Ethereum & Solana: MEV & Beyond

Transcript

[Music] welcome to Uncommon Core who we explore the Big Ideas in crypto from first principles this show is hosted by John chabo co-founder in general partner of DBA and me hu strategy lead at flash BS and advisor to the light of da today we are very excited to bring you an interview with a former searcher um he taught so many people about the intricacies of me uh I think one of his most famous threats is the whole Saga of sandwiching and re sandwiching and

salmonella attacks that were happening on um ethereum proof of work days um his nickname among other Searchers um was bot C Miller which then became also his Twitter name now he is a product Steward at flbs is actually the product Stewart at flashboards welcome Robert Miller thanks so much for having me I'm excited cool yeah welcome also to you John um how's it going excited for our semiannual podcast recording yeah I think like we really got to um question what we're doing here with this frequency uh that's not the topic for today today we're talking about ethereum

we're talking about salana we are talking about me um and yeah we're trying to connect it to all of the things that uh we're working on inside flashboards so we have um Builder net uh a new protocol that we released on ethereum layer one there's roller boost um with flash blocks and um all of these things um really strongly relate together they relate to the to the um ethereum road map debate they you know like really strongly relate to to rollups and the way that we want to scale and innovate ethereum and they also strongly relate to other ecosystems and what we can learn from them so why don't we start here John you recently

wrote I think it's not probably um wrong to say already the post of the year um uh like you really I For Worse you put your finger on uh some of the issues in I think in the ethereum ecosystem in the ethereum community where um really I feel very reminded of something we went through with Bitcoin back in the day where really it wasn't really clear I think what people want ethereum to be and you have very polarizing sides that if you ask five people you know what is what is ethereum you get five different answers they all

wanted to be basically optimizing in a very different direction and it leads to this you know almost yeah it leads to this governance by it's almost hard to describe what it is exactly right it's kind of in in any case there's no unified direction to what the ethereum ecosystem is doing so why don't you start us off with um explaining what is what what is the issue behind what we have now uh come to know under the name like ethereum Northstar debate yeah sure um you know I I I think you summarized and teed it up all there

of um the the main kind of recognition I think a lot of people in ethereum have had kind of progressively over the past year or so is that um compared to most of these other M major ecosystems so that's Bitcoin salana Celestia go down the list of all the new competitors on all these Dimensions um there's generally less Clarity and cohesion in kind of the ethereum ecosystem on just what are the primary objectives of what we're building what's the what's the purpose of all of this stuff um you know is it focusing on you know we want to build this permissionless money is it you know building a great Tech platform um and then what does that look like you know is it building the best execution environment for everyone in the world to use that's permissionless and censorship resistant or is it more of this kind of

hub for launching a bunch of rollups and they can have all these different degrees of trade-offs on them um all all of which are valid things um it's just that there's more realization of we're at a stage where I at least in my mind I think it would be quite helpful to just try to get more General agreement just within the community Brad broadly so that we can more effectively pursue any or potentially even all of those different goals um as opposed to kind of we we tend to argue in circles on a lot of these things when we're feel like we're having lowlevel technical debates but they're really debates at a higher level of just disagreeing on what the end game that we should be building for is um and concretely I would say the stuff that will end up relating the most in particular to um a lot of what we're

going to talk about today um and where a lot of this manifests is kind of this General I would say debate or disagreement on what is the prioritization in end use cases of L1 versus L2 um where historically you know all the activity on ethereum had been on the L1 still the large majority of it today um by value um and stuff like that um and so I mean for example with you guys I mean that that's where all of the Mev infrastructure has started out that's you know it's the most mature meev ecosystem um in all of crypto by uh by quite a wide margin um and so now over the past year or two you know as people are really starting to shift at more scale into scaling up these l2s um we're getting to that point where you

know it makes sense for you guys I mean you're starting to launch L2 products now for people to deal with me there and um at least now early on that it it does still look quite different from uh what we see at the L1 and so part of the question is I mean uh you know how is that going to evolve over time and then from the ethereum perspective it's you know what is the level of prioritization that we actually have between these things you know do we want to kick everyone off the L1 anyway all the users going to be gone and then you know perhaps that changes the types of activity that we need to think about and deal with there um over time and I would say directly the trend has been obviously you know users and volume on average are moving towards l2s um but

still by far you know the biggest Hub in crypto is actually still ethereum layer one um so it it is still actually the primary place to deal with M and it it is generally pushing the Forefront of new designs and research as well as you know as you guys see with buer stuff like that you're trying to push now you one thing that I would add there is is that even though e is generally the hub for new designs and me research R&D Etc that was not supposed to be the case with the rollup boost sorry with the rollup Centric road map we were supposed to have a bunch of different Innovation going on on l2s that would make its way back down to L1 eventually we'd have it would sort of serve as a testing ground for things on on uh uh l2s first to go back down to L1 and for

whatever reason that hasn't played out that way and one thing that we're trying to do with rollup boost is to reignite that part of the road map and sort of let a thousand different rollup uh uh Innovations Bloom to make their way back down to L1 a concrete example of this would be like testing out fossil on on l2s uh before we roll this thing out on on L1 yeah I mean I maybe I'm biased here but I've been I mean for those who have basically followed my journey I started out in Bitcoin um in 2017 and then I was very heavy like in the Bitcoin like research ecosystem and writing a lot of stuff but

increasingly like fell into the ethereum rabbit hole because of Defi and like I discovered what you can do on ethereum because of defi around 2019 uh I would say and from there on I spent most of my time there um but as someone who was using defi all the time so in Def is summer and Beyond I was very heavily uh you know I was trading lending borrowing yield farming trying out everything and it was very clear that this stuff even if you looked beyond the kind of you know unsustainable yields it was very clear that there was something revolutionary

going on with protocols like unisoft and makeer compound um a uh you know Lio and so on that uh something genuinely novel and if you kind of compared it to how much if you looked at at every single relevant metric how like who who's paying the gas fees on ethereum where's the tvl in ethereum uh like all of those things basically like defi was the number one app by far right Defi and in particular like the kind of high net worth defi that was really what ethereum was about for me

but when you were talking to ethereum developers to ethereum researchers more broadly then you would almost never hear them talk about defi right you would hear them almost you know talk about this entirely these separate types of applications you know like Dows and public goods funding and you know privacy Solutions which had some overlap with defi as we now know um uh you know there probably like a ton of stuff that I'm that I'm forgetting as well um but it really seemed to me like we were living in these two parallel worlds that were not really overlapping

and you know if you ask me okay how did we get here I think this is where the chickens are basically coming home to roost you know you had this incredible product Market fit you didn't really try right you put out this general purpose platform you found in insane product Market fit with defi but you never even knew who your users were and why they were using ethereum and you know where all of your tvl and your your transaction fees and your all of you know the daily active users were coming from and now the apps are starting to leave and you know people are like asking okay like this is clearly not the best chain anymore for defi right um and yeah I think this like

this is my personal take how I think we ended up here yeah I I I would broadly agree with that and I think that has been more of the realization progressively over the past year um is just trying to probably rep prioritize some of this stuff realizing that hey this is I mean where a lot of the product Market fit is because yeah I I mean similarly what what are the things that I concretely use ethereum for um is very consistently it is still generally I mean some variation of high value defi where you know that that is still the place where you know most people trust the most and where I'm going to leave you know a large percentage of my onchain net worth um and that I'm going to do simple high value transactions from time to time um and I I I do them less often than I

would I mean as a result of you know generally you know not scaling and prioritizing the old one and stuff like that um but yeah there's a realization that you know this thing is maybe even stickier than we expected it to be I don't I don't know if that's the case um for people from a few years ago of realizing that yet no people really do want to use this thing um and so there's some level of re prioritization of you know this stuff is going to stick around it's probably a really valuable thing we probably should prioritize this a bit more um and that it's it's very also just not mutually exclusive as well with obviously having a bunch of l2s um in fact I think it's the opposite I think more people is starting to come around to that realization um that you know

scaling the all one doesn't mean you're abandoning all twos it's just the obvious you know thing that you should do for the the hub for all of the l2s that is supposed to be a useful Place yeah I we can already see that there's like core Defi and that can be on layer one then you can have more satellite defi you know the the clearly like lower value use cases like you know mcoin trading where gas fees are abstracted and maybe the average trade sites is more like $500 to $1,000 and that's something that you can perfectly do on Layer Two right and this you know this stuff doesn't need the composability as much because you know people are not collateralizing their meme coins I hope that it stays that way um but I had another thought um and I'm

going a little bit off track here so I don't know if you two guys like if you two saw that uh you know polygon and uh like you know polygon team and um and J and Moro and uh and other folks were putting out this proposal for actually putting some of the uh polygon Bridge into landing markets right so obviously this is not a new idea uh um at least um a blast has done it before um I think you it'll probably become a thing uh where you know you just tap into this additional kind of source of yield and like Capital efficiency um in the future but it just the the reactions that you

saw to it is you're on the one hand like you have uh you know a strategy which is you want to become this unified Bridge aggregator right right with kind of the the E layer just this is kind of this uh it's it's this bridge really for a lot of rollups that like outsources or it handles it bundles a lot of the accounting that is necessary which then makes among other things like uh interoperability between these These Chains easier but then you know you you you you have this proposal to to be the unified Bridge which it's just saying you know we're the best Bridge you can trust us you know we're the most we we're the most safe but we're definitely safe enough

but then on the other side you're kind of doing this you know proposal to say we actually like want to re hypothecate you know all of your funds and it's just it's just a big mistake in positioning right like in particular if you pursue this kind of strategy of like you want to aggregate all the bridges you can't also say oh but like let me also be you know do this thing that's theoretically good but maybe not for you right and the very same thing I feel like is true in ethereum where defi is this incredibly high value use case um um but you're kind of pushing people to the layer two because you're not making the layer one any better right and so it's just

falling behind uh in terms of progress in terms of ux and you kind of think but people can use U Can can use the layer to but the people don't want to use most of the layer twos yet with their high value use cases right I mean I use layer twos but I have 90 like way more than 90% of my net worth that is on chain I have it on ethereum and I I I just have to I don't even have to to look at the data I can just look at myself would I put those funds on a layer too no these you know these are all still controlled by governance not to their own fault but you know it's just not like you just it's just like too early way too early

to even think about deprecating the layer one and I mean that's what we technically doing right if you stop improving in a fast moving market then effectively you are deprecating and so yeah just got reminded of this when I saw uh the polygon proposal and then realized this is the exact same thing that we have been doing with ethereum layer one I think it's also not uh zero or 100 I think we're mostly arguing about the degree to which we focus on L1 versus L2 and it it it sort of felt like it's focused 100% on l2s and and that's kind of the problem um as opposed to we should focus 100% on the L1 because I think in the long term most people would agree that you couldn't have the world's

