the sequencer in some sense we're abusing or like you know leveraging uh the sequencer in a way that's like slightly unexpected um and is is is like uniquely powerful and once you see it it's kind of hard to to unsee

it um but uh I guess the this discussion for today doesn't have to be like around base rollups specifically um we can talk about more generally like decentralized rollups and shed uh sorry decentralized sequences and and and shed sequences my I have a very strong belief that you know the you base will will have a a decentralized sequencer within you know X number of years I think it's bound to happen there are some like technical limitations which I detailed in the AMA uh one of them is around security turns out and this something that people don't really appreciate but the sequencer is used as a security training wheel whereby if there is a bug in the fraud

proof or the snog verifier this bug will not get exploited because the only entity that's building producing the blocks will make sure that the blocks are Kosher and don't exploit the bug for block for bug to be kind of exploited like it needs to be the block producing process needs to be permissionless for the attacker to go build a block that actually exploits the the bug um another training wheel that uh centralized sequencers are used for um is me front running protection um so right now what happens is that users basically send that transaction directly to the to the sequencer the centralized sequencer there is no public men pool uh so and so from day one there was no no sandwiching and you know this is this a great

feature to have and in a decentralized context it's much more messy like okay who do I send my transaction to uh you know it could be some Anonymous person on the internet um and so you know there's there's various Solutions here you could have like these fancy ined men pools like SU or or you could go down this this this direction of of having the wallets have their own ofas and like you know public and you know private deals with with the builders and whatnot but that has it its own set of of tradeoffs because um now you have you know uh exclusive order flow and you have these a lot of opacity behind the behind the scenes um and then the the third reason if I can remember for why it's used as a

as a training wheel um help me out oh yeah fast ux um so you have you know a centralized sequencer which which is you know running on a us with extremely high up time um you know the IP address never changes it's like all optimized for everything for latency and frut and it's an amazing quality of service and replicating that in a decentralized context is extremely difficult you know especially if you talk about a project like like arbitrum which I believe have has block times of 250 milliseconds that's like that's even faster than Solana um and you know this it's very very difficult to do decentralized

sequencing at at that kind of of of of pace um and so yeah it will take time to to get the decentralized sequences but uh but but they will come I think the the more pertinent question I guess around incentive alignment is the the shed sequencer where you're opting into a shed sequencer where the token is not yours and so there isn't this this indirect uh flow of MV to your token uh but then this goes back to the arguments that I put forward one uh the ratio should should move in favor of congestion fees uh because of these better dexes two um we're going to have experimentation with tickets and so

there's some you know some non-zero chance um that that the execution tickets will work out and be successful um yeah vertex Edge is now live on say Network say users can now trade 50 plus spot and per markets with the unified order book liquidity of vertex Edge across arbitrum blast mantle and SE the next 6 weeks 850,000 say tokens will be up for grabs as trading rewards every single week exclusively on say join vertex onay Network now on vertex Edge that's fex protocol.com mantle is a

layer 2 blockchain with one of the biggest onchain treasuries in all of crypto and they have a 100 day campaign called metamorphosis live now what this is is it is a campaign for their liquid staked eth token me eth as well as their upcoming liquid reaking token C me eth where you can utilize this token on the mantle Network in a variety of different apps some of your favorite reaking protocols and defi protocols and earn yields in the form of powder which will eventually be turning into a governance token Mantle's been around for a while we like them because they use their own form of Ian da uh they have a good ecosystem and they're growing so check out this 100 day metamorphosis campaign

and hope you guys enjoy the rest of this video there's two aspects you mentioned about about why the me is going to reduce and and base fees and congestion fees are going to be the majority you mentioned better dexes and internalizing the me and then rebating the LPS but then you also mentioned private order flow and and encrypted men pools which would encrypt the men pool so that you don't even know that the me is there that seems I I I see that as like there there's some trade-offs there because um there was a someone mentioned to me at ECC it was Matt from block native who did block building for a long time and he mentioned that uh there's a

