205 - The State Of Ethereum L2s

welcome to bankless where we explore the frontier of ethereum's layer 2 rollups today on the show we have Georgios constantinopoulos the CTO and researcher over at Paradigm and Andrew hang CEO of conduit both of these extremely smart gentlemen have unique Vantage points over the future of ethereum's Roll-Ups that need to be shared with the world here is what you're going to hear on this episode today what is the state of ethereum's Roll-Ups in 2024 what's going right and is still left to do how will rollups recompose with each other What

mechanisms are there to help with rollup composability and do rollups even need to compose with each other at all or is that narrative just totally overblown what about rollup security multic client fraud proofs and multi ZK provs why Georgio thinks we get them this year and why Andrew thinks we're entering a golden age of layer 2 deployments and why Georgio thinks we're entering a golden age of layer 2 experimentation I learned so much in this episode as soon as as I'm done recording this intro I'm going to go back and listen to it before we get into this episode though we disclose Ryan and I hold investments in

some of the layer twos mentioned today we also hold eth you can see all Bank list disclosures at bank list.com disclosures now let's go ahead and get right into the episode with Georgios and Andrew but first a moment to talk about some of these fantastic sponsors that make this show possible Bank list Nation I am excited to introduce you to Georgios constantinopoulos the CTO of Paradigm Georgios is an enjoyer of rust and has helped build our eth and op wreath a rust based execution engine for ethereum and the op stack he's been on Bank list before talking about me when

we first discovered its implications all the way back in 2021 Georgios welcome back to bankless hi David and thank you for having us Andrew hang is the CEO and founder of conduit which is a rollup as a service provider many of the layer twos which you have likely used were spun up and supported by Conduit including Zora Avo public goods Network and many others rasses as they're called as infrastructure supporting rollups both present and future have a unique vantage point for seeing the evolution of layer twos which is why we're bringing Andrew on the show here today Andrew welcome to Bist thanks thanks for

having me excited to chat today I think the whole entire ethereum ecosystem the crypto ecosystem would really just like an audit of the state of layer twos the state of rollups there's many many Roll-Ups going in many different directions with different design strategies different holes uh and there's still a lot of like unknowns for what the future holds I think for the scaling of ethereum while rollups did give us a lot of clarity for how ethereum will scale it has also given us a lot of questions mainly in the world of fragmentation and and security so

George maybe we could just start with that can you just kind of give us the vibe of the audit of layer twos how are we doing what's going right what do we still need to work on just overall give us your sentiment check uh yes of course so starting from the basics a it exists and it's real which I think is only its own a remarkable achievement after many years of hard research and hard work and uh I have pulled up here L2 beat which is a great website that many people reference

these days and you can notice that the current tvl in layer 2's amounts to almost 20 Bill a number that has over 10x almost this year um uh the risk is in a lot better spot we're suddenly starting to have rups deploy more fault proofs in 20 24 we're seeing the security councils going live we're gradually moving towards the so-called stage two decentralization in rollups um if you keep looking at l2b you will

notice that there is over 20 or 30 maybe rollups live of various types optimistic Zay some are not even rollups granted some might have the layer two or the offchain data availability but overall things are growing technology that is advancing things are being deployed which I personally find uh really exciting that is on the core Tech side now there's a whole ecosystem being developed around rollup Services there's companies like espresso that are

building shared sequencers for people that want to Outsource their sequencing needs there's companies like conduit which we'll talk about in a second that are building the infra for people to deploy more and more rollups there's other companies building alternative data availability layers and you know there's the evolving rollup stack and the layer two stack that we inv many years ago that is finally starting to play out in production now are we done no we're not we still have a lot more to do the ecosystem still depends on permissioned uh fault proofs it still depends on

permissioned sequencers on in general the stacks are nent we still have a lot to do but that is for the coming years but overall super excited and I think today is a great day to be having this episode because we're almost at a pivotal point in the scaling story for ethereum and its ecosystem what are the big problems that you identifi Georgio as things that we still need to work on in the layer 2 space obviously security uh is one that you mentioned and brought up and will be an ongoing thing that we will need to

work on for a while there's been a recent focus on while layer twos are scaling they are also fragmenting uh which is one of the big problems what would you say are the big obstacles the big research obstacles or maybe engineering obstacles that the ethereum layer 2 space needs to focus on more of course um I think we're actually way past the research phase um this year has shown that we're entering this deep productionize where all the research is mostly done and right now we're just