Financial system just running on L1 at least under the constraints that it is trying to optimize for you need some way of scaling it makes sense that you do that in these different domains it makes sense that you try to aggregate it back on the ethereum's Bas layer and use its sort of unique capabilities um to do that and so directionally I think the RO of centric road map still makes a lot of sense you you're going to need it to scale we're just sort of arguing on how much in the short term do we prioritize scaling the L1 anyway while directionally you still want to go towards the rooll of centric road map in order to put the world's Financial system and a bunch of other forms of computation uh and apps on ethereum

through the RO of centric road map in the long term too y yeah I would agree with that my my general like quick summary of this kind of whole general area is uh it kind of comes back to I mean when I when I had written uh using Josh meme Josh's meme of you know we're all building the same thing um I I think that most people have certainly the realization over the past few years that like Mo most of the endgame architectures for any of these chains whether we're talking about ethereum salana Bitcoin Celestia like go down the list they're all generally converging on a like somewhat similar model in the long run um that doesn't mean it doesn't matter what you build today um it actually means almost the exact opposite is that potentially the most important thing is the path dependence and you

know what is the timing and degree of prioritization on these different dimensions of you know what is the timing and priorization of scaling the L1 versus you know making it easier and more amendable for l2s to launch on top of you what is the timing and prioritization of caring about your base asset versus just trying to build the best tech tech platform um so if you actually do think that we are all building towards a relatively similar architecture in the end game which is where I think directly we are going there's some differences between them but actually the most important thing becomes you know what is the correct strategic way to prioritize and sequence these uh different um parts of the road map along the way there um and that's really what a lot of the conversation as Robert was describing kind of is now is

you know is a little bit of a reevaluation of the timing and sequencing of something these things where we said you know stuff like L1 scaling where we just kind of just ignored it for a few years and it was more of a ah we'll get to that and you know 5 10 years when we have ZK and all these fancy things there's a little bit of a realization of oh no this may be strategically valuable and important um to kind of bump this up the prioritization um and do a little bit more of this now um because this is clearly where this value and we're bum get a lot of limits right now on what we can provide in this thesis that we're all building the same thing do you guys have any thoughts uh whether that means that we will have a bunch of ecosystems that all

look relatively the same in terms of the layers that they are playing on like do do you think for example that you know Celestia like right now it's doing da um you think it's also going to go into like you know maybe consensus and other parts right and you know maybe uh you know Solana will uh you know also go into like all three do do do you think that uh you know every chain will eventually or every ecosystem will eventually do everything or will there be some aspects of the gener blockchain stack that will be you know abstracted and and unbundled into its own shared layer I think effectively everything

ends up doing everything as the boundaries between these things start to become very blurred um and so I'll explain that a bit more concretely I mean starting with ethereum I think it's pretty easy to say I mean if you just generally play out what ethereum is doing right now and it prioritizes everything it basically is doing everything you know we're focusing on the moneyness we're focusing on having an L1 execution environment we're focusing on you know having a you know a robust consensus layer that gives a lot of data um to all these l2s like it is generally doing everything that was kind of a part of the point of the whole Northstar thing um if you look though at you know is someone like Celestia going to end up doing everything I think that's where it comes down to the lines between these things get very blurred of

the I mean concretely the the thing that Celestia does not do today I mean they're a consensus layer that provides a bunch of Da the thing that they don't have is they don't have aable on execution um and so they want to add uh ZK accounts which is just minimum functionality to verify ZK on the Bas layer um and so that allows you to launch Roll-Ups kind of you know similar to the way that you do on ethereum more so um actually a little bit more towards how ethereum wants to go with Native rollups which kind of gets into the point of you know the boundaries between these things blur I think when you play this out on a long enough time Horizon of you know 5 10 years which is part of why I mean people like Justin um who have generally had a very long-term view

I would say uh directly in their prioritization of the ethereum road map you know he talks a lot about you know these you know native rollups um that you know we're going to add some sort of op code in the future to verify ZK on the Bas layer um and they can really be much more kind of one with the base layer where they are you know uh directly the same as the layer one evm they can upgrade in unison you can do those things and so at this point if you add those things and you know it's you know it's being like actually natively verified within the evm execution and the data is also on the ethereum data layer you know what is the difference between this native ZK rollup and the L1 um you know effectively it does become you know a programmable execution chard

you know what what is the difference between these things um and so I I agree with that on a long time Horizon that as ethereum starts to enable and add those things in you know 5 years whatever it ends up being those do end up looking quite similar to the L1 if you put that spot on the tradeoff spectrum of you know they're pretty just one for one compatible we don't have governance on this thing we decide to just use eth as gas token effectively looks like the same thing um and so I think that Celestial will have those you know that that's effectively what we're talking about here is you know there's going to be these verifiable execution environments that's use its data and you know what's the difference between this thing being called the rollup on Celestia and be called being called the

L1 ethereum is going in the same place the the important path dependent Advantage is that those things do not exist today obviously um ethereum has the L1 execution environment which is definitively different today from you know Bas or arbitr or optimism or any of these other things it is a different state space with very different uh trade-off assumptions and so I think that they're going towards a generally similar place um and and even salana I mean they have their own L L1 execution environment but you know anyone can go launch an L2 um and they're going to have more of those over time and so if people are launching those I mean you almost get forced into a rollup Centric road map to some degree on a long enough time Horizon because anyone could just go launch an L2 on salana and it's a big

data pipe why not um as long as there users an assets on the L1 you want access to those in a trust minimized way but you know we need to customize for whatever reason it may makes sense to go launch an L2 so it's the same thing of you know they're they're all ending up in a general similar shape um and then the big question becomes okay if we're all ending up in a generally similar place in the long run what is the area where we have a big advantage today that you know if we're thinking that the thing that matters path dependently is strategically what do we prioritize in what time timeline what is ethereum's really big Advantage is that thing that we've talking about that we all want to use ethereum for and that we don't want to go somewhere else for um it is the

fact that it is still by far the most trusted execution environment um for stuff like high value defi that is still the default place where you're going to keep a lot of your money and do simple defi um and do stuff like that and so I think there's more of a realization of that of yeah a lot of the stuff does look similar in the long run but very definitively for the next five 10 years I don't know what the timeline is there is a big difference qual like qualitatively and quantitatively in what you know ethereum L1 is providing versus the these other places so we should lean into that area where that is the one very big area where etherum does have a definitively large Advantage so what uh we already touched on it a little bit but why what what's the rational behind ethereum bedding so

hard on rollups and in particular not these kind of enshrined rollups but you you could call them you know the private Market rollups and what are the remaining issues with this road map I mean I think that the role of centric road map if you go back to the original post it kind of lays out two things that uh were objectives the first was to increase the scale of ethereum and the second thing was is to increase its in Innovation and to get many different teams building on ethereum not just ethereum core developers but a bunch of different people trying out a bunch of different things that look a lot different than what ethereum core developers had been doing thus far and

uh that is really interesting because you could have parallelized Innovation on top of ethereum that would then feedback into the L1 once someone figured out something that that worked like a new OP code as an example or a different um uh an upgrade to the EV evm that you try out on L2 that makes its way back into L1 and I think if you look at it you know we have been unequivocally successful on the First Dimension uh the first objective which is to scale ethereum you go to L2 beat you go to roll up. WTF you can see we've increased transactions per second you know manyfold I think it's something like 25x the last time that I looked we've increased gas per second something like

65x uh uh from what it would be just at l on um you know you can transact on base at sub scent transactions to do swaps which is kind of ridiculous if you think about the Heyday of uh 2020 you're swapping on Unis Swap and it would cost you $1,000 uh at at Peak in order to make a single trade um and now we' we've had a 10,000x from where we were at those times of of congestions being able to swap on base and uh so we've been unequivocally successful on that first goal of a scaling ethereum but if you look at the second goal of sort of increasing Innovation on ethereum and enabling parallel innovation of many different

teams working on ethereum I think there there's been a lot that l2s have benefited from the L1 basically everything the L1 has done uh has ended up making its way to l2s go look at the op stack it has blobs as one very concrete example but basically nothing has made its way from l2s back to to L1 um so I think we've we've kind of failed at the second goal of getting a bunch of different teams working on ethereum um there are uh really interesting things that are going on um so there there are a couple Innovations uh that I feel like are opinion IND people are taking now stuff like Mega eth sort of pushing the um the trade-off Spectrum to its absolutes stuff like Eclipse a different