lot of private order flow and encrypted men pools which protects people from me but it makes it very difficult to price a block or Price the execution ticket perhaps and so I'm curious when you extend sort of this this private order flow like you mentioned metamask is no longer sending their order flow to a public me pool you know I think it's on the on at the rate of 60 to 70% if we if we see that number creep higher are you worried about the the pricing for for Block space or blob space or is there are there other ways to get around that yeah I'm I'm aware of you know mt's concerns and I'm personally not too worried um part of the reason is that um

what he's talking about is this Edge case when there's more demand within a slot than the gas limit can supply uh which happens you know fairly infrequently in the context of VIP 1559 so on the R1 for example we have a ratio of 2x between the gas Target and the gas limit and so for the vast majority of slots something like 95% um you can fit all the transactions that came in within this this block like the 2x buffer acts as a dampener uh to to absorb um to to absorb the transactions and then for the 5% of of cases you basically um devolve to a first price auction where if you want to

get in then now you need to start bidding up with the the priority fee which is what we had prior to eap1 1559 um and in order to do this bidding like properly you need to have visibility as to uh you know what what is going on here and um the reality is that like a lot of entities have that visibility right like for example relays they have visibility as to uh you know how many transactions are out there in the quote meol you know the big Builders also have visibility the wallets also have a great deal of visibility and so it's a matter of you know just surfacing this information so that uh in this you know Edge case then we can do uh we can do proper pricing

let's sort of take take this a little bit further into its logical conclusion you mentioned that a lot of this is is uh you know on on the time scale of X number of years in order to achieve decentralized maybe shared sequencers we're we'll get there those will come but there's certain upgrades to to ethereum that are going to have to take place for instance a hard Fork could you kind of talk through like what the next hard fork for ethereum looks like and and you know you mentioned that Unis swap V4 hooks are part of that what is h how does that work why why is that included in the next hard fork and then what are some of the knock on effects of that right so broadly speaking like 90 99% of you know great deal of the

considerations around sequencing have have nothing to do with hard Forks they're like it's a purely permissionless space anyone can you know spin up sequencers it's at the application layer and to a very large extent it's a social coordination game like how do we agree on a shelling point for this canonical etherum wi sequencer um now you you did mention the next hard Fork which basically I believe um will have this this optimization this gas optimization um that is very very useful to Unis Swap and so uniswap before actually depends on this gas optimization to be very gas efficient and you know Sorella depends on you before and you know

the sequences not receiving all this me depends on things like sella so there is a a chain reaction um I don't know exactly when the next hard Fork will happen it might have happen you later this year or earlier next year and then there will be some amount of time until you know liquidity potentially migrates away and so yeah it it will take some amount of time before the the incentives you know become more more advantageous towards uh towards opting into a shared sequencer but again like decentralizing doesn't really depend on this Unis swap stuff because if you have your own token which actually base does not right and so Bas is a bit of a unique uh a unique thing but in any case um you know fix ing the

security issue that I mentioned is is is actually difficult because we don't know how to build fraud proofs and uh you know ZK VMS without bugs these these these are very very complicated pieces of infrastructure that almost certainly have like critical vulnerabilities and so like the Reliance on these um on these training wheels is is is is important today and like one of the strategies that we have to fix the security bug or to at least greatly mitigate it is to have client diversity so instead of one implementation of a zvm you have three of them and you know they all need to agree in order for things to go through and it's very very unlikely that the exact same bug is in

the three different implementations uh you know another approach could be formal verification which is this heavy duty mathematical tool that actually proves ma mathematically that the the circuit is correct for example and you know we can do that and we will do that and you there's a somewhat of a clear a clear line uh clear plan to actually do that and and the foundation is allocating a lot of money it's allocating $20 million towards actually uh completing this this this huge engineering effort uh but it will it will happen uh and all of these things happen on on time scales that are maybe uh longer than than just the next H Fork which should happen in a matter of months got it and okay so coming coming through and

through then to this like conversation that that has with with Max on the pod on on Bank list about improving the L1 and then kind of like this Reddit conversation like when we're thinking about the the key priorities from your perspective with regards to this kind of client diversity and this upcoming hard fork and how you're thinking about um the like the broader landscape of a the including Bas sequencing like where are you seeing the top priorities for rollups as you see them today in terms of their next step to get to the next stage of their development and where are you seeing like kind of like butting heads with other people in the ethereum

