Modular vs Monolithic Scaling - Panel w/ Pavle Batuta, Maxim Legg, Marija Matic

okay so I guess we're starting uh hello everybody and welcome to the modular versus monolithic scaling panel so I'm glad you're all here I hope you're gonna have a good time talking to you know enjoying a little talk I guess I'll uh introduce myself my name is Pablo batuta I'm a CTO at Moonstruck I'm a blockchain developer researcher you know I've been in this crazy space for about five years and change and I'm very passionate passionate about this topic I hope I'll be able to you know convey it in the right way to to you and I'll pass it on to the rest of the guests and panelists so next to me hi I'm Max I'm filling in for James who's on the card uh from Super chain uh opening crypto since about 2016. I've worked on some fairly big infrastructure projects building mining pools validators and specifically indexes um building lots of indexes for many companies and this is our our business now is super Jane which is which is changing the way we do indexing specifically around decentralization and speed um uh and we've been running super chain for uh since December 21 so uh but obviously won't talk about that too much today James is talking on the other stage about us today so I'll leave it with him hi my name is Maria Matic and um I've been in crypto since 10 years ago as a spectator and the last five years or more I've been a crypto analyst and uh around maybe locals know me since I run a bit welcome channel on Telegram okay so um maybe I can open up with a question which is uh uh right out the gate what is what does modular scaling mean to you I'll take this um so modular was was first introduced through um the Celestia or the lazy Ledger white paper um talking about data availability as a means to scale the blockchain so that instead of loading all of the data through the evm and executing is that you could have a data availability proof and then use the proof of that through the evm as opposed to loading all the data through it and then it came into the concept that you could actually separate the execution the consensus and the block building um and the data sorry um and so that's ultimately the idea um is is basically conceptualizing this but when we look at ethereum we we see that as as the monolith or maybe we find out that we did see it as the monolith and that you know we could look at modularizing that it was that as well or seeing it as part of this this modular stack that's being sort of conceptualized and developed at the moment so uh when we talk about modular projects uh yeah Celestia consider themselves as pure modeler project and then we have we can call them modeler first and then we have ethereum which is model it but trying to become modular and many others and also we have these modular first projects like maybe Cosmos which maybe celestians are not going to consider truly modular so these are there are layers to these and um so it just depends on what are we talking to but let's say that a modeler is at least trying to offload or uncouple execution from consensus so we have four blockchain functions and these are consensus that that availability execution and uh settlement sorry so at least one of them for example execution if you uncouple from consensus maybe we can say yeah that's a modeler so now I mean if you want to talk if ethereum has become modular when it's going to become modular is it gonna you know offload all the execution one day to other networks or but yeah now I think now it's more of a story about not about monolith versus modular but more about monolith uh modular first and those who are trying to become modular yeah this is going to be a bit weird because I'm also answering my own question but basically uh yeah I would agree with with the panelists basically we you have this journey that started with monoliths with this idea that you have one signal single Universe of transaction one single well-ordered transactions history and this is a very good thing and all these things the words are capitalized by the way because you can have amazing effects of interconnected and uh transactions that form one big Unity you have the ability to do flash loans you you have the ability to do very Advanced defy and it would be good to maintain this uh uh this property inside a blockchain however it is very difficult to achieve this and I think that modularization is a necessity across all projects I think that what we have and what we're going to see is that most monolithic projects that that are embracing this um this approach are running into trouble so yeah I would agree that basically you have to have some kind of majority in the stack if you want to scale and this modularity is either implicit where you have execution charts you have like a like a blockchain that tries to have execution charts and then have implicit communication between these shards or you have explicit modularization with like separate environments and these separate environments are managed by different domains they're managed even by different operators in case of Roll-Ups and so on and so forth so basically um I don't think that monolithic scaling has a like a it has a well-defined path ahead of itself but it does preserve this this property which basically allows you to have atomicity and locking transactions from different smart contracts and preserve this property which which is very very beneficial so um I guess the next question becomes um speaking of Roll-Ups specifically how do Roll-Ups fit into this modular scaling Vision across ethereum and across different blockchains Okay so if we're looking at Roll-Ups the need for Roll-Ups is we're looking at optimizing our our gas costs we're optimizing like the amount of transactions we can push through let's say ethereum but do it I'd say that the um the tokens that we wish whether the value is locked is on ethereum and we want to then push that out to a layer two to run some execution where our application lives and then right back the states the new States back to the chain so this is the idea of the the roll-up and you have a couple of different types where instead of writing the state back you can just write a proof uh back so this is the the idea of the validium so this then produces uh this can produce a lot of data and this can basically be passed back into ethereum and ethereum then has to process that again so you're basically