VM uh and then we're we're sort of just starting to see more purpose built rollups that are doing stuff that the L1 could never do like uni chain fast block times using a te sequencer or worldcoin trying to prioritize human transactions but there's relatively little Innovation on the l2s relative I think to what the original aspirations were uh and none of that stuff has made its way back to to the L1 yet as far as I'm aware maybe someone can tweet at me and tell me if I'm I'm wrong um so we sort of failed on on that second goal um I do think that one one caveat to that is is is that there are a lot more very well-funded teams and Foundations now that are are

funding public goods sort of in the eth space there's there's like tooling and uh ways that people can transact on on ethereum and it's l2s that have been funded from this and that is one piece of innovation that I think you can you can point to um but not not something sowh fundamental as an upgrade to the ethereum protocol itself and that's why we we made rollup boost uh to try to make it a lot easier to innovate on these Ops to try to uh Foster a lot more experimentation on rollups that could make its way back to ethereum on its L1 um get people to be doing stuff that's a lot more purpose-built to what uh exactly they're trying to do on on their l2s okay before we talk about rollup

boost I think like in a lot of ways like for example but you mentioned rollups are not innovating that much on the execution environment right like so far what we've seeing is mostly the same clients the same evm um you know the same everything that you have on on if layer one and I think this is contributing to the fact that people are not really perceiving uh using a rollup as a kind of trade-off like as a as a kind of you know I'm I'm I'm giving up something I'm giving up some security but I'm gaining also something that's very significant for me right and I think you definitely gain uh lower latency and you gain lower transaction costs and those two are very material but

um now we are seeing like on other chains that it is actually possible to innovate more on the user experience right and like I think uh like Hayden from unop is keeps using this analogy of the blockchain as the server um as this like server with special properties and like why should there be one server right like you would choose the server that basically best suits the backend infrastructure of your business and I think it really speaks to the degree of customization that we going to see in the next generation of apps and so if we kind of you know put this in a kind of General context of crypto uh the reason why we're even noticing uh we even

noticing that you know people are maybe taking their business elsewhere they they're taking their TV elsewhere they're trading elsewhere is that maybe for the first time ever ethereum has what I would call like a real credible competitor and I don't mean to I mean base is competitive with ethereum and like unit chain is competitive with ethereum and arbitrum is competitive with ethereum on like some level and I think this is what you know some people notice at least like subconsciously because like everything's competitive with everything but I'm talking about like a real like competitive ecosystem where there's like actually like no linkage at all right and so John you've been following the salana ecosystem way deeper than I think the two of us have um yeah what what do

you think like is going on like how do you compare Sana and ethereum today um for someone who is spending most of that time in eum may not have all of that context yeah happy to um yeah I think they're a perfect example that also seg that shows the kind of path dependence of what they're prioritizing even if you do end up in the same place um and that we'll kind of segue into some of the next stuff of um in particular I mean even if you think that we end up in this multi- chain worlds um there there's a strong argument that just focusing on I mean even directly as I was saying for ethereum you know you should still focus on the L1 today just I mean in practice you know most of the stuff that we can

do today that we need to do in crypto today kind of fits on one chain um for the most part um and so that's been their entire Focus early on of you know we're just going to focus on scaling this one chain as much as physically possible um and then we'll just take it from there if other chains pop up other chains pop up whatever but our concern is just building this kind of P you know most scalable execution environment price Discovery you know engine for all of global Finance kind of thing um and in that I would say the the directional bias and difference between salon and ethereum has at a high level very often been um and ATT ties to a lot of what I was saying before if ethereum is generally more long-term focused and salana has a much more short to medium-term focus of

you know the world is burning down we need to ship the thing right now otherwise like we're going to lose in the next six months you know we'll be bankrupt in three months kind of thing um and and might be like an inherited urgency from actually almost going bankrupt and I think more than once you know yes like L like literally when they launched I mean it was like absolute bottom of Co crash and they're like oh Are we going to like run out of money and go bankrupt uh like shortly like that that has been their entire history I mean as Kyle would put it I mean like they're basically like a war time all the time um that that has always been their history their culture their kind of thinking um and one of the areas I would say that specifically manifests in which

is a lot of what we'll end up talking about today um is generally I would say their approach to meev and decentralization on a lot of different dimensions compared to ethereum where ethereum's I mean broad summarizing approach has been you know I would say particularly epitomized with PBS and the merge where the the directional trade off and decision was made that you know we are going to heavily prioritize decentralization of validators we want to offload this complexity all of this you know scary M stuff to these different actors so that we can keep validators dumb and decentralized um that came with with I would say like reasonably severe short-term trade-offs of you know we do have a relatively centralized Builder market right now there are concern have been serious

concerns around censorship for the past couple years you know validators can't even enforce censorship resistance most of them if they're going to be Outsourcing vimf boost you know we're trying to add these things you know like Min bid helped a little bit over the past year you know we want to add inclusion lists at some point but definitively there's a fair argument that this was a like a bad thing on many dimensions in the short term because we think that this is the necessary and important tradeoff for the long-term health of the network that this is how you keep validated centralization because that's how you keep the properties that we want in the long run salana has generally made the opposite trade-off of we are going to prioritize we need to ship all this stuff today we need a win today otherwise we're going to be dead in five years and none of

that stuff matters anyway um and so you know they have not gone down this Rabbit Hole of you know doing everything possible to you know how do we uh have a perfectly level of playing fle field to keep you know maximum decentralization on the validator set um and so specifically in regards to mbv a lot of what you've started to see lately has been very clear showing that in the past I would say a week or two so I mean kind of walking through the quick history of you know salana from the start did not have a public men pool um when it had a public mol arise it was built by a third party it was jto who built it it was I mean for ether people I I'm sure you've heard of them at this point but I mean the flash bots of salana effectively

they effectively added it can you explain what what is the or orber as well like what is the experience of a chain like why do you even need a public manool what is the experience of using a chain that doesn't have a public manool because it like when you only come from eum you might think well we're clearly moving away from public M right now right like uh we used to have it it was terrible for everyone people got sandwiched all the time uh and now you know like 80% of Dex trates are going through private M Pools um so like what is the what is the rationale here yeah for salana well so it's it's different in salana in that you know it wasn't that

they had a bunch of you know private you know Builder specific type of men pools in the way that you do on ethereum today it's just that specifically they didn't have a public men pool for everyone it was just you know you would just forward your transactions to the next leader and they forward it to the next I think three three leaders it's the next four leaders so the next 16 slots I think if I remember correctly um so you're just forwarding it to the next few leaders as opposed to having this pool where uh everyone in the world can kind of see your transaction so from the user perspective it actually doesn't look a lot different it's not like your you know Salon users where you know you had to decide you know do I want to use the flash Bots protect or some other Plugin

or whatever am I sending my transaction to someone else or am I using metam massmart transactions it wasn't like a it wasn't like that um it was just a a different mechanism where you're kind of just forwarding directly to the leaders and there isn't this public meol for everyone to see um and so goo did add one effect ly out of protocol where you know you basically have this out of protocol menol um where anyone can see it and so the result of that is I mean you got this type of stuff that you got on ethereum as you know as you started to get more usage on the Chain there started to be a lot more me in particular types of me that people didn't like a lot of Sandwich attacks um so around March um jto made the decision that all right we're going to get rid of

the public men pool um to try to mitigate these sandwich attacks that are going on right now on salana as you know Meme coin was picking up in the early part of the year um and so that that I would say shows the and you know I I my general view when they did that at the time was I I didn't think it would hold up super long but H why not like try it um you know this is culturally more in line with salana you know just go try the thing and see what happens um and the kind of trade-off there is what you are doing is you are create you are getting rid of that easy ability to democratize me across all validators such that everyone has easy access to me and you are creating an incentive for these private out of- band solutions to

start to pop up and you know this gets to all the things that you know flash Bots has been talking about for the past I don't know five years whatever it's been of you know if you don't democratize this m what's going to happen you're going to get the side deal pop up you know validators with a lot of Stak should go partner with you know whoever is good at extracting all the me they're going to make this deal they're going to make more money because they're extracting me they're going to get more stake because they're earning more Soul they're going to compound their stake and like this flywheel continues where you centralize the network on the person who is most willing to extract um and that is what you have seen happen in the last eight months or whatever since uh they pulled the public menol in Sala um

is in particular you have this one very large operator um who has been doing exactly this um who has access to a lot of stake and who has you know had these out of bands this out of band ability you know outside of gto2 they're just sandwiching users a lot and they're making a lot of money doing so and so now there is basically debate with ins salana of you know okay trying to quantify this to figure out how bad is this um because that I mean all of this does end up coming down to how bad is this you know if the centralization risk is you know the bad guy gets all of the stake in the network in two years that's obviously very bad we need to like address this immediately um versus a if

they're earning you know .1% more than everyone else and you know it not really meaningful amount ah maybe this doesn't matter um directly I mean it seems like it matters that this like a pretty meaningful amount of money that they're making right now and so there's a lot more um kind of discussion in salana now of okay what do we do about this and there's a mix of you know short-term Band-Aid type thinking of you know do we just throw back the public men pool in um which is actually directly I mean more in the ethereum line of like okay it's sort of admitting that we can't completely root out this problem right now um we need to at least democratize the MV so that we're not centralizing the network in the meantime and then we need to start getting these more out of protocol Solutions whether that's

validat or wh listing so that we're not just forwarding it to the people who we know are sandwiching um do you start having more solutions like you guys have built on ethereum um you know like flash Bots protect even if they are more permissioned early on you know trying to push users into these more I'm sending it to people that I trust so I'm not going to get sandwiched their Band-Aids are not great but you know they help users while trying to preserve validator deenal in the me can you can you explain on what level exactly the saging is is happening so is it the like so the the users are sending their transactions where to to gito directly or to the no most users are not sending to jto directly um I mean they're they're

interfacing with an application you know whatever our PC that they're connected to um and that is supposed to be sent to the validator there's like a list of who the next validators are and like let's say every app knows that list and so they you send it to the next validator um but validator does not forward it to someone that's so normally when you submit a transaction on you know any application um that RPC they're they're going to send it to whoever is the current leader and then that's going to get forwarded to the next few leaders on salana and so let's say that you are the sandwich well if it's getting sent directly to you you can just sandwich them or you can have if you're the sandwichera deal with some someone else who has a lot of stake and so you don't

have a public mol with these transactions but you are peing at all the transactions that these uh validators are getting effectively and so when they send it to the people who are willing to sell it to you then that you know whatever percent of stake that they have you can see those transactions and then you could sandwich those and send them directly to them um is is effectively what's going on either you need to have the stake yourself or you needed a deal with someone who has the stake um or I mean if you had an RPC provider leaking it to you you could also do that um but concrete that's what it is is access to St the saging is happening directly at the validator level basically because transactions are send clear text to the validator okay so I mean it sounds like you could

relatively easily I mean talk about easily but you could introduce a private RPC and then uh you know have yeah but then you need to Outsource all of the block building right because you need some way for the validator to blindly sign that the the block that they're building and how do you build that on that's a good question but I think you know either you and I or John and I should have the goo guys on the podcast uh soon I think because that's a very yeah just like interesting debate how I think you know the Solana Community is looking to solve that problem yeah because you you really have the double the double whammy here right you're not protecting users from me and you're not

democ like you're not democratizing access to me right it's you're really in you know between these two problems you're really choosing kind of the worst trade-off point right now um and yeah I think that is not so great maybe right so maybe they they have some ideas already like what path they want to take I think it'd be interesting also because uh the jeto folks have been put in a similar position as flash Bots to make really difficult decisions about what to do with their infrastructure where there really is no clear answer so you know it removing the GTO pool didn't work sandwiching happens anyway um but it it was extremely not clear that that