ecosystem who have a different perspective um and like you know putting that together how do we kind of how do we align on the best next steps to make rollups to get rollups to the next era of their development yeah um I mean I think that rollups are do actually doing you know pretty well um they're you know moving at a as a steady Pace uh I mean one thing that you know we've we've been disappointed as a community is that you know for example fraud proofs took you know way too long uh to to to come to to to production um but you know in in general I think we're making a lot of progress and as I mentioned um there's going to be these incremental steps or intermediate steps

uh like having shed sequences within specific ecosystems um and shed sequencing is actually not the only thing that that needs to to happen there's there's other um surrounding pieces of technology that that need to to be built out for for this vision of of synchronous composability to really be compelling so let me give a few examples um one is that um having shed deposits would be very very nice because when you want to do a transfer from one L2 to another L2 the canonical way to do it today is to go through the L one so you do a withdrawal from L2 to L1 and then you do deposit from L1 to L2

and even though you're just touching the L1 minimally just to do a withdraw and deposit you're still touching the L1 uh which is which is suboptimal and so really what we want is this layer of shared deposits basically some sort of rollup that does one thing only which is to accept deposits on behalf of the other rollups and then now you can do L2 to L2 transactions without going for the L1 and that's going to be extremely extremely gas gas efficient and again I think what's going to happen is that within ecosystems like the super chain um or polygon or whatever it is um we're going to see these experiments happen and they're going to be parallel experiments and we're all going to be able to learn from each other and then ultimately at some point um the

experimentation will have happened enough that we can pick the best ideas and kind of suggest and try and standardize some sort of ethereum wide solution where all the rollups and their competitor ecosystems can feel comfortable uh opting in another aspect that is important um is real time proving in order to have synchronous composability you actually need two things not just one you need the shed sequencing that's pillar number one pillar number two is the real-time proving basically you need the ability for the rollups to not have to trust each other they need to have proofs that you know things are are working as expected like withdrawals are actually

happening and deposits are actually happening as as expected and the the communication of trust the trust language if you will is is proofs um and by real-time proofs I mean like proof that can be generated within the slot so at most the latency can be one one slot and so if we're talking about base sequencing you know the latency can be at most 12 seconds and so um we don't have SNS that can that can prove you know zkv m within 12 seconds um actually there's a someone today um you know shumo from nebra actually told me that he has like a straight line visibility towards 10c is uh proving time for his ZK VM which would be amazing and he he

thinks he'll get it ready you know within within a few weeks and so you know the tech is moving extremely extremely fast um the other piece of good news is that we have an a really you know somewhat compelling um a training wheel which is tees which provide realtime proving out of the box and so there's various projects out there there's a puffer for example there's T1 um that are looking to use tees uh to experiment with this synchronous comp while we wait for the snarks uh real time proving to come to to come um and it's it's unclear how long we'll have to wait for the real time proving but my like bullish take is that uh in a Eur will have real time

proving uh this is partly because if you look at the Historical Trend it's like exponential uh you know Improvement is much faster than Mo law it's call it you know five five to 10x better every single year which is insane uh and and you know um another reason to to be bullish is that we have SN A6 coming there's three different companies axial cisc and fabric uh Each of which you know have started the tape out process or have already manufactured actually in the case of axal that KN Asic and um you know once these as6 get get tested and integrated into systems and whatnot then uh we'll see performance go through the roof um so yeah there's there's kind of

all this tangential infrastructure that need to be built and then the the last one that I'll mention is is AG layer uh where um if you have multiple transactions that form a what I call Super transaction so you have these constituent simple transactions that form a synchronous super transaction um you want to make sure that it doesn't get unbundled um and you can only execute one leg but not the other and this is what Ager solves for is this notion of of cross rollup or cross L2 bundle safety uh which is also built on top of SN so you know there's a lot to to do um to to get to this like shared sequencing and and synchr compos Vision but the initial steps of decentralizing purely