borrowing ethereum for its security layer and settlement Leia um and your ultimately allowing your application to scale through through many many many more transactions um so this is the original idea now it's been more conceptualized so that you can actually break up the different layers and build a completely independent blockchain so it's not just execution and consensus um off chain like on a layer two you're now basically pulling that back into the ethereum stack with uh the new improvements that we'll have in the upgrades coming so uh specifically we're looking at blob transactions to basically facilitate this this was looking at how we we Shard ethereum's data how we could this is um uh this goes along the line of the proto-dank shotting which is to come which is the um blob transaction that I think is the eip4844 which supports this so that instead of uh executing let's say 10 megabytes of transactions uh through the evm you can execute those separately off chain or on Layer Two but then provide just a proof that that's happened and it can be processed on the evm so it's a very very lightweight operation but you've been able to process 10 megabytes worth of transactions which in theory could get you up to 100 000 transactions per second so that's like a really nice way to scale and I think the innovation has really been pushed into the sort of modular blockchain stack as we're seeing the ethereum now catch up and you know look at adding in these proposals to basically meet the same expectation as such so I think the the idea from ethereum's point of view is that um these layer 2 Solutions are more of a temporary scaling solution for now and that ethereum will catch up and be able to to capture a lot of those tx's so I will have to see what happens with that so well you said it all Max I'm not sure what I could add to that yeah well we're waiting for uh next upgrade and um what was unusual for me was um there were some proposals that implicate that layer one users will subsidize for the cost of transactions for layer 2 users I'm not sure what is going to be in the upgrade in the end but uh we now have different topics to talk to talk about because um if getting transaction cheaper for layer 2 users means getting transactions more expensive for layer one users than where does it put layer one in terms of being used as will the community ethereum community decide in the end to maybe completely or or somehow move the execution totally to other layers so these are kind of interesting questions for me well to to follow it up [Music] if you have a situation where everything is on Layer Two you have a problem because you're now fragmenting the community essentially different layer twos have very different properties and they are sort of mentally mapped into the same category by which you're just saying okay instead of settling on layer one instead of settling on ethereum or maybe it's another chain we are on layer twos and layer twos are very they can they can be very different they have different properties they have different levels of security different levels of centralization and this is um can be a bad thing the advantage is that and I think this was a sort of a I would even say a brilliant move by brilliant Evolution strategy for ethereum is that to Outsource all of this development Acro across a very large number of teams a very large number of projects that take different approaches and compete with each other now as the problem with the increasing costs of layer 1 transaction might be that if you want to post proofs on layer one then this might become cheaper and cheaper and cheaper first with prototank sharding than with tank sharding data availability shards this can become a very cheap operation and it might allow rollups to post more frequently to have more storage uh specifically available for that purpose but the downside and potential danger is if we observe and of course that's going to be my next question what what is the what is even an L2 for you uh it Rhymes uh the the situation becomes like if I want to bypass the roll up if I want to post the proof uh for example a validity proof instead of the sequencer because the sequencer is offline or it's censoring me or is producing an invalid transaction then if uh the space and the cost of transaction execution or transactions on layer one becomes prohibitively expensive then I am effectively being denied service I am being censored and there's nothing I can do about it because the value of my transaction is much smaller then the amount I would have to pay to actually initiate this transaction and um there's already something uh to that regard happening on bitcoin where basically the cost of closing a lightning channel is becoming prohibitively expensive because the demand for Bitcoin quote unquote block space it's a little bit different on their on their in their land uh is becoming prohibitively expensive so I think that you cannot basically scale to l2s without also scaling through l1s and I think that exec sharding of execution is also important it might not be as important in a monolithic sense where you're putting everything on one blockchain and validating everything through one sense of consensus nodes and one set one set of execution clients but regardless you need that blog space to be able to bypass the Roll-Ups otherwise you just end up with a bunch of side chains there are actually not sharing security with anything other than posting proofs via data availability which is only half of the equation you also need the ability to bypass the actual the actual roll-up for it to be a valid L2 and this is a segue into the question like what is a valid L2 in your opinion uh anything that really is is is not ethereum that can write back to it I guess is is a rough definition in my way but if if you're thinking to go along with your question which is about uh parallel execution or sharding um the idea is that why would you want to do this so so currently we're limited to about 10 TPS on ethereum and if we go to uh different chains we can get hundreds of TPS and then if you go to some uh parallel execution chains like let's say Salon or sui as a new one or from like the Libra sort of children chains uh you can get in tens of thousands of TPS because they're doing the the the the process of parallel execution this this ultimately means that if you have two transactions that have different contexts then they