would play out like like it is uh or like it had had had happened um I think it it didn't work sort of because there was sufficient levels of decentralization in salana to the point where you can't just go around and bonk everybody on the head and say don't do that you know this harms the reputation of the protocol or we're going to yink your stake or or something like that but that was not at all clear uh at the time that they made the decision and um I I can appreciate from being in similar positions that uh uh this is hard to to do so it would be fun to have them on have a conversation I feel like we can get into um some interesting philosophical conversations about how you approach these sort of infrastructure level decisions um and and doing that from an

interesting kind of out of protocol kind of in in protocol uh uh perspective too and the one thing that I want to want to add to the conversation is a few years ago uh Phil one of the uh co-founders of flash Poots the The Godfather of Meb he came up with the original idea and the flashs 2.0 flashboy 2.0 Paper he wrote this phenomenal blog post called meev what do and he's um unfortunately not well functioning enough at uh running a website to be able to keep his blog up so you can't find it um you have to find it on archive but I've tweeted it out before and he lays out this argument for why we need to extract meev for the economic security of the protocol and

and it roughly goes that um uh no matter kind of what assumptions you're making about your cryptocurrency economics are probably Central to them because if we were willing to rely on social assumptions or honesty assumptions only then you might as well make a permissioned uh protocol where you're relying on some form of um you know gating and reputation for who who's going to be the proposers for these these things uh but that's not the world that we're in economics do matter and we don't want to rely only on identity and and Trust And as such we need to incentivize people to behave the way that is desired and leverage economics and in the long term if you

want any form of robust Economic Security then you need every validator extracting available me at about the same rate because a validator that can extract more me than another validator is inherent apparently going to be concentrating economic rewards and as a result be able to exert more influence over the long term than people who are not extracting me at at the same rate um and this was kind of the original Insight behind the decisions that flash Poots has made from the very beginning and then I think what has been driving a lot of the ethereum uh protocols decisions around me that you need to have uh every validator in about the

same amount of me for Economic Security in the long term and if you're leaving me on the table then uh you are providing a way for an obvious kind of deviant to go out be aggressive uh extract more M grow larger and be able to then extract more me and and snowball from from that place and so that's the design philosophy that came uh uh uh or that led to things like proposer Builder separation um and I think it's coming to roost also in the salana case too uh where you know we were trying to rely on um some assumptions that were were not economics it turns out that economics actually do matter in salana as well uh and now we have to deal with potentially

a a a um validator that we don't like growing more Stak than anybody else and need to figure out how we can get everyone all all proposers making about the same money um the final thing that I want to add to add to this is that I feel like ethereum has also been developed in on hard mode which is to say that we have been making the assumption that uh we have maximal decentralization from the very beginning and sort of Designing around that right and if you you can assume that you can go around to all validators and tell them like no like stop that like don't include these types of transactions or process things in that way then things get a lot easier and we just assumed from the very beginning in ethereum you can't do that and from that I think you iterate enough times and you get to

something like proposer Builder separation um salana took a different trade-off point I think it was an interesting one worth playing out the experiment but they sort of found that it is sufficiently decentralized enough that you can't go around Bonk everybody on the head and tell them don't don't do that thing um so it's been very interesting to see all this stuff play out uh and I encourage everybody to go back and read this block post from Phil from literally four years ago that made this argument for why you need to extract me for economics that is now playing out in the real world as we see one validator to extract way more M than everybody else and potentially become a very large problem for the protocol yeah yeah and I mean and kind of echoing of what you said before I mean yeah I mean

it it's difficult I mean a lot of this it's it's non obvious at the time even even in hindsight what is the right thing to do at any of these points is a lot of trade-offs I mean the J the jetto team's amazing um and no one thought that this was going to last forever it's just a matter of how long was it going to last and then what do we do um and then kind of I would say segueing into um as we kind of talk about the long-term Solutions here um I I would say a lot of of the philosophical split between at least a lot of people in salana and a lot of people in ethereum is I'd say there generally broad agreement uh across both that you know the way to fix a lot of these issues is by increasing competition in the number

of parties that are able to have an input into block production um I would say the the main split between a lot of the approaches between the two side is at what level of the stack does that happen um in particular there is a lot of uh you know hope on the salon side that the long-term solution to this ends up being you know increasing that competition at the proposer level um so it ends up being you know looking like more like multiple concurrent proposers instead of multiple concurrent Builders which is in some sense what Builder net kind of is um and so you know like part of what totally will argue as the long-term solution is that you know it is effectively that kind of like whitelist thing that I was saying of you know you know validators differentiate on like how they are executing your

order flow you know if I know that this validator is going to sandwich me I'm not going to forward my transaction to the leader in that slot and so you know that doesn't work well obviously as a user if let's say you know half the leaders you know are sandwiching and there's a long time in between those leaders so the way that you make this a competitive market and you know actually have the economics generally in your favor is you try to maximize the leaders per second and so that's a mix of lowering slot times and having multiple concurrent proposers because now if there's a different leader every 100 milliseconds and there's multiple concurrent proposers I you know there there isn't a real tradeoff off to me saying oh I'm going to send it to the good proposer not the bad proposer that I know is going to sandwich me and so

we've made a competitive market and so hopefully this doesn't that depend on the ordering rule here too because you maximize the number of concurrent proposers then any given proposer could sandwich you no depends on how not if you are intentionally only sending it to the leader that you know has a reputation of effectively not sandwiching you like if if that the yeah like if at the current time you know there's 100 millisecond slot and I you know I know it's proposer a and proposer B and I know the proposer a sandwiches and I know the proposer B doesn't I'm just going to send a proposer B and I'm not going to send it to proposer a that's completely aoral to the idea of having multiple concurrent proposal right it's just no it's it's not because it's the same idea of the way that you make the market the most competitive is

by effectively maximizing leaders per second and so that's that's a mix of lowering slot times and increasing the amount of leaders at a given time both of those factors do the same thing yeah I think the devil ends up being in the details of how you compose together these leaders 100% because if you if you have one leader that has the last look on everybody else then like okay great like you sent it to the one person that's not going to sandwich you but the one dude who get has the last look on the Block is going to do it anyway uh so it doesn't really matter uh uh so I'm I'm not really sure that that yeah I'm not very compelled by the AR argument at least in the absence of seeing the exact steps for how that composes into an actual block yeah yeah I yeah I mean I

would say on both sides there is a level of implementation details for the kind of end solution um so that that's I mean I I'm at least teeing up this is the kind of you know I would say philosophically a lot of people in salanta have this direction of we want to maximize proposer uh competition at the proposer level um and then kind of work out the market there versus I would say the ethereum side is more um particularly on your guys side of you know we want to just dumb decentralized validators they don't need to think about anything keep it more simple and then we have this competitive market at the Builder level I mean so but they they do acknowledge that like the user can never be sophisticated enough to choose like specific proposals right so they expect the basically to select the

proposers for them okay so they basically require the app to be way more sophisticated um it's kind of antithetical to Solana's existing positioning right which is like just it's it's easier and cheaper to develop on Sana than on ethereum I mean an RPC can just handle this for you yeah okay so you okay well you are introducing an RPC in the middle okay well that's fine yeah yeah I mean there's always going to be some transaction routing service you know whether that's the you know the the applications the rpcs the wallets all of them will have a level of control here of you know who do we want to send these transactions to for the users and presumably they're going to send them to

what is in the best interest of the users um and that's going to be a competitive market over time if people can choose who to send it to um and so the way that toally kind of use it is basically our job is to just make that market as competitive as possible and then it should be pretty easy for you know we let the wallets the application like the the uis the rpcs all of them they can you know fight it out in that market then um and try to offer the best solution on top of that yeah to build off that John in the idea of involving multiple people in Block production I think essentially everyone agrees that you want to have as many actors as possible involved in Block production um and it's kind of interesting to talk about the different

philosophies uh so in the ethereum world we're trying to have proposers be as Maxim ly dumb pipes as possible um and to have the the uh uh kind of proposer stake distributed around the world ideally uh in as many places as possible but this just to be a dumb job of signing a header essentially that you're Outsourcing to a market but we still want as many people as possible to be involved in Block production and so how how do we achieve that if our proposers are kind of dumb pipes there's kind of two ways on the road map itself um you know an idea I think it's called rainbow staking is that we could have a new

class of actors that have a really dumb job uh which is just to look at the men poool and to say hey I've seen these transactions in the men poool and uh the act of doing that in the protocol produces a list that block Builders then need to include in their blocks in order for those blocks to be valid um and so you you sort of remove and cordor off this role from the rest of uh uh what it means to be a proposer today you create a new actor that's probably called like an includer or something like that in consensus um and it should be pretty lightweight all you need is the mle and then to be able to sign a message and and propagate it around um and so that that's one kind of idea the other idea

that we have for doing that is kind of what flash Bots has been building for a long time which is how do you get as many actors as possible involved in the block building process uh we could talk about that with Builder net but we are working off of the the thesis that you have many different people all around the world that are interested in making money through something like me searching or being able to offer kind of features and services on top of transactions being able to process them in a certain way maybe you know account abstraction would be an example of this and can we leverage that economic incentive can we leverage that economic gravity in order to co-build a block uh in many places around the world and um

that would be nice because we have a lot of people that are constantly trying to extract me and we can use them for some reasonable geographic distribution to different preferences and get uh many different people from many different places with many different compliance policies building a block ultimately getting you censorship resistance and so the kind of thesis being that you can leverage me Searchers and other me service providers in order to get multiple concurrent Builders and censorship resistance which is kind of ultimately the whole purpose of doing multiple or having multiple people involved in Block production and that's a very differentiated thing um or differentiated strategy relative to just you know let's get as many uh concurrent

proposers involved as as possible um and that goes back uh to the original ideas that we had around decentralized block building and kind of uh Builder net is the first concrete step toward towards that change in Gears here a little bit one of the nice things about Solana is its uh its user experience especially its low latency which it uh it can afford you to like a different way that that blocks are being processed uh by validators so um Robert can you tell us how um how we kind of uh you know are learning from this idea and um bringing it to the rollup ecosystem yeah AB absolutely uh so one thing