seeing production as deployments of things that we've known for a while um so right now the things that I think are very important are as you said security so right now um you can check the state of the security of the ecosystem system just by counting the multisig uh that that govern all the rollups as long as we have all of these multisig it's going to be hard uh for us to consider the ecosystem like really mature and secure now what are we doing for that uh we're

moving towards multiple fault prooof implementations for each rollup we're moving towards delays uh on any kind of power that any Security Council must have we're limiting the power of every Security Council to let say only when there's a layer one hard Fork um we have a lot of work to do on decentralization right now most if not all rollups are sequenced by one sequencer usually the one by the labs entity that or the foundation entity

that the company runs um the project that builds them um which is okay given that there is the layer one fallback option however it also means that the system can go down as we saw many times this year with various projects so overall we also have work to do there so one security two decentralization now three there is one not as spoken about topic that is the tooling which is all that we think about at paradig and for

tooling the problem is going to start manifesting when people switch to different op codes when people switch different pre compiles there's a lot a lot of areas where things can start to change and right now people are not um ready for the tooling is not ready to support this evolution in the layer 2 ecosystem for example if somebody builds a new OP code for their Layer Two because they want to experiment now they need to go into the solidity compiler and edit the compiler and figure out a way to expose

that the user and people just don't have the expertise to do that and somebody some mistakes will happen if we don't prepare for that so we really really really need to make the tool link like ready and robust and modular and extensible so that it's ready for this Cambrian explosion of layer to Innovation over 2024 as the barer to entry Goes Down And as the stacks mature and that's part of like the tooling that we're also trying to build uh with our teams uh in an effort to make them extensible and ready for Layer Two

Andrew I want to turn to you and get kind of a similar perspective from you and what you see over at conduit but before we dive into the same more or less the same set of questions uh with your perspective maybe you could also illuminate what your perspective is what does the perspective at conduit give you uh you are a rooll up of as a service provider you talk to a lot of layer 2 teams uh doing different things similar things and so maybe first uh illuminate for us the vantage point that you have in as your role at conduit and then

we'll kind of go into the same set of questions that I just asked Georgios yeah definitely so um as you said we work closely with a lot of the later two teams we work with folks that that want to launch chains um we work with Integrations that need to integrate on those chains and so I think it's a very kind of like pivotal point in the ecosystem that affords us the opportunity to really like see a lot of different things um and importantly see what's kind of on the bleeding edge here and and how we can help enable that um I think the tldr here for us is like um if a year ago I think nobody was really

thinking about many different rollups and um I think uh it was viewed as much riskier I think today it's in some ways kind of like becoming the default um and we're very excited to kind of help with that transition um and I think for a variety of reasons both economic as well as um frankly just at a technical level enabling kind of new applications to be built um we're really seeing kind of the rise of the modular blockchain and I think um in some ways it's kind of the cosmos thesis but playing out on ethereum um where people do want

Sovereign chains but you kind of have those interoperability standards that um allow folks to kind of transact across chain and across RS in a way that kind of makes sense um and so very excited to kind of be at the center of that and and um kind of playing playing a role and um you know all the complexity georgeos was kind of mentioning around fault Cru and removing multisig and upgrades and um it's just one difficult for the labs found or foundations or whoever is running them to do themselves like it takes a lot of time and not to mention

if a random kind of Dev team wants their own kind of rollup um you know in some ways being possible and you folks kind of mess it up all the time and their Bridge gets hacked or something happens and I think one of the benefits of like a Ras provider like conduit is you know you get all that same Great Tech and all those migrations and like upgrades kind of seamlessly and kind of like AWS instead of focusing on building the best data center and like all of this kind of undifferentiated heavy lifting um you get to focus on kind of the important thing which is like building a great

application for your users and so we're very excited to help facilitate that transition MH and so what are the strengths of the layer 2 ecosystem right now what's going well for teams as a whole that that you work with and then also what are the some of the pain points what are the hurdles what are the difficulties that that teams are experiencing yeah so I think the um so one I think the first question when it comes to around rollup is like how do you actually stand this up in production in a kind of secure reliable performant way um and I think a lot of rollup