for other reasons like you know um around uh you know regulatory considerations or around like mematic considerations or maybe you're launching a new token and and you want to have a you know some sort of credible story uh for that token then I I think decentralization should be an extremely easy step um and to my knowledge um all the big all the top rollups have have signal that they want to decentralize um and you know I I believe them I I don't think it's just like the this uh this this vacuous you know virtue signaling I guess one exception is Mega if where they're saying you know we're going to you know go all in on the

centralized sequencer the centralized sequencer is going to be a massive massively sophisticated um you know server and uh and that's going to give us you know 100,000 TPS um now my I think this is you know an interesting and and valid take um my kind of counterargument to to to that would be that with propos a builder separation you can have the builders be extremely sophisticated and kind of run the sequences on your behalf if you will um and what what will happen is that if if it so happens that there's 100,000 CPS uh with like Fe paying transactions

then the the builders because they're competing with each other they necessarily have to build blocks for those because uh if you don't then you your your blocks will be less valuable and you'll lose relative to the other builders that that do um and so in some sense The Searchers and the builders that their job is to just sweep up every all the M that they they can find if and if they they miss out a few pieces of Meb here and there then they suddenly become uncompetitive it's very Cutthroat um in industry and so if there's you know 100,000 TPS of demand then all the infrastructure to support 100,000 TPS uh will will be put in place um I guess like one like potential counter argument

to this counter argument would be that PBS itself like has a little bit of overhead um when you basically when the proposer kind of requests a block from the Builder there's a little bit of latency that's that's that's added there and so if you're aiming for you know one millisecond block times which is what mea is advertising then uh you can't get that you know at this at this handoff point at the the slot boundary if that makes [Music] sense cartei is your home for app specific rollups at the Forefront of developing new tools and platforms that

enhance the capabilities of blockchain Technology Cy enables applications that are flexible and built to scale build on a full Linux environment and use traditional software libraries but in blockchain development go to cy. that's cy. to learn more about building your own app chain today Everclear is pioneering the clearing layer the clearing layer can almost be thought of as splitwise when you go on a trip with friends but for the blockchain what I mean by that is when users are sending Crossing messages there's often times fillers and solvers who have balances left on different chains that they need to rebalance and

reset their positions this is where ever comes in ever is pioneering a collaborative approach to the chain abstraction future and they are willing to work with interoperability Partners solvers and other providers in the space to build a interoperable and chain abstraction future check out Everclear ever clear.org and welcome back to the Lup yeah I'm not sure if you've heard of Rise chain as well um yes so rice chain yeah um interestingly wants to become a base rollup so there may be like the the Mega E but based they based me and mega yeah which I think is particularly interesting and then we had a pretty

technical conversation with uh The Tao team building gwith and kind of talking a lot about cross cross bundle me and the composability benefits and then we're also pretty big fans of what Spire is the Spire lab's team is trying to do with their base stat um you know I'm a I'm a big fan of the idea of based l3s over based l2s which settle so basically h i mean you've probably heard this a bajillion times but the big problem with Bas rups is a 12 second uh kind of block times on theum yes you can do pre-ms but we need some sort of L1 IM uh improvements which kind of Max was going on about on Bank list okay well why don't we just skip that and just rely on the two-c pre-confirmation of arbitrum

one or optimism Etc and build based l3s that settled L2 so I think that's going to be a particularly interesting idea inspired and pull that off and bring the base stack that they're trying to build into uh like rups as a service providers that'll be uh also particularly cool yeah absolutely um I think as you said we're going to have l3s have a Shar sequencer which is the L2 uh Within These ecosystems and a lot of the infrastructure to you know the a layer the shed deposits all of that stuff will you know hopefully get get figured out at the L2 uh before it has to be figured out for the L1 base sequencer yes um one like very like Tiny correction um and it's it's kind of a common you know

misconception is that uh we need to um make L1 changes in order to enable pre-confirmation like um somehow people think that it's part of the road map in the sense that you need to do a hard Fork to enable prec confirmations actually you know amazingly uh you don't need to do any hard Fork uh you can just use the infrastructure as it is today um and you can permissionless have these cryptoeconomic commitments where basically as a user I will talk to the next um block producer and I will get a promise from them to include my transaction and the specific position so if I'm doing a unisoft trade I know exactly um what I'm paying for my trade and um if this promise is not um