ultimately can be executed separately without any state dependency between them so um I think currently some of the limitations are that when you're executing across many contracts the the data that you want to change or the state that you want to change is held within those contracts so in order to do it atomically without double spending is that you need to sequence those transactions going through the evm even though they may not be related at the moment they currently are so with parallel execution you could do this and there are some uh techniques that could be used in ethereum now um namely access lists that you could use which aren't necessarily used at the moment um or we don't see them being used but then if you look at other chains and how they work is that note state is stored within the account of the user not necessarily underneath the contract which now makes it a little bit easier to basically Branch out and run sets of executions on on specific user accounts interacting with contracts so it's easier to grab smaller atomics around that small part of the data as opposed to having global state which now has to be sequenced all in order globally at 10 TPS so I mean that's I don't know if that answers the question or not but it's some information about parallel execution I guess it might be useful so so the original question was what do you consider as L2 right well it's a it can be a state Channel or roll-up and now of course we have arguments about whether side chains are Altus or not whether commit chains are l2s or not but uh I I feel to see why does it matter so much like why this definition is so important what's the point from what angle are we looking at why it's important well um here's the thing how do I heard about different things are different wow now what's important is to understand like what you're dealing with and I think the the different scaling technologies have wildly different properties and they're being bundled into the same category so that they can appear uh like the ones that have certain undesirable properties in certain cases can ride on the coattail of those that are quote unquote better or quote unquote more secure whatever you want to call it like personally for me in a marketing sense right yeah it's like it's like 5G basically you know it's just you you just use 5G because it's better even though it's like a like a wildly different technology and well different upgrade than 3G to 4G to 5G it's like a it is a marketing term but it shouldn't be like Layer Two is something that shares security with with layer one and this sharing of security is something very very specific it basically means that uh what if a layer is um a blockchain or anything that orders transactions and executes transactions then layer 2 in my definition and something that I ascribe to is something that you can bypass on layer one it's that simple so if you have a side chain or a commit chain then it's not a real layer too whatever you know in in this definition because it exists fully and independently from layer one it doesn't need layer one other than it's like a nice add-on to have it's it's a it's it's utilizing it a little bit if you have uh let's say a validium or something that that post transaction data on ethereum in this blob Space Storage or posts courses on the data availability layer if you cannot use this data to reconstruct a transaction and execute it on layer one then this is problematic because you can still be censored on this layer regardless of uh of how well implemented how well executed it is and I think that um scaling Tech needs to converge to this for it to retain the properties that layer 1 has and scale in let's say uh censorship resistant neutral and Equitable way yeah in Broad sense I would just say a layer 2 or what I tend to imagine as Layer Two is something is a layer which offloads some of the work from the layer one so it makes uh it's just yeah it offloads the work and now these are the in the case of modular blockchains it's also it becomes even more complicated because you know you're gonna in the future you're going to be able I mean even now through maybe Cosmos Hub you're able to rent a security from Cosmos Hub or in the case of Celestia they work heavily with the optimism stack uh you you will be able to print security from any blockchain you want which you consider secure so it's quite interesting and then you have all these in modular economy it's get it's getting even more confusing what represents you know which category you're going to put something is it going to be an app chain is it going to be l2l trees I I it's going to get even more complicated I think like before we were all calling side chains as Altus but now it's not anymore but then in a year or two it's going to be complete mess yeah I think the um the idea that as as a developer with your you know to improve your experience if you're building an application let's say it's the the new trading decks and you wanted to build a chain specifically let's say with with no gas allow your your traders to basically execute on on tokens that that you can Bridge across to your app specific chain you can now choose all of these different um you can understand the different concerns maybe you start on ethereum with the evm but then you want to run the evm on a different uh consensus layer um so this is like it really becomes a choice um and obviously to sort of Branch out and to scale across like to scale all of the different use cases that we have um to build a layer one now is far easier and there are far many more choices than just say forking Bitcoin or or ethereum whereas ethereum now is actually the implement the client implementations are far easy to work with and are far more abstracted so that you can replace different types of layers like we just seen with the ethereum upgrade from proof of work to proof of stake this was a module that swapped out it's fairly easy there are many different types of consensus mechanism you can add in there as well so again it will get more complicated and it's just it's purely up to it depends on the trade-offs you want to want to go down as an engineer as a developer and also like what is the end user experience do you want people to make multiple trades and have them sequence based on the context of what they're trading as opposed to globally sequencing everything where obviously the slowness kicks in because ultimately