Frameworks make it easy to spin up a test net U even like a local vers but there's like a huge gap between that and like something that is ready to custody and kind of hold user funds um and that's really where conduit comes into play again making that seamless so you get that at the click of a button all of the work we've put into reliability security performance um Etc um you just get for free and it again it makes sense for us to invest in because we run like hundreds of these across meet and testnet um and so like those

kind of small percentage points that might not matter for you really matter for us and that means you're getting the the best offering I think typically after that the next thing we see is is frankly just like um for lack of a better word pmf I think like I think like early on in the narrative it was you know we just launched a chain and it's going to work and it's a new thing and therefore we'll have users I think very quickly it's become clear that um you know Ro will need to differentiate in what they offer um and I find that

exciting because I think we'll start to see um instead of like clones of your favorite defi app or like you know the the 10th or 11th clone of like Unis swap we'll actually get something new and differentiated and I think like one example that I've been you know kind of was a dark horse for me but like it's been really exciting to see play out is something like Zoro Network where you know they're very focused on kind of the collecting side of things and that network has grown in a really interesting way where suddenly they have all this data around like mins and art that um you know only exists on the zor

network and I think they'll be able to create really compelling applications and new behaviors on top of that so you're saying that you think that there will be a trend away from the highly General layer tws and layer twos as a as an ecosystem as a category are all going to shift the Overton window towards more specialist specialized Niche layer twos that are optimized for more narrow use cases is that is that what you you think is going to happen I'd say we see a bit of both I think It ultimately Ely depends on the

brand I think something like base for example right huge brand a lot of access to you know retail users that just makes sense as like kind of a generic chain that has everything um but I think for kind of your average Dev team that uh you know like a startup right they need to build something new and differentiated um and aren't going to have the same distribution or or brand benefits that some of these larger organizations have and I think the only way to differentiate yourself is really on the go to market and what kind of uniquely is happening on your chain and in terms of the state space um and

that's kind of my recommendation to teams is you know actually building something novel versus um just trying to be kind of your 100th DJ uh defi chain so I want to ask the very big question that I think a lot of people are asking in the ler to space which is few Roll-Ups or many rollups and there are Arguments for both sides here the the few Roll-Ups arguments are the fewer Roll-Ups you have the more net composability there there are so you know there's fewer different uh chains to be fragment Ed around so like you know the more liquidity aggregation

fewer you know fewer networks and the drop down on metamask and so uh more composability is just good ux and also rollups individually have costs and so if we aggregate everything together you can consolidate the cost in order to save money and so these are some of the arguments for why there will be a few rollups but then there's also arguments on the other side which are that technology costs always get cheaper and that's something that conduit is doing it's making it cheaper for for layer twos to exist and that will only Inc uh uh improve over time there's also uh

rollups desire for sovereignty that we know that this is a powerful force games for example will probably want their own chains so what do you guys think which is it is it few rollups is it many rollups Georgio I'll start with you and Andrew you'll you'll follow along yeah of course so there's two axis on the demand side one is cost and the other is customizability or being free you called it sovereignty earlier being able to do whatever you want on cost how to think about it from or how I would think about

it is that rollups is a an elastic scaling solution you add more compute the more load arrives so if we end up having so much compute demand then probably there will be many rollups because it is unlikely to expect like we saw from all the past L1 let assumes that one chain will be able to accommodate the world's compute or that's where I come from at least other some might disagree and that's perfectly

fine um now on the side of customizability of course that comes off at odds sometimes with the perfect horizontal scaling thesis which means that you know sometimes when the demand might be enough for five chains maybe there's 10 chains because people want some extra customization that you cannot get in the other place for example a Big Brand like coinbase or I don't know say Starbucks or someone um for some reason maybe they would want to be on a separate area why maybe there's like a

lot of customizations that they want to make or maybe they want to own the brand or they want to do specialized airdrops and whatnot hard to tell um so I think it is hard to bet on few as in you know one or five or 10 um so and I would probably expect a power load distribution to nobody's surpris on where the demand gets uh kind of like concentrated let's say in the top whatever but I expect a very very long tail of chains with also varying duration because imagine a game could be

played over a day on a you know people have called these flash chains one day rollup popup rollups whatever you want to call them um you know maybe you play a game for a day um an onchain game that game maybe it has like stupidly High State growth or whatever so it would never make sense to actually put it on a real Network and then just checkpoint the result into a layer two or a layer one or something else so the whole co-processor thesis that many people have been putting out uh it might also apply in the rollups and that would

enable thousands and thousands and thousands of rollups but also with a very small duration power low a lot of activity but the duration of each chain might change demand for customizability also might affect that MH the reason why I like having both georgeos and Andrew here is we have Georgio the researcher and Andrew the market founder uh and so one of the perspectives I enjoy here is that conduit is uh tapped into what the demands of the market are what can you tell us Andrew about the few rollups

versus many Roll-Ups conversation and in terms of what the market wants with your clients and needs uh over at conduit for sure well I guess one point zooming zooming out and kind of tying it back to what georgo said earlier I think like if you believe there's only going to be a limited amount of demand for crypto compute then I guess like the kind of few rollup um kind of world makes sense if you're really bullish on crypto and you know new applications taking off more and more demand I think like by necessity you're just going to need many many of these different kind of crypto