honored then I know for sure that this end will lose money that they'll get slashed and potentially even I could have a a crypto economic guarantee that I will I will get compensation you know if you if you pay for insurance you'll get the the compensation and this relies on basically smart contracts where these um these sequences with deposit collateral and on offchain agreements so these are bilateral agreements between the user and the the the the next r one proposer and it doesn't require any hard for I think I guess what now that I think about what you alluding to um is this idea of not really doing pre- confirmations but doing another kind of Dumber thing in some sense which is to

just reduce the slot times so if we could reduce the L1 slot times the L1 changes you want to keep ethereum IM mutable and just keep it how it is oh no no no I I I love the idea of of reducing this lot of times um I'm I'm one of the accelerationists I want you know ethereum L1 to to have all the goodies um and uh you know I think that there is low hanging fruit to uh to to improve the the the R1 um and if we can reduce the slot times that that that's amazing because as as you know Max mentioned and as you mentioned it would make the base sequence are even more compelling Justin we didn't talk about um current rollups and I know that

you didn't want to touch too much on base rups we've got a couple more minutes left and I think that this is a conversation that I heard um actually on the podcast with Kyle Sani on Bank list which was um we didn't we talked about uh current centralized sequence sequence rollups opting into a decentralized model or decentralizing their sequencing or opting into a shared model which is kind of one and the same but two different kind of ways of thinking about it what we didn't talk about is something like a current uh rollup going into a based model um and kind of what that looks like and the big concern again I think comes down to losing the revenue so the amount of Revenue that

say arbiture mix in a centralized sequencer is significant if arbitrum was like let's be a base rle up and inherit all the liveness properties and the composability properties of ethereum L1 you know they would be subject to losing their business would be inherently uh take a a downturn now in today's world there's still no sequencer fee redistribution to token holders or stakers of any kind I assume for some sort of like uh regulatory purpose I do I in my assumptions I think that's going to happen or else I would see a lot of applications on these general purpose l2s if you will kind of go launch their own app chains which is a big part of like our kind of thesis I think is that

if you're an L2 and you're not going to figure out a way to reward your Builders which I think a lot of these teams do through grants and things but there if there's not a consistent distribution philosophy then why would they kind of stay there they could just kind of go and do a do their own app chain and figure out the interoperability or the composability whether through a base model or through something like AG or through these kind of like intent based interoperability standards and so kind of coming back to the Crux of it is do you foresee any of the current large rollups or just Roll-Ups in general transitioning from the centralized model of sequencing and not going to share or decentralized model but rather transition to a based model right um so these things are actually not mutually

exclusive they're decentralized shared and based so I would actually think about it as a progression you have like the the most straightforward decentralized sequencer which is one that you control with your token and then kind of that you know that should be a fairly easy step um from an incentive standpoint the next kind of level up is that you opt into decentralized sequencer which you don't control and presumably the reason why you would do that is because it's a shared sequencer that's uh that that gives you composability benefits and here you know there's a bit more of a question mark around incentive alignment but you know like the decent the the the shed sequencer more likely than not is going to be decentralized because it needs to be credibly neutral um if it's just you

know some some centralized server is not a compelling uh shed sequencer and then there's kind of the next level after that where it's decentralized it's shared and it's specifically the L1 um now the the question you had around incentive alignment you know this this worry of of giving up the fees that's actually not specific to base sequencing it's it's at the second Rong of the the the shed sequencer and the the basically the argument that I have is kind of this this two pronged argument prong number one that the ratio of congestion to conention is going to get better with the better dexes that's going to take some time but that's what