when you're making transactions you're effectively loading all of that information sequencing that in the same execution environment so this is obviously where you know optimizing the amount of data that gets loaded will improve speed massively and then also being able to separate that out into many different parallels and whether that's at the execution layer at the data storage layer um or like what you actually use as part of your consensus um you know it all really um it really depends on on what you want to do whether it's you know security scalability or even if you know you have data that is completely ephemeral you don't need to keep the history of it you know there's lots and lots of different use cases and many different solutions for all of these so um yeah so the way I see it uh there is a whole Forest of solution to this problem and it's going to grow even bigger because there is a quite a bit of an incentive to try and produce different solutions and try and experiment in this field there's there's a lot of money in scaling and I think this this will continue to be so for quite some time but then we might see a bit of a convergence because in the end the use cases the use cases allow for certain things and as an app developer you will converge towards liquidity because if you're doing if you're doing um uh Financial applications you want to do stuff that on chains that have uh liquidity that have that supports your application and if you want to do something that requires uh scaling on a much larger degree then you will perhaps use something like a validium or something like that but in the end you will you will converge towards the change they have uh liquidity and you will converge towards change change to have uh good bridging and that is my next question because one of the things that that you sort of lose out when you abandon this monolith is the ability to seamlessly and efficiently uh communicate between different contracts to to just send a transaction from one contract to another and it just sort of works and I think that whichever definition of layer 2 we can accept the problem becomes that then that if I want to send basically something cross-chain I run into a whole plethora of problems that arises from basically ad hoc Solutions so my question for the rest of the panelists is where do you see how bridges will evolve in this commercial space since the mic is with me uh well I've seen actually two interesting topics in your question uh one is related to liquidity other is related to composibility I guess I will leave compostability to maximum because I think he's uh better at that part and when it comes to liquidity and I would connect it to the network effect and I would like to say just a few words about arbitrum and uh how important from the community view how important it is um to gather this network to get this network effect which they managed through the token airdrop farmers and but also through perp Texas because they managed to capture that sector seriously because there was a huge hole we had the dydx which is a perplex which is not composable and there is a huge problem within D5 Community because we didn't have a purple Decks that you can that you can build anything on top of that and create a whole ecosystem some maybe I don't know Asset Management walls or whatever you want to connect to that or Landing applications so when the arbitrary came all perp Texas started using it and that's where and they have a very good Revenue model these are profitable platforms and all the biggest decks except perfect access except the ledx are right now on our bedroom so our bedroom right now has a huge use case and they had they also did a very good thing of Distributing their tokens to not just to users but also to those so all these dolls that represent certain platforms certain dexes are Distributing their tokens these tokens as reverse to the users we we're gonna see how much of these liquidity and uh locked capital is gonna is gonna be sticky how much how many users are going to be sticky uh once the rewards dry up but considering that these projects tend to be quite profitable I think that our bedroom has a very nice shot through that sector but we're gonna see you know so I think it's going to be what I wanted to maybe touch upon is a network effect and how difficult it might be for some other projects ltus to you know take their share of the pie um [Music] yes so in this is an interesting um part of the ecosystem is that when when we talk about layer ones and we talk about bridges ultimately the idea is to move the token from um chain a to chain B now what we're seeing with modularization and and the way uh bridges are set up is that you you have basically a small validates a set that effectively has a multi-sig control over the tokens of that bridge and obviously we've seen many exploits and hacks and so on around this now what we're also seeing is that bridges are turning into their own execution layers their their own blockchains themselves so this is now changing the dynamic this is becoming enabled because of the modularization that ultimately you want to be able to take from um like purely from ethereum you want to be able to execute within a smart contract on one chain and have that settle on another chain and this is now being enabled through through summer chains um that we see in the ecosystem that we have like gravity Bridge we have um axler all offering that these types of uh let's say modular Stacks behind their solution um and this is where we were seeing um a lot of movement there are also projects that are looking at sharing liquidity across different chains so basically allowing you to basically have access to liquidity immediately across chain um I I think this Con I first saw this concept I think with with RV v3's portaling system where you could basically pull the liquidity across many different chains and move it to where it's needed the most so if you had an account on on ethereum you could move it across to to arbism quite easily uh without even having to um do the bridging yourself I think obviously when when you are bridging your tokens it's very difficult because you need to have gas on the other side of of that bridge and this is like a common problem we have um and obviously there are nice solutions that we can use like um with li-fi allows us to basically go across the bridges root for us and give us tokens