that's interesting about salana is it streams its state as it is producing a block at sort of regular checkpoints and that gives you the ability to um uh know that your transaction has been executed earlier than when the full block is is built and there's a couple reasons why technically at least for op stack chains you can't go below the current 2C block times and so if you wanted to have a faster use experience like salana does or or like arbitrum does or many of these different chains do how do you you get below below that and um how do you also push the boundaries for uh kind of node efficiency uh as well the answer and ends up being in what we built with

flash blocks that as you are building the block every 250 milliseconds we stream out the state from um a block Builder and so as a rollup user you can know what the state is of uh of a rollup every 250 milliseconds in instead of waiting for these two second checkpoints so every 250 milliseconds stream out the state stream out the state stream at the state eight times in these two second Windows effectively taking what was a 2cond uh user experience and turning it into something nearly instantaneous 250 milliseconds and just brief note since people usually ask um you could definitely get the 250 millisecond

number down further if we wanted to we chose 250 milliseconds uh because that is about what someone's reaction time is going faster than that has um diminishing marginal returns and uh we wanted to start at a place that was relatively higher because it'll be easier to decentralize in in the future and the the final thing that uh we're taking from this and um using it to to or uh final sort of architecture Innovation that we're we're learning and using in in Flash blocks is that if you are streaming out the state regularly then as a full node you can be ingesting the state executing and verifying it as it comes out to you instead of waiting

uh for that two-c window doing it all at once um and so by kind of continuously executing and verifying State instead of waiting for uh these two second checkpoints you are able to process more throughput um and so it's actually a scaling innovation in addition to a user experience Innovation to and that's the core idea behind flash blocks we talked a little bit in the beginning about the like the motivations for why we have the rollup uh Centric road map but also some of the current drawbacks of uh that that rollup still suffer from so where where did the idea for for rollup boost come from so the idea for rollup boost came

from uh us at Flash looking at how do we make it really easy for rollups to innovate how do we make it really easy to um you know add the latest innovation in architecture or or sort of add a new proof uh system or Outsource to a special third party and there was no way to easy easily do that within um the op stack at the time if if you talk to somebody like conduit or one of the other ROP as a service providers they're maintaining like a dozen different gu Forks it's like a monster right now of uh of needing to support multiple different things multiple different da layers etc etc um and so the notion behind roller boost is is

this lightweight sidecar that a sequencer runs that makes it really easy to tap into external Innovations and that was something that we were kind of uh forced into figuring out how to build because we were trying to innovate on rollup architecture with Unis swap um together we put our heads heads Heads down and we asked what would make a rollup the best experience possible for Traders um how can we optimize a roll up at a fundamental level to make it a really great experience for for trading and one of the very uh uh the one of the first things that you think about here is you have really fast block times and so you go and you ask okay like how can we get fast block times Well natively in

the op stack you can only go to 1 second at the moment um that's not that fast of course we can go faster than that and so you you forc to back out into this innovation of intermediate State streaming with with flash blocks but it's not easy to build that inside of um the op stack as a fork and you don't want people to maintain many different things and so you get this lightweight side car that uh enables many different extensions that people can um add on on on top of their rollup and there's a bunch of other uh uh things that we layered on on top of flash blocks um when uh backing up from the question of how do you make this the best best experience possible for trading so we also layered on no reverts so if your transaction reverts is not included on

chain um you know imagine if you had to pay a scent every single time you typed the wrong password into your email account or something it's like a completely ridiculous user experience if you think about it in the abstract um uh we added on um te kind of uh proofs of and verifiable uh priority fee ordering and that's really interesting because if you have verifi able priority ordering then you can build your application in a way that uh internalizes me that isn't possible if you know that the sequencer is going to sort of arbitrarily reorder your transactions or it might censor you but if you have this guarantee that you're going to be ordered in um just a simple priority fee ordering method then

you can build your applications in a way that internalizes me um uh and you only get that by layering on a trusted execution environment to um so those are two more things that you get verifiable ordering and Mev internalization and finally uh by also using a te you get privacy um so uh uh we we actually as the parties that are ordering transactions in a block we don't see your transactions until they are committed inside of um inside of a block uh and so there's sort of no point for for Flash Bots at the moment to be able to mess with your transactions or Sage you or backround you or something like that you get an actual guarantee of

of privacy uh which should lead to better user outcomes so these are kind of the unique Innovations of uni chain that make it more purpose-built for the use case of trading um uh and why we build roll out boost it's uh flash blocks with intermediate State streaming no reverts me internalization through priority fee ordering and then verifiable ordering and privacy on on top of that yeah I for me I think what what all of this is saying actually or like what made me realize I think the power of something like roller boost is just like how much of both the experience like the user experience the develop experience

but also the security of these chains that we're using is actually determined by the sequencer process right like the block building process and you could you could like what is already important on etherum layout one it's just like the whole me pool and you know routing of the transaction and the block building on Layer Two it's actually kind of that time like that times squared why because the like the sequencer is effectively like a lot of things in in one right it is it is like a um it's a block Builder but it's also the validator that basically gives you some kind of economic commitment against the block

that they that they building building and then that that they have to post to the layer one right and like you're trusting like the sequence on Layer Two is actually controlling much more of the overall blockchain experience than even even the M Supply Chain on on layer one is and so you have this really big like you could say well you know this is a really big problem but you could also say well this is a really big lever to then go and and do like a some like really good stuff to to differentiate and like improve the experience for your users and yeah I think that's that's why I got really um excited about this yeah I'm really excited about that

too I think it it's interesting a lot of the reaction on Twitter was is like oh yeah this is an an app chain it's you know Unis swap is an app as a as a rollup instead and I think that slightly misses the point um almost that it it it's not an app chain because it's not just a single application um you know it's not like there's an TR deck or something like that but it is a rollup that is purpose built for a specific purpose and that is making trading as great of a user experience as possible and then it's kind of sequencing rules and how it works is backed out from that and I'm I'm super excited uh uh to unlock the next wave of features and and products um once people get in this

frame of mind um because there's a bunch of different stuff that you can do with the SE sequencer that I feel like people aren't doing today one super simple simple job um that uh I'm excited about releasing at some point would be allowing people to just schedule sort of sequence or cron jobs in the future maybe that's too technical you you could tell the sequencer hey here is one eth call this contract every block for the next 10,000 blocks and then have your game iterate every every single block instead of relying on um outside actors uh and so there's all sorts of stuff that you can do with transaction ordering in in the sequencer once you start thinking how do I build this for my specific purpose I think world coin is another interesting example of this where they have this form of identity

that they're trying to prove that people are humans and so they're going to prioritize people who they think are humans in um in the block over people that they don't think are are humans block Bing for humans yes exactly not not for Bots um it'll be very interesting to see play out I'm I'm excited for it yeah yeah yeah I mean you were early you were earlier saying like a flash block has 250 milliseconds and there is a rationale for why you know it's 250 milliseconds and you know it's the user doesn't really notice the difference if it's 250 milliseconds or if it's any lower than that so I I mean one of the one of the things that like roller

boost is also doing for these Layer Two is is it is a form of PBS right so I I just said a sequence I is really block Builder and a proposer so you're kind of trusting both so what we're doing here is we're unbundling that role right and we're kind of establishing a a similar relationship here that you know that um we actually have on on the layer one and one of the one of the things that ethereum is doing that salana is not doing and like other chains are not doing or like that's actually not fair to say so Solana is doing it but it's just doing it in a different way is the idea of the me auction so what we're actually also doing in

these 250 milliseconds is uh we're bringing me auctions to Layer Two so uh there have generally been two uh auction models on um or two sequencing models on Layer Two they were like prevalent one is just the first come first Ser sequencing so transactions are ordered in the you know order they they arrive at or they are ordered by uh you know there is a notion of kind of discrete time there's a notion uh of blocks but transactions are ordered in uh you know by the priority fee that they pay and I think the priority fee ordering in general you could call uh like you could call that almost like a mini me auction

right because it it has some properties of an me auction in the sense that the Searchers uh have some ability to express uh you know how they want their transaction to be executed what they don't have is for example to say you know I want my transaction to be in the specific part of the block or I want like these two transactions to only execute together uh I want them you know to fail you know if they don't execute and so on right so there there are basically uh I think things that were missing that uh that like make like a priority fee auction into a fullon m auction that over time I think can also

uh you know make their way to the layer two and it uh it it has just a lot of benefits right it has a lot of benefits for for Searchers has benefits in terms of more Revenue U for the chain itself and it has uh also significant benefits like for users um like both because their own transactions no longer revert but also because the overall capacity of the chain goes up and so I want to quickly kind of hear you comment on the idea of you know like reverts and just like how the idea of block building and M auctions in general like it has been I think very mislabeled as you know okay this is just you know to take money from

users and like democratize it to validate us and so it's like it's you know you could say it's like it's a necessary evil for decentralization but I think very few people have actually seen it as well this is also a scaling technology yes for sure so you you either you need to pick one of three you can choose an M auction in some way you can choose that your m is extracted through latency instead whoever is fastest whoever's collocated whoever's figured out you're messed up networking um or whoever spams the hardest on the Chain those are your three options and you know you have some people doing the first comfor serve thing and so they're choosing latency and then most everybody else is sort of

sleepwalking towards spam without thinking it and the reason is is that if you don't have an offchain auction an M auction then your your auction happens on chain instead where uh since there's a private men poool and Searchers don't have any signal on what is happening when and the chain is extremely cheap to use as computational resources the dominant me strategy is to implement your Bot as a smart contract and then just Hammer the chain just absolutely hammer it as hard as you can um do your searching on chain and probabilistically you will land some me extraction uh uh you know when someone

creates an Arbitrage or when someone creates whatever opportunity is it is that you're trying to to do um so the way that this works in in practice is you know user trades on base they create an Arbitrage opportunity and then a dozen me Bots send literally hundreds of trans transactions that are each just churning trying to find arbs on chain and one of them succeeds the rest of them are using gas uh just using space that is not economically productive um uh and means that other users can't use that gas in instead it effectively lowers the overall gas limit with um these actions that are they're just a waste and it is even worse if you think about the fact that every full node in

perpetuity is going to have to replay all of um all of these Searchers that are just turning the chain over and over tens of thousands of times you know millions billions of gas probably and um we've been saying this for some time because uh our experience on eal one led us to have this Insight where before flash Bots was live in the meev GU days you had Searchers that were wasting a ton of gas on chain trying to extract me and that was clean up once we had an M auction um and this is even worse than the contact an L2 because if if gas is extremely cheap then you're willing to use uh you know a boatload of it and you can see this happening in practice with base where recently up to 30% of of their 20 to 30% of their blocks is used