I think the industry will be will be moving towards and then prong number two which in some sense is Insurance in case the first feis doesn't work out which is that let's assume that there's tons of meev well then we have execution tickets as a potential uh technical solution and maybe there's like some other thing that we didn't think of um for example there's this notion of execution auctions which is also a way to capture the problem with execution auctions is that um you need uh the ATT testers to come into play um so bab calls it a super based rollup where not only only do you enroll the L1 proposers but you also enroll the L1 ATT testers um but yeah like the design space is kind of open and the the motivation for

exploring this this kind of muv redistribution mechanism will depend on the amount of muv got it coming back to the to the original uh Point prior to that question about you wanting to have L1 have all the goodies is there like a p a perceived uh balance between act economic activity on rollups versus the L1 and kind of like what what maybe raising the gas limit could do for rollups and L1 users like do you see a a balance that you have or a mental model for going forward how you want to see economic activity be uh kind of dispersed amongst rups and the L1 yeah so I think we should um try and scale

the L1 you know as aggressively as we can within the bounds of decentralization so at all costs we want to preserve you know what's unique to ethereum which is this extremely strong decentralization but within those bounds we want to you know Max the the gas limit and it turns out in my opinion there's there's a lot of low hanging fruit like we could uh you know increase the gas limit by orders of magnitudes and um you know part of the reason why I say that is because in the end game the layer one evm will be snfi and once you have it snif the testers no longer have to reexecute R1 transactions they can just verify the SN and so for them doing their job of you

know voting on the tip of the chain on the valid tip of the chain you know will be extremely easy you can do it on a on a phone or or Smartwatch um and um it doesn't matter what the gas limit is the gas limit could be you know there could be no gas limit it could be like you know a trillion gas per block or something and you would still verify your knock in one millisecond um but you know there's that's the end game but there's there's intermediate stages where we have uh you know statelessness for example um or you know one like really low hanging optimization is this idea that we we can delay the state route to the beginning of the next slot so this the state route

is basically provided in some sense at the end of the current slot and providing it at the start of the next slot is essentially equivalent uh but it just buys you an extra 12 seconds um and so executing these these transactions is no longer part of the critical path and so we can we can uh increase the gas limit and just be be be more efficient um now one of the things that I'm extremely excited about is this this notion of a native rollup once we've SN ified the the L1 evm we can create a pre-compile which will essentially like verify these SNS and verify arbitr uh evm State Transitions and so now

anyone can spin up an instance of the L1 evm and it's kind of like execution shotting except that it's it's so much better than execution sharting because the vision for execution shoting was that there's 64 or, uh 24 uh ethereum um you know copies of the ethereum L1 and with with this pre-compile you can have as many uh rups as you want and the other big Advantage is that they're programmable don't have to be exact copies of the R1 if you want to have your own you know governance wrapper or you know public goods distribution or you know your own gas token or whatever it is you can customize um the uh the logic around the evm uh

pre-compile and what I expect will happen is that all the existing uh you know evm equivalent rabs will adopt this pre-compile uh because it gives them two benefits one is that um um the the evm implementation will be bug free it will be literally equivalent to the L1 by construction um and two it it it won't need to be upgraded whenever the L1 as a hard Fork so you know we're talking about this L1 hard Fork that adds you know some sort of new gas optimization for UNIS swap well what needs to happen is that all the um evm equivalent rollups need to you know use their governance in order to update their own

implementation of the evm they get rollups will get that for free if they use the the op code because it will always be one for one compatible um with with the R1 and I guess the the third benefit is that you've reduced you know potentially you know hundreds of thousands or millions of lines of code down to one line of code which you don't have to reimplement you know your SN circuit or do these crazy fault proofs all of that complexity just evaporates and you just call this one up code and guess you know as a as a rollup developer like anyone can now become a rollup developer and spin up a new rollup and it's going to be a a very compelling future and now if you consider the um if you consider you know L1 execution as being the sum of you

know all these L1 instances then it's effectively like unbounded amounts of of execution the only real bottleneck is going to be data availability in other words it's going to be bandwidth I think we have like just a couple more minutes and you bring me to like first of all that's very exciting um that that that Vision I think ZK VMS play a big part in a similar uh role of reducing the amount of lines of cod needed for these kind of circuits happen and the generalizable nature of ZK Gams will will be very helpful in this progression while I do think there's going to be as you mentioned some more like different types of custom circuits involved um now at the end you talked about data availability being the bottleneck and I think you kind of read

my mind because uh we've been kind of playing back and forth with the idea of data availability whether or not it's a commodity and the idea behind um where sequencing is the ordering and execution is the actual execution of the transaction and actual execution including the value the RO of operators get the get a lot of the a lot of the me data availability for ethereum doesn't get a whole lot data availability layers like Celestia IA Avail they're not getting paid that much in fees for kind of what what they're doing but my question to you actually is more around uh whether whether you view these like modular data availability providers like Celestia I ail Etc as