Compu environments so just as an argument for like industry growth as a whole I think it's almost somewhat bearish to believe that there's only going to be a couple of rollups um to serve all that demand um and so you know here at conduit we're we're bit uh we're very bullish on crypto and kind of believe in this bul Centric world but kind of thousands or hundreds of thousands of these rollups um and so just kind of from from that argument alone I think we're we're very excited about it in terms of like what we're seeing from a market level I think ultimately uh one customization is a good point but I think even more than

that I think there are large economic reasons to launch your own rollup and you know I think the biggest factor that we see today um is that when you launch and Deploy on another chain you're essentially paying rent to that chain um and by deploying your own chain you get to like internalize those fees um and not to mention you have more control over your your own ecosystem it's narratively kind of a great thing to do um you get to customize you get to maybe build your own ecosystem on top of that right so you have your own L2 with many different l3s on top other types of applications and so I think just from a

pure economic argument and kind of like a pure sovereignty argument there's just this incredible kind of demand to launch your own block space in the same way that if you look at web 2 today right it's not one big Global computer or like a couple big Global databases like every company has like their own application um and like if you look at Facebook right like they have a ton of apps and whatever they built on top it's this like gigantic system like that could be one big rollup or it might be like multiple rollups um but then you have this long taale of other companies that also have their own you know rollups um

and I think like if I'm thinking through like what the future kind of holds for crypto it's um that model seems a little more clear to me where I think like it's unclear that every application needs maximum compon ability at all times and therefore needs to pay all this rent and like all these other things to a single chain um it seems clear to me that kind of like asynchronous message passing or like if you look at you know apis today right it's like kind of this asynchronous kind of web hook kind of thing or like you just have an Integrations API that happens less frequently than kind of normal um but

then you collocate the logic that's really kind of important um and you know I think the scaling argument this kind of like demand for customization this um demand for like economics um is really driving what we're seeing in terms of this Cambron explosion of chains yeah the if we're going to see thousands and thousands and millions of chains we need infrastructure that we can like copy and paste right we need highly replicatable uh infrastructure in order to make that happen uh and this is where a lot of the uh battles are being fought in the layer two space from all

the like the super chain standards the the optimism super chains arbitrum orbits polygon supernets zkc and Hyper chains uh the way I think about these things is that they're all economic zones because the block space is very alike inside of a network right and so like the op stack main net is very alike to the Base main net right and so these have a relationship with each other that's more close than for example like base is to arbitrum and so these this is how I think about these things economic zones uh and they can engage in trade

with other economic zones right like arbitrum can trade with optimism via a bridge like a cross but trade is going to be easier inside of the optimism super chain trade will be easier inside of an OPM orbit and it'll be a little bit more costly to go between these things this is my perspective for how I understand it George yours how do you think about the whole evolution of super chains yes uh maybe to give you a bit of a cynical take to start um anything that we say in this conversation will probably be more speculative and more an

expectation of what is to happen given that there's none of these systems live yet in um there exists one system called Astria that was deployed a few weeks ago but it's still in test net and it even had some issues at the deployment um when they deployed their Shir sequencer so it might be worth zooming out and thinking what problem are we trying to solve um as a third place um and the problem you mentioned earlier David it's how do we make these different cities talk to each

other in a cheap way without introducing too much additional layer of trust uh when is this useful um one would think first and foremost on defi you know or on transfers I am on chain a you're on chain B and we want to talk to each other without having to think even um about uh where am I sending money to right um and as you said earlier we don't want to be in a world where I go to metamask and I pick from a drop down of 55 or 100 or whatever rpcs that's

terrible user experience and honestly we would have failed miserably if uh we're in that world so the super chains or the shared sequencers or whatever you want to call them they come in as a set of solutions that try to address that by allowing you to interact with one uh endpoint as a user one place um and the sequencer smartly will route to the right area whatever transaction needs to happen um and all of these super chains roughly have that same vision that they

want to abstract away the communication inside of their own ecosystems now there's solutions that achieve that for heterogeneous systems like espresso um and they introduce they require modifications to each of these Stacks to make them compatible with each other for example to make the arbitr orbit stack shared sequenceable with the optimism super chain stack um espresso needs to modify with the same modifications both systems to make them compatible and to