up by you know one or two Bots that are just searching for me um so you know base is increasing the the gas limit by millions every single week and then Bots are just eating eating that gas crease uh up and so at some point there's a tradeoff where you actually get more juice in scaling from implementing an meev auction and cleaning up these uh all this wasted gas than you do by increasing the gas limit more and we might get to a place eventually where we do have a billion gas per second but 999 million of that is used by muv Searchers looking for arbs on chain because is so cheap and they can make it

all back with uh a single successful Arbitrage and the final thing that I want to connect this back to what were saying uh uh before is that on l2s you can Implement MV auctions in a way that is still user aligned so you you don't need to have sandwiching in order to have an auction to clean up all this gas spam you can still have an auction do it in a you know 250 millisecond window like we are doing clean up all of the spam on chain uh but without your users getting getting wrecked and in part this is because the um kind of form of proposer Builder separation that you can Implement on an nl2 does rely on some assumptions that are just not pure

economics um so on on L1 we only want to rely on pure economic uh assumptions that's because we're sort of optimizing for this extremely decentralized situation maximally decentralized situation on nl2 we can take more opinionated um tradeoff decisions and have just governance choose who the Builder is um as an example and that's kind of a social assumption not an economic assumption but because you can do that you can assume and make these um make these protocols where there is no no front running as an example uh and we're going one step further and using uh kind of verifiable computation and privacy through tees to um uh layer that on on top of that social assumption as

well that there there won't be any front running that you can only only back run in these auctions um and then you get the other guarantees that you want like sensorship resistance through the l12 um so I think it's interesting kind of way to connect it back to the early discussion around protocol decisions PBS Economic Security Social Security Etc I think it's the admission that like some of these some of these things is just like it's better to do them offchain uh instead of onchain but if you don't give people the the kind of release valve to do it offchain in a way that that works for them uh then they are going to do it on chain and it's going to be pain you know for for your network and for like your the the kind of long-term like costs

incurred on like State growth and and like all of those things and yeah so I think that's the angle that we've been taking on a lot of these things is just like people have incentives and their incentives are going to dictate their behavior so like you better build a solution that like aligns aligns like what they are doing like aligns their incentives and what they are doing with like what's tolerable for like all other stakeholders right yeah the the story behind the bundle if it's interesting to people originally was is that in in 2020 on on ethereum you had a public men pool and we had a lot of um defi activity was

kind of going exponential at the time and it was all happening in the public men pool and so you would get these um priority gas auctions where I saw a user making a transaction that would create an ARB and then I need to play this complicated networking game of trying to be the first person after that transaction in the public men poool um sorry in the block that the minor uh uh eventually extracted but then it's proof of work and so you don't know who the minor is um they're all around the world it's like this weird complicated game now of having the best peer-to-peer uh overview and you you had this kind of iterative game being played back and forth with other other Bots too where um

since it's all in the public men poool if I try to ARB this thing someone else sees that and they can front run me take my ARB and just pay a little bit more and then I'm trying to you know bid over them and so you would get these games where people would be tit fortat trying to outbid each other and that would waste a ton of gas on chain as um you know the the waste of that war between multiple Bots essentially um is eventually included and it's very inefficient because your your sort of ability to spot transactions that create me and then to capture me didn't translate into your ability to extract me to you you actually needed to be super low latency have a great view of the peer-to-peer layer um and so they

were extremely high barriers to entry and what uh what the bundle did was is it let you be very precise about the ordering that you wanted um so you you know I care about being behind this transaction in particular and here is my uh ARB and it turned What was a um what was an allpay first price public auction into a sealed bid first price auction too where you're you're sort of competing on your payment to the minor instead of competing on your ability to um you know one up the next person in the fastest way possible and have the best view of the peer-to-peer layer and the barriers entry there are way lower so many more people could bid and it's a more efficient auction um since you're incentivized to bid your true value with

a sealed bid auction and as a result it was way more profitable for miners to run the software than um just building in the public menle and the way that the software was originally built too is you would you miners would um multilex they would just choose is it more profitable for me to build a mempool only block or a block with a bundle at at the very top of our block um and so they sort of always made the most money um from from running the original software and that's where the original idea came from was kind of uh lowering the barriers to entry and increasing competition um by increasing your precision by changing the game from peerto peer and kind of latency basis and then cleaning up spam on chain as well um from the tit fortat

that these Bots were playing all the way back in in 2020 I think that was a Phil idea also originally um hats off to him that that should wrap up uh you know the discussion about uh layer 2 and moving to our latest announcement which is actually Builder net on layer one and you know there is a fun fact uh behind this name which is it was actually your idea John yeah I'll be waiting for my royalties I appreciate it John was uh roasting me before the proposal for our other names and and how bad he thought that they were and and initially Builder didn't really resonate with me but it did come out on top yeah in the future when you transact on

ethereum you're going to pay you know like a 0.01% like Eternal royalty to to John for the name for the naming rights that we had to rent from him it builds for that are sending it to theu the community Builder or something else so well John suggested John net also but uh that that was beat up by Builder net yeah so um what is what is Builder net uh so Builder net is our latest announcement on the layer one it is a decentralized or you know more decentralized block Builder Network for ethereum and it kind of has has two things that are really interesting that

I I want to call out the first is is that multiple people operate this it's no longer just flash Bots running a block Builder we've actually deprecated our centralized block Builder we no longer run a centralized block Builder anymore and instead we have this block Builder that is jointly operated by The Flash Bots team the nether mine team and uh the beaver build team you know the largest block Builder by market share on on a um uh and this is really interesting you you can get this because everyone is running their block Builders inside of a trusted execution environment where you get some guarantee of privacy and you can share private order flow really sensitive information amongst the

operators and you get some guarantee on the code that they're running so the first thing is is that we have actually taken a meaningful step towards decentralization by introducing multiple different operators of this infrastructure instead of just relying on a single team like flash Bots uh or the BBU team to to operate this stuff instead the second thing that I think is really interesting about this is is that the Builder net uh network is actually refunding the value that it is able to capture um so you know Builder net is bidding within the me boost auction and whenever it wins it takes that Surplus that um between the value of the block

that it could create and and the value that it bid to the proposer and it refunds that back to users up to kind of um you know some semblance of the second price that they should be paying so if you've overpaid for your transaction you could have paid less to get included inside of our block uh we will refund that Surplus back to you and I think that's really interesting for a few reasons um the first reason is that it is essentially upgrading ethereum's gas market to be something like second price there sort of caveats on that but but roughly second price and that's a really unique differentiator that no other protocol has not salana not any of the rollups um of a second price gas market second reason is is that this is a open permissionless rule you can go and you

can you can look at it on um our docs on the Builder net docs uh uh and it's kind of obviating the need for these offchain opaque agreements uh in order to get the value of your your transaction these exclusive orderflow offchain agreements um instead and so it's an open transparent kind of answer um to off- pay exclusive orderflow deals that are going on um and hopefully something that is is uh more Community aligned as well since this Builder is is going to be paying out uh refunds from the value that that it actually captures um final thing that I want to mention as a part of part of Builder net is is that it it provides us this basis to be able to

launch other interesting products in the future and and so in particular I'm thinking about the relationship between flash blocks that we talked about earlier where where we have these te block builders that are ordering transactions on l2s in Builder net um so in the future we want to let uh rollups Outsource their sequencing to Builder net you know with a drop in extension and out of the box then you get decentralized sequencing um by leveraging these block builders on L1 that are um already building blocks inside of a inside of a te and so it's kind of drop in decentralized sequencing on on the road map too which I'm I'm really excited about and that's a short summary of of Builder net um anything you want to dig into there sure yeah I have actually some

thoughts on like refunds because it see like it sounds I think when you when when you kind of hear for the first time it's not immediately clear like what problems with the with the market structure it actually solves right and so there's two motivating problems for why Bild is giving refunds and so the first one is me is still not pric into transaction fees right so I think we had you know an idea like a long time ago a couple years ago actually which is like you know why can't we just uh you know solve um the M problem through better fee estimation and it's like a

surprisingly kind of s like it's it's just an overall like really really good idea right because if your transaction can be manipulated in a certain way that lets like a block Builder or Val that extract in money from it that in theory that is a transaction fee that you pay right it's it might be an invisible fee but you're paying it nonetheless right and so you know that's how I mean for example uh like Robin Hood or like other kind of you know retail brokerages for example can afford to have users trade without commission why because their trades have me right and that's how they can like sell their order flow to market makers you know get paid for it and then like Finance basically their operation

from from that on the back end and so it's like a long story short but if you're if if like you're you know you're paying like a $50 transaction fee or like let's say like a $5 transaction fee because that's more what we hope I think to have in the future on ethereum um and like your but your trade is paying like you know a $50 fee because it's getting sandwiched you you're technically paying like a $55 fee right and the same is actually true if you're making an inefficient trade and like in theory uh you're paying a $5 fee but your trade can be backround for $20 you now technically you're enabling that background to happen because you're the one moving the pool right and so in theory also your transaction should be

worth something you know between like $5 and $25 because you are actually creating the me here that then somebody else again can Harvest and so it would be really nice if we kind of would have this like overall view of what a trans action is actually contributing to a block and that is what the what that is what buildon is doing because everything's everything is happening inside the Builder and the basically the Builder can compute how much each transaction is contributing to the block it does knows the price that it's actually paying to the validator and can thereby kind of effectively you know tell you how much you should have been like how much you paid how much you

should have been paying and thus how much you should be refunded so that's the first problem the second problem is what is what you touch on with like these exclusive orderflow deals and and so on and it's um no let me restart so the second problem is what I would call the Dual auction problem so while the first problem is responsible for a lot of these private orderflow deals where for example telegram Bots are you know making deals with Builders and you know getting the kickback on the extra priority fees that the users are paying or like the the background ability of these trades the the other one is really one that's more kind of native to the Mev supply chain and that is involving

basically Searchers and and block Builders and so uh right now when you look at the Mev supply chain you actually have two auctions going on at the same time and so Searchers are technically bidding for inclusion with the block builders in a sealed bid auction and then in the second stage these Builders are bidding uh for the you know to make to to to be the highest bidder with the validator at the relay and while the first auction is a sealed bit auction the second auction is what we would call you know an open outcry auction or like is a public auction because uh as a builder you can actually query what is the most um what is what is the currently highest bit uh