on the other side um you know these are nice improvements But ultimately what's happening is that you could have your custody held in ethereum and you could execute on say Arboretum on on on on mycelium or something um and then bring all your profits back let's say within within the same transaction um you know that's probably the power of what where the bridges are moving to and and I think uh the modularization of blockchains really allows us to go from being a very simple mint and burn Bridge from one side 30 minutes later then you might see your tokens if you're lucky to actually you can you can build a full consensus execution layer to allow you to move the tokens across to different chains and to execute and to come back um you know and and to be able to do that immediately to be able to do that um more securely I think that's where a lot of these uh sort of bridge type Chain Solutions are are going so maybe to add just uh speaking of this modular approach one of the advantages is that Roll-Ups have if they are Tethered to ethereum is that they can use ethereum to prove to each other that a certain block has been executed with certain certain State changes so you can have essentially the same solution that you would have in a monolithic stack which is basically you have a shard in this chart you have a contract you have an account especially today with uh with account abstraction basically you have a you have a smart contract that gets locked and then executes uh stays Frozen with some State executes a transaction this transaction is routed to another chart or in this case to another roll up and then in this case performs a transaction hits another contract it's another contract and then finishes and returns to to the original originator so you could in in theory have a a very tightly coupled system that uses proofs basically between chains so you can say okay I've initiated a transaction on this roll up here's the proof you have a certain waiting period Then the transaction resumes it starts fires off on on a different roll up then again this role approves to another roll-up that uh the continuation this transaction has has occurred and then so on and so forth until the transaction finishes until you run out of gas the advantages uh in this case are that a you're using the same token for gas which is uh you don't have to worry about the the end of the bridge problem uh you're not having enough tokens to complete the transaction and you have this one layer that everybody can agree on that proves to every other roll-up every other Shard quote-unquote that the transaction has transpired so essentially I think that um this is this will go beyond just liquidity Beyond just the concept of tokens I I think the end game is that you have this seamless sort of bridging between Roll-Ups through a shared standard it's very cool okay I have no idea how much time we have so uh I think we have five more minutes so um uh questions and closing thoughts but first questions here we go so um so aren't roll ups really just fixing deficiencies in get yes yes if you have Geth and if God was good then none of this would be a problem but unfortunately having Geth be good enough is impossible so here we go anyone else don't be shy what do you think of Wrath um next question I'm not I'm not that very intimate with like different execution clients to be honest yeah I I think the ones that we're I mean the one I I really like is evm I think um this is been tried and tested and used everywhere um it's it's probably like the go-to all of the the smart contracts despite all of the issues you have with let's say security or ergonomics of of development and building is that it is actually a proven system um we've seen a lot of um copies of the evm being used in different scenarios uh one of them which I'm paying close attention to is looking at how cbdcs are reusing this technology there's a case to be released on Enbridge which was a Chinese um uh International settlement system between countries in Asia and that was used in the evm and obviously tried and tested contracts that we use in in ethereum in open source this is a permissioned permission chain between between these countries but it just shows the power of of the evm and and that you know if you wanted to trust x amount of billions or trillions that these countries are trading and settling it immediately um then that's like real validation real proof that it is a solid system and then you look at the Innovations of um let's say fuel as as one of the other uh sort of modular execution layers this has um it's it's using rust as as its language um so there's some nice ergonomics than say solidity well depends on on what your preferences are um but also it allows this parallel execution to happen as well um so that you can actually execute many many more transactions um could could the evm do this could we change ethereum to copy a similar model um there are hints that yes you it is going that direction there are some changes in the solidity language specs that are that's allowing us to build in these Primitives um but it's going to be very very slow for ethereum to catch up so I think this is the nice thing about modular is actually gives applications and developers um and businesses the ability to actually prototype and innovate on the ideas they want instead of waiting for the chain to catch up with them they can go ahead go out explore do frontiering with new technologies and then come back to where it actually matters where actually the the tvl is where the actual um where people want to keep their assets where they feel that they're secure so that's my opinion uh it's nice that you mentioned the fuel as well because I think these are the projects that needs to be constantly checked and and followed like Celestia fuel Cosmos of course and based on the Lessons Learned there I think ethereum will kind of try to follow I mean Network effects with ethereum is still the largest so yeah is that it have you gotten any questions did you enjoy I hope so in any case yeah I uh thanks everyone for your time I hope you were entertained and maybe learned something new and uh we were very very glad to have the opportunity to talk to you all and if you want you can catch us later in the in the conference and we'll be very happy to answer your questions so thanks you thanks again thank you very much for your time and um that'll be all thank you foreign