what extent that will be feasible or not uh is TBD and is an exciting area overall one uh point that is worth uh unpacking though is that none to our to the best of our knowledge none of the systems offer um let's say the Holy Grail of synchronous calls across all of these systems um there's no world where you know you can say a calls B which calls back to a and and does a lot of like things together

um unless you're in the design where you you're basically one chain um and that is where the optimism super chain design is going toward um for the most part what you get from all of these shared sequencing designs you get Atomic top of block um in inclusion which is useful for me um which we had talked about uh two years ago David so the idea there is that the shared sequencer is able to guarantee

that five transactions will always be at the top of the block and these five transactions that will be on top of block a and not top of Block B will be extracting some kind of Arbitrage uh opportunity that existed and that makes money and that's a valuable service to be offering and that could be a valuable like infrastructure protocol to be running however to go beyond that you know in terms of conditional execution let's say I send the transaction on a and that means that the transaction also delivers on B and stuff like that I don't think we

have any design yet that is soundly implementing that um and finally a general area that part is still in the research phase I I would think that it's um the the way that people are trying to do it without doing further modifications to their systems I think is not there yet um the most promising design that I have seen is one in a blog post by James preswitch uh which describes a certain way where each blog commits on every other Block in the super chain again research phase um so I

take back what I said earlier that we're done with the research phase there's that uh there's that component that is not figured out but it's also worth understanding that it might not be worth figuring out fully you know maybe the Holy Grail um solution of feature set nobody needs and maybe you can get away with something you can ship like in a year or in six months um that solves the real problem which is not actually the composability but the decentralization the problem is that every one of these systems is run by one person uh whereas ideally there's

not that much of a problem as you said earlier we have a cross we have you know 10 Bridge protocols to do all the transfers people will figure let the market figure that out but the part about the decentralization is more critical especially as you know more and more infrastructure gets launched and we put our kind of like Trust on few intermediaries mhm George could you go into shared sequencing a little bit more uh just what composability benefits does shared sequencing give chains and does that conversation change if we are

talking about uh all these chains inside of a single like um uh setup like a all the op stack chains that are going to be a part of the super chain these can shared sequence with each other in some degrees and produce some composability benefits and then there's also the potential like you said of optimism and arbitrum shared sequencing and also getting those kind of composability benefit overall what does shared sequencing get us right so it it solves you composability and some to some degree

decentralization composability though it solves to a small degree as described today and TBD whether it can solve it to a larger degree and uh surely shared sequencing in the same ecosystem in the same flavor is going to be cheaper easier more compatible where it's trying to make two different ecosystem talk to each other will be harder and might not even be you know desired to some extent yeah cuz that's always what I've uh thought is going to be difficult because sequencing is the Golden Goose

for layer twos like that is where they get a lot of their fees so why would arbitrum and optimism want to give up their sequencing fees to Espresso in the name of um a decent but marginal amount of uh uh um composability benefits right I think jury is still out there and I think depending on who you ask you will get a different uh response um from my point of view um the optimism ecosystem for example wants to build a mod around a different set of let's

say infrastructure components or values for example the governance and the entire process around the law of chains whereas the it's kind of almost acknowledging that a the biggest part of the technology is open source is to be given away it is not really a mode uh or if it is a mode it's a very weak one that's going to go away over time um whereas the real mode is elsewhere so that would be one take uh and also remember that to opt into the Super chain ecosystem and Andrew will tell you

that very well um you pay a rent uh back to the back to the super chain the optimism Collective and uh that is where uh it comes in effectively optimism is offering a service that is the shared upgrades the shared governance and the shared sequencing which is going to be in part um mandated by optimism smart contracts and it's happy to take a cut for it and all the rest let the ecosystem figure out because it's hard to pick a solution yourself maybe you don't even want to build it yourself and

Andrew can cover more on how this looks like from the you know business perspective of how do rollups actually you know because we've done this with multiple customers so far yeah yeah I would definitely want to get there because I think that that conversation of uh chain uh synchrony chain composability and chain governance I think is one of the most interesting ones and I know that Andrew conduit you're like right at the heart of that thing the heart of that conversation but George one last conversation before we I asked that that question to Andrew is universal composability a dream like every single layer two across ethereum

different constructions different setups even if it's like 10 years in the future is that a pipe dream or how far can we get there h i I think it is hard to beat against the mad ethereum scientists just to say it as a as a prerequisite so um you know David it's hard to bet on a 10-year time Horizon my view is that we don't have any fundamental let's say problems to be solved but there's also a degree of like a prioritization and I think that we also