at a specific relay right and so uh you know this basically gives the Builder a much better ability to monetize blocks uh like among among other things it allows them to to to basically uh like more specifically bid um or on the value of that transaction right so it's it goes back kind of the nature of these two auction formats where a second price auction really encourages someone to bid the their true value in the auction and the first price auction encourages you to bid only as much as you need to to be ahead of like the next person bidding right you just want to be like a little bit ahead and so that is why uh Searchers have

increasingly integrated uh either into becoming block Builders themselves or uh basically making the same kind of deals with block builders that we are seeing you know the telegram Bots make so it's just a deal that like I want to technically bid also in the open auction and not in the seal bit auction and so I want you to to bid on my behalf and then pay me back my fees right and so what we're seeing here is something that's been actually in a lot of ways the driving force behind kind of the last like 12 to 18 months of the market structure uh and yeah the idea I would say uh with build is doing this but making it much more democratic

and much more transparent and much more credible on the way that these rules are actually enforced inside the block Builder I see it as an upgrade to PBS in in many ways the um proposer Builder separation through Mev boost fixing a bunch of its problems so the first problem being this vertical integration pressure that you have if you are a Searcher you can kind of more fine-tune your bidding by being a builder and so you have this um incentive to vertically integrate into being a builder instead of just competing at the Searcher Searcher level which is not desirable for a bunch of different reasons and now you can just Outsource your bidding to Builder net and Builder net will bid on your behalf

it will refund you up up to to the fair value of your transaction sort of the second second price in in quotes um that's one thing second thing is is that it's trying to revamp the kind of dimensions of competition within PBS to be more productive for the protocol and for users so the main dimension of competition right now is on order flow you're trying to go out get these wallets get these telegram Bots get these users to use your your private menol your Builder instead to have an edge in the auction um instead what Builder net is trying to do is to create this kind of open Sanda box where anyone can access order flow where anyone can compete on the best way to merge it

together or providing uh kind of new services and products on top of orderflow too um that that's kind of the ultimate vision and initially Builder net is um sort of dumb monolithic Builder that is run by multiple different people it's just the same merging algorithm multiple people are running it in multiple different places which is nice for some reasons but we want to go beyond this like I said to create this sandbox where people are competing on um providing more value to users by providing and providing new products and services and anyone would be able to do this too in the future um so the dimensional competition would be a lot more productive than today's

competition yeah so so one kind of uh follow on question I had like leading from that which also ties back a bit to what we were talking about before is I would say um so I I mean directly kind of on Builder net um you know as this thing progresses over time the vision is very much you know uh Builders can share orderflow with each other early on and cryp between each other you know progressing into later even you know sharing partial blocks with each other you know you kind of have this kind of snowball block building kind of on top of each other effect um and so bringing this back to the conversation before of kind of like mCP versus MCB um what what is the value of this kind of you know really moving towards more multiple concurrent Builders like having a lot of

Builders actually input into the same process um particularly if ethereum does move to which seems sort of likely at this point um to implementing something like fossil um which for The Listener I mean is effectively a a weaker version of multiple concurrent proposers where you know you have a lot of validators who are sending it incl list um and then you have one proposer who still has a proposing Monopoly potentially um you know kind of acting as a view Canon alizer you know putting those together um but you know all of the inclusion lists that everyone else has submitted or respected and then you could just have whatever build there you know Outsource that that they could reorder the block how they want um so so concretely in that world where you know hopefully we have a bunch of validators

who are contributing to censorship resistance you know uh ensuring that you are going to get your transactions included um what what is the additional value there of having you know a lot of block Builders inputting into this process still at the end as opposed to just having you know why can't we just have one Mega centralized block builder at that point um just you know perfectly efficiently optimize this whole thing um especially you know it's going to be obviously more difficult to compete with on latency um and stuff like that so yeah I'm just curious like what is the the additional value out of that kind of having multiple uh parties uh joining in the building if we also do that the proposal level to some level I'll start and I'm sure fasu you're going to have have things to say

about this too I think the the the concrete value is that we think it's going to be the best foundation for applications like defi that need a neutral platform to coordinate Financial activities around the world and there's kind of two parts of that the first is is that uh I think it will be more censorship resistant than you can get that with just fossil um I think it is unclear what the impact of fossil is going to be uh I don't want to come off sounding like I'm a bear or something like that because I'm I'm not and I think we should still go ahead and do do fossil anyway but uh in the very first versions of of fossel as a proposer you're sort of sticking your hand out

and you're saying please sir include these transactions and uh that may not be the what you want to do as a as a proposer especially if those transactions don't look that great to the you know jurisdiction that you currently reside in um and so without some level of privacy on top of fossil it's it's not actually clear to me that um how many proposers are going to be uh submitting inclusion lists and whether these parties want to be doing this as well and how effective it is um I think it'll work regardless like there will be ethereum is geographically distributed enough that you're going to have some people around the world that um arguing uh you have enough people and enough

geographies with enough enough coverage of different compliance policies that it will improve censorship resistance as well um but I just don't know to what extent in in the limit um and I think that decentralizing block building is still interesting because you can use this financial energy around capturing meev to get more people in more places involved in the process of building a block in a way that is going to be less um you know less sticking your neck out so to speak then uh I am a proposer please include this transaction uh uh that uh my geography does not want me to include or my jurisdiction doesn't want me to include that's the first thing I think it it still improves on sensorship

Resistance um second thing is that uh I think you still don't want one monolithic Mega Builder because it provides a choke point for rent extraction from users um so having a a single monolithic Mega Builder they will get all the order flow they're going to be able to capture a significant portion of that value instead of providing it back to users or providing the fair amount back to to proposers um final thing that I want to add to this is that I think by involving more parties in the process of building a block you get more points of coordination and more points of competition that are ultimately going to create more value for users in in the protocol than you could ever get by

having one mega block Builder one mega block Builder is never going to be able to come up with the best ordering um in and of itself there's always going to be some Edge case that is going to produce a Better Block pack more transactions extract more value be able to refund more value users give better trades uh and one mega block Builder is not going to be able to serve all of the different features and products that people will want in in the future you really want a platform for open competition and coordination among as many people as possible with as low of barriers as possible so that you get this uh maximal um maximal coordination and competition that produces the most value for users and for uh proposers and provides this

foundation for for Innovation um so the tldr I think is is the decentralized block building is the best platform possible for financial activities uh by preventing um choke points for censorship and rent extraction and increasing you know places for Innovation and competition yeah so that's a very good answer I have some thoughts I would say on top of that and so I think in general uh I don't know John if you were at lion uh actually in uh in Bangkok uh the one around Devon yeah I was there briefly I was there for my panel and that was yeah so like vasili one of the co-founders of Lio he was giving basically this talk

that was both a review of the last year of Lio but also like an Outlook into the future and he was reflecting how long he had been working and staking uh he had been St like working he was previously the CTO of P2P and they had been working and saking even like years before even uh you know ethereum then completed the merch right and then there were the first sing pool inum and so on and back then Lio was designed uh or like ideated as this end game of staking right this idea that like if you only have Lio you know no there's no more in needed and this is like the one siiz fits all solution right and now like he when he was on stage he said like I no longer

believe in in any end games at all like the market is always evolving ex preferences are always changing we need to position so that we can be flexible and that we can like serve like many different use cases on the same shared foundation and I I thought this was so fitting because it's the very same thing that I feel about the M Supply Chain it's very hard to design this like one monolithic Mega Builder that like works the best for everyone and the reason is you're already seeing you're already seeing the block Builder being kind of unbundled and disintermediated in a bunch of different ways actually and

so uh if you look at you know what all of the following things have in common you know like a layer two um uh cop uh you know unop with hooks uh like any kind of app specific sequencing um like at the end of the day you know these are all basically like the user has a transaction or include also me share and like me Blocker in that list as well the idea is just the user basically you know has an intent or a transaction and then it's processed of chain into something bigger than it originally was right and in the case of a roller like a lot of users actually have you know uh transactions and their transactions get

like process like they get sequenced and processed and bundled and then they get posted to the layer one as this data blob right in the case of cop just the same you have like in some cases like many different traders who want to trade actually in the same block and they all get put together into into one transaction right it's all like the transaction is constructed by this network of solvers of chain uh and then they're competing in the auction then it you know it gets put into this one transaction and gets put on chain like me share you have a transaction gets back run it gets put into this bigger bundle um sella you know the idea of like every unisoft pool can have its own sequencing right and like it can auction off the the top to

Arbitrage the block to return lvr to lpce really like What's Happening Here is that block building itself is becoming unbundled right like already today when I'm uh when I'm when I'm trading uh in in in cop like I I have a trade intent there are multiple servers competing for for what I want to do and then there's an auction uh you know it gets packaged into this bigger transaction but then like often times these transactions then either get sent to M blocker or flbs protect Because by the time that basically the solver auction has cleared and then you know it gets sent to the RPC and it for very often again time has passed and maybe a

new Arbitrage opportunity has opened up or maybe it's even arbitr Arbitrage opportunity that was only possible if you're like specifically order like two like two transactions in a row right H and then all of a sudden there's another background possible you know on this overall like C bundle and um and then it's forwarded to the block Builder and by the time it's at the Block Builder you know maybe the pool balances have changed again right and there's like a third background possible on the same original like trade intent right and so my point is uh like first of all like already we're seeing tons of innovation like on the front end of like how user transactions are processed and it's impossible for one Builder to satisfy

all of that uh and you know there's this very strong kind of drive towards like more and more efficiency in the supply chain and I think the only way you actually do that is you if you have many small parts that can all interoperate and that that can all compose with each other on this same shared foundation and I think this is really the beauty of what like we can do with these tees is that you can have a shared mol uh and you can have orderflow sharing between different parties um but it can all be done on the same order flow right like there can be a big pool of order flow and everybody can kind of process it and can contribute to it um

and everybody is also like refunded for their contribution to the block right that's how it connects again to this idea of the refund uh because if if you can make a bundle more valuable then you should be compensated for it right and that's I think how we can really leverage the idea of you know Searchers and offchain compute and uh you know and those things really I think to the to a degree that you know eventually the execution of crypto transactions can be way way way better than it could ever be in trefi because they only had these really like monolithic kind of execution matching engines and I think I think uh in part