collaborative to the ethereum road map or as competitive in nature to the ethereum road map with regards to like post EIP 4844 um and and the way that ethereum is trending by Outsourcing execution to rollups and kind of being the main data availab ility layer for them at least in theory um so curious how how you're thinking about that yeah so I think um you know in the grand scheme of things they're collaborative in the sense that uh they accelerate the development uh you know it's yet another experiment it's also yet another option for validium so as you said the validum can become celestium if they consume Celestial data availability um and that's great that you have this this

this this option now my particular thesis is that while you know there's this this amazing accelerationism in the short and medium term in the long term it's winner take um and that's for a couple reasons reason number one is that I believe that you know a single platform call it ethereum layer one can provide enough data availability for the whole world um and you know I have this this kind of uh backof the envelope road map where we can get to 10 million transactions per second within the next 10 years uh and that's uh that's enough for the whole world it's a 100 transactions per day per human um so you know it's it's it's it's it's

plenty um and then the the other kind of uh reason why I think is when it take most it kind of goes back to this notion of of synchronous composability um and and and network effects uh which is that um if you have an external data validity system like Celestia you're introducing a new sequencer right like Celestia has a sequencer like one block after the other and so now your your celestium is no longer synchronously composable with you know your your your typical base R up uh and so that's going to add some friction and you're going to have to

fall back to this asynchronous model blah blah blah um and so so I like you know I I I I love the celesa team I think they've done a great job you know accelerating the space but I think ultimately it's going to be when it take most and my default position is that you more likely than not eum will will win that uh that business opportunity now um one interesting question is um is data vality going to be a a commodity in the sense that it's going to be like dirt cheap um or is it going to be like you know a meaningful source of of value and like I would actually lean towards the the second uh Direction and the

reason is that because of like this this decentralized decentralization constraint that we have with ethereum we're limiting consent participants to home internet connections and if you think of DK Shing it's basically a multiplier over the the the home internet connection which is your your your your your limitation and the multiplier call it to 100x okay so basically all the world you know Financial activity needs to fit within a 100 home internet connections and you know that is a lot it's enough to do like these 10 million transactions per second but it's not like you know crazy aw you know like big data stuff and so my personal feis is that the world in

general is going to want to put as much as possible on on on on ethereum and there going to essentially unlimited demand um and you know that that's you know partially because of this this this induced demand where you know the more Supply you provide like the more demand there is uh just like you know bandwidth like every time your your your home you know your ISP gives you a better internet connection you just start using you know higher definition video whatever it is and YouTube automatically you know will consume the appropriate amount of of bandwidth and I think this is what will will happen like you know businesses will just every time ethereum scales they will just you know put more and more of their business on on on on

chain um and then the you know the other aspect is around this this this these Network effects of of of shared security and and shared composability um and and shed synchrony I should say um which um which means that there's a lot of value to just being on the Manhattan of the incent of value like you don't want to be like in the suburbs and have to take the taxi every day that sucks uh you want to be like where the action happens because that that that's that's where your business can Thrive um and so um you know the what I think will happen roughly speaking at least what I hope will happen is that um there's going to be um you know 10 million transactions per

second and each transaction will pay less than a scent in data availability fees and the math is just so EXT dream when when you're operating at 10 million transaction per second that even a tenth of a cent is $1 billion per day of income so for every tenth of a scent that's $1 billion doll per day so let's say it's like half a cent that would be5 billion do per day of of income for for ethereum um and that's great because you know it's the you know kind of like very large business you know like like in VI like apple like we can we can muster like a multi- trillion dollar valuations purely based on the flow um and then you know this is a whole

different topic but on top of this base valuation of of trillions of dollars you know there is this opportunity for epha to be money to be collateral uh as economic uh security but also as economic bandwith for decentralized St conss blah blah blah long story short you know there's a very exciting potential road map over the next next 10 years thank you Justin for coming on and sharing a lot of awesome insights appreciate you answering our questions thoroughly absolutely thanks for having me