that's why we are fighting so hard here at fleshpots for these open competitive platforms because I think the thesis is that we will in the long term be able to outc compete trfi by offering better user experiences and better prices by having more open markets with lower barriers to entry but that's not going to be the case if you end up getting a single Mega monolithic Builder um uh of course and to link this back also to our original Vision with Suave I think we saw into the future to some extent uh that all these apps were going to want their own kind of custom ways of of ordering things that you would have um you know uh collaboration and

competition amongst different parties to to extract value like the solver case being being backround as an example and trying to create the the idea was to try to create the ideal platform for these parties to be able to um cooperate uh together and compete together against other parties as well and that that was where the notion of suave gu came from which was um some uh R&D that that we did of can you express what you want and how you want to be executed in a smart contract in solidity um it turned out that that was uh uh not the exact product that that uh people wanted to to use and what I'm really excited about

with Builder net is I think we found um a better fit with stuff like the bottom of the block searching inside of a TDX that we did with rsync um where you can have a Searcher deploy their entire rust bot or python bot or whatever you're using and a node inside of this te and then have other parties that don't trust them stream in state um so the Searcher can do useful stuff for you initially um you know our sync one of the top block Builders is using this to have their blocks back run by uh arbitrageurs instead of just trusting you know one or two different parties um to do this inste so it is opening up the bottom of the block Arbitrage market and I think

it's an early example of how you can enable coordination um amongst untrusting untrusted parties with tees and how we might one day get to a more uh distri Ed Builder as well one other question I just wanted to ask kind of broadly um kind of wrapping up a lot of the conversation on this is just curious if you can kind of walk through how you see the progression of Builder net over time I'd say both on kind of just uh uh on like I would say like a technical level and a literal you know what do we expect on like usage and kind of stuff of you know um like you me like early on there are like more trading wheels on it it'll be decentralized more over time um like I know in the post um it like Builder net

relies on centralized infrastructure for functions like transaction Ingress and data storage um like like how stuff like that gets alleviated over time um and then also just kind of the stat and the status and kind of timing of just general usage of it um and any predictions if you want for how many blocks or Builder net making uh by end of next year oh man uh yeah let let let me uh think on that if I want to give you a prediction only to be quote to we did on it uh in year um so I I'll first start on sort of practically how this thing works and it's it's training wheels um practically you can imagine Builder net as uh or a good place to F to see how Builder net works is just by imagining a

centralized Builder to begin with put it inside of a te um and so it's a centralized Builder you put it inside of a te and so you get this guarantee on the Builder that is running inside of it you know the exact code that is processing your transactions like okay you got that great now with that what can you do well since you have this guarantee uh on what code people are running and the Privacy that a te gives you you can have other parties that you don't necessarily trust running the te as well um and so that that's the point the Builder net is at where we have multiple different people running this te Builder and they're all sharing order flow amongst each other and they're coordinating on the Block bidding

process C they're not competing with each other in the me boost auction but it is kind of the single Builder code base running inside of a te that uh is just sharing water flow amongst these different Builders and it's distributed around the world and that's very nice uh even as a starting point because you know different parties work in different jurisdictions they have different compliance policies and so flash Bots as an operator um you know has one compliance policy we have we censor certain transactions but other operators Builder net have different compliance policies uh and they they don't censor the same things that that we do and so even as a starting point Builder net is more censorship resistant than any individual Builder could be today um and

sort of improves on the censorship resistance of the meev Boost Market um even given that it's kind of this single centralized Builder code base with three parties that are operating it and the other really unique thing about um Builder net is the refunds that it it is processing and we're we're trying to give people kind of something akin to the second price there's caveats there but something akin to the second price on their transactions by refunding them up to the fair value of what they could have paid the minimum that they could have paid to still get included but the way that you calculate that is by running an ungodly amount of simulations um essentially uh kind of waving my hands a little little bit here but you

run a ton of simulations to see what transactions conflicted and how do these things affect each other and where does value really come from inside of a block um by trying many different orderings and seeing how these things relate to each other uh so you need a kind of archive of all order flow that you considered in a block and precise timings for what you considered and when and then you need to run just a ridiculous amount of simulations on on these things um and so one point of training wheels initially for Builder net is this idea that there's an archive of order flow historically uh that flash Bots initially is running um uh uh and we use that you know really just to

process uh refunds by running all these simulations on a regular basis and then paying out refunds um we'll upgrade that in the future to be kind of pull instead of push so right now if you go to Ether scan there's I think it's build- refunds doeth we're paying you know hundreds of thousands of dollars out to users through this ens address in the future you'll just be able to pull whenever you want um from the Builder the the refunds that you have and they can acue over time so that's one training wheel the other big training wheel is this notion of a builder Hub um and the the sort of long story short here is is that you need some notion of identity what is an

actual Builder net node and um where should we be sending order flow to and have we verified that these things are actually running um tees in order to share order flow amongst amongst each other and that uh kind of identity management today is fairly centrally managed Again by flash Bots as the kind of stewards and creators of this uh project and it's very feasible to do decentralized identity um we just haven't gotten there on on the road map and so over time we'll reduce the role that we have in um kind of the setup and identity management of the process to be you know completely decentralized uh as well as doing things

like governance too um uh in a more decentralized fashion but initially that's one of the other big training wheels that exists um final thing is that uh you know a lot of times in Block building you need pretty fine grain access to logs and data analytics on what was happening when you were building a block and when so there's um more debug and logging access than I would be fully comfortable with at the end game because this is still a system that we are iterating we want to improve the block building make new algorithms make it more competitive etc etc etc so that there is a kind of debug and

logging function that we have access to um after blocks are being built not not kind of inflight I think it's cool seeing how ethereum is progressing here on a lot of the M stuff um and it's not entirely but I I mean like it to some degree comes back to that we're all building the same thing to me certainly not exactly but I mean it's interesting that you see in both salana and ethereum I mean I would say the the the general trend on at least you know block production how do we solve a lot of this m stuff over time is you know basically how do we increase competition and get a lot of parties involved in Block production um over time um and there are I there are very different approaches to

you know how Salon is thinking about that today versus how most I would say ethereums are thinking today um but I would say that that is very directly the trend that we are headed towards in both ecosystems that both seem to be pursuing as kind of the answer um and so it's been very interesting to start to see uh both strategies and this is probably the first time that you'll really start to see it play out more because obviously I mean Sal doesn't have mCP or anything live um so getting to see kind of live on ethereum is finally a testing ground of you know starting to see this you know private building and more collaborative building over you know the next year or so um as building net rolls out and we just get to see very very soon even the live usage of the the

early stage stuff is going to be great um and then it similarly it's the same thing I would say with roll of boost on the l2s is a little bit back in that direction of what we talked about earlier on is I would say there is some level of disappointment in uh some regard for l2s to ethereum of you know not as much Innovation as maybe you would have liked on certain different dimensions of you know go try some stuff out and you know we can learn a lot of lessons maybe some of that comes back to ethereum um I think rollal boost is one of those things that you know as l2s are finally getting more mature they have more users now they really need to think about muv and get serious about it um and same thing it's this is one of those

areas where it's going to be a good testing ground and you'll probably see these uh these two areas kind of start to converge again over time um but yeah excited to see both of them I think basically everybody agrees you need to get as many parties involved in Block production as possible for competition reasons and censorship resistance and to kind of put a tea on it I think what we we think is is that um to actually get these things you want to be meaningfully de geographically distributed too so it it it sort of doesn't matter if you have many people involved in um block production but they're all in one data center in one jurisdiction you want many people across the world with different compliance policies different preferences all kind of

collaboratively um uh uh building a block instead of doing that in in one place and that gives you the constraint of trying to build a a really performant chain while also having things be meaningfully geographically decentralized to come back to kind of the ethereum road map and how much we should be focusing on L1 I think you want to improve the L1 to the greatest extent that you can without sacrificing Geographic dist distribution um people from all around the world kind of running proposers because that's where you get real censorship resistance from in in the long term and I think um Builder net is a really interesting uh bet in this direction of can we harness

the energy of M People all around the world running the software to get many people in many different places uh involved in the process of of building a block and then can we use that to let rollups Outsource their decentralization and other features to um you know these people that are already running this te software to uh to kind of try to put a bow on all of this stuff yeah I think for me the takeaway is I think people have been looking at m at this like really ugly duckling you know uh like okay you kind of you don't it's so ugly you don't really want to look at it but you know you kind of have to if you want to keep your chain

decentralized but I think it's becoming like increasingly clear what we've been saying for a very long time this is like a really really really instrumental part of your blockchain design uh like you should be thinking really like me first and like the way that you know block building works and like Meo design works and these just these just influence such a huge amount of the final user experience uh the developer experience the security of your blockchain that you know is like just very very hard to to kind of quote unquote like you know Outsource that or like ignore it right in in in in the kind of the way that you think

about uh your protocol right so I think you have a huge the the positive the positive take is you have a really huge lever to differentiate and to control the user experience and to improve it uh through the way that like you design your sequencer and the design the way that you design your mol and the choices you make in Block building and when you have a kind of interface like that established there's so many things that we didn't even get to touch on today right like from te co-processing to like multi-prover systems in in bridges like you know AI agents like really really like cool stuff that you know some way

or another are like all connected to to block building right and like the idea that like time matters a lot in in the block building process and like for it matters for financial applications most of all right so if you're ethereum like you want to help defi uh or like you if if you're like if you just you know want to help defi app succeed then I think one of the big things you can actually do is allow them to do this kind of just in time computation that happens as part of the block building process like I think we're still like a little bit away from that becoming a reality but I've been such a big uh like bull or like believer on this idea for many years now that like so much computation should actually

happen off chain just in time as part of the block building process and I think I'm optimistic 2025 will be the year that I think we will finally start to see some of those things uh out in the wild okay and with that uh thank you very much John uh as always and uh yeah Robert thanks for coming thanks for the really fun conversation thanks for having me it was great talk to you all next time fun thanks guys thanks for joining us today as always always nothing we say here is investment or legal advice the views expressed by the course are their personal views alone please see our podcast description for more

disclosures if you enjoyed this episode please feel free to subscribe and share it on Twitter thanks and goodbye [Music]