Radina Talanova, L2Beat: State of L2s | ETHSofia

[Music] my name is Rina tanova I'm a product owner at el2 beat and if you haven't heard about el2 be it's a research and analytics company uh focused on ethereum scaling Solutions or the so-called L tools uh we are a public goods company we provide all the research and metrics that we have for free and we mainly focus on the security of those Solutions and the risks that come uh with uh using them so uh I'll start with some context um so why do we need l2s [Music] um some of you have probably heard about the so-called scalability trilemma um the thing is that uh there are three properties uh that um are desirable uh for a blockchain this is capability decentralization um and security um so uh the TR trma states that blockchain cannot have all three at the same time uh and to give some examples uh Solana for example has a high trut uh low cost uh but um it sacrifices the uh decentralization because of the high requirements for the nodes uh Bitcoin is another example where uh it has uh high security uh and decentralization but uh at the cost of um low uh number of transactions that can be included in one block and then uh ethereum uh ethereum at first prioritized the security and the decentralization but then started facing scalability uh issues and um to solve this uh ethereum uh to come up with a way uh to push all the all three buttons um and uh released their Centric CR map um that includes those l2s uh as a main focus so uh what l2s are um l2s um inherit the data availability the consensus and the settlement from ethereum but they um do the transaction execution of chain um here in this example uh those are the solutions on the left so those can be either rups that use ethereum for data availability or validium and optimium uh that uh have an external solution for data availability uh volumes are the ones that use ZK proof systems and optimium use optimistic or um fra proofs so um I'll start with some uh ecosystem metrics uh to see where we are when it comes to l2s um first of all on the left you can see the current value locked in all L2 Solutions it stands at uh around 36 billion um and then uh when it comes to activity right now there are um around let's say 300 uh transactions per second um settled on uh layer tools um compared to um 14 um on ethereum uh so the scaling Factor it basically shows how many more transactions can be executed on l2s uh is uh 22 um in the table below you can see the top Solutions based on the total value locked right now arbitrum is on the top and so uh if we're looking at those charts things are uh looking okay I mean activity is moving to l2s value is moving to l2s so uh all seems good um there are uh over a 100 scaling Solutions right now uh most of those are optimium which is not great and I'll explain why in a bit uh but the good thing is that uh they make up for uh really small percentage of the total value Lo so let's say that um most of the funds are still on rups so uh it's not that big of an issue um and then if we look at the technology Stacks uh most of the scaling Solutions right now are based either on the optimism Technology stock or the arbit from orbit one uh but overall um all projects that are building some kind of unique technology are aiming to build ecosystems around their own stock so uh more stock are expected to um be created uh right now as an example we have the polygon cdk stock star cakes uh we also have the zc casing stock uh and some other smaller ones um and a final set of charts uh here I just wanted to um say that um the majority of the TVO uh comes from um solutions that are the launch through the op stock or the orbit stack um and same for the activity uh it's close to 90% uh so uh so far things seem okay um but um let's see where we are actually with the scaling road map uh this is the road map that eum Foundation released um couple years ago um and the part that is interesting to us uh is the second one the search it deals with scaling uh and uh outlin some mouson for the uh Layer Two Solutions so um here uh you can see that the first one uh was already completed uh so this year in March uh EAP 4844 was released with the Denon upgrade it allowed for blobs uh to be submitted for ethereum and it um is something that's important for data availability because previously uh submitting all the transaction data as a layer to to ethereum was uh super expensive uh right now the costs are um I would say around a cent per transaction or something like this so it's very cheap um but uh it's not over I'll mention why in a bit um once this um uh got released uh 30 around 30% of the solutions started using blobs uh and that's 70% of the robs uh and you might be wondering uh why if it is so cheap uh more solutions have not started doing this yet and the reason is that uh right now only six blobs can be submitted per block and they uh you can submit a limited um amount of data in each block um and there is a target that's currently set to three blops so um whenever um the amount of blobs goes above this target the cost for publishing the data to blobs increases significantly um and we can see this in this example uh I think at this point in time layer zero launched their air job so there was increased activity on arbitrum um and the costs here for all solutions combined for a day uh jumped from uh around 20K to million dollars uh and that's a pretty significant increase and that's why some solutions might be hesitant to um um using ethereum as a DA layer uh because if more solutions start doing this then we'll probably start reaching this target way more often uh and there won't be a big difference between using call data versus uh using blobs uh to post this data so there are couple of improvements uh that are planned here and I'll mention those a bit later uh but uh let's look at the second part of the scaling cow map um and um the second part says that the first Milestone is uh for l2s to have uh limited training quills and then to move to having no training quills and uh we at l2b uh came up with this framework that um allows us to evaluate the l2s on a couple of uh risk properties in order to be able to track the progress towards better security and decentralization um so now I'm going to uh go through each of those properties uh and we see if any progress is being done there so when it comes to State validation and why is State validation important um so um every time uh there are some transactions ethereum has to be aware of the changes in the state of the Rob um so um in order for this to happen there is an actor called a propos proposer that is submitting State roots to ethereum uh the thing is that um someone has to uh guarantee that those uh State roots are valid and they uh do not represent an incorrect State because if that's the case then funds can be stolen uh so that's why we have proof systems um and the thing is that right now most of the ZK rups uh have working proof systems they're not perfect but they're doing the job however we cannot say the same for optimistic cups um when it comes to optimistic crups the only one that has a permissionless uh proof system is op Manet uh you can see that arbitrum um has fraud proofs but their proof system is permissioned uh usually uh that's the case if there is some kind of uncertainty uh about the security of the system uh and that's why only a permission set of actors can submit those uh State roots and can challenge them um op Manet released their proof system uh recently and the rest of the op stock chains have not followed yet but hopefully they'll do this soon um when it comes to the zik robs as I mentioned um the systems are not perfect because uh this is an example with Linea in their documentation that say that um the proes Do Not prove 100% of the evm op codes uh and that's why again there is a permission set of actors that can um submit those otherwise um there is the risk of um funds being stolen uh this an example with arbitrum uh they currently use the arbitrum classic protocol U so they have a WID listed uh actors that can submit or challenge um the state route um and uh the project currently can be subject to delay attacks uh this means that uh withdrawals might be delayed however arbit Trum uh is planning to make some significant uh upgrades uh with the launch of both uh it's expected to come to main it I believe beginning of next year um and uh with both they're supposed to remove this permission set of actors because they're going to introduce bonds so uh all the actors that uh want to propose or challenge route you have to St a bond that has a very significant size so this limits the number of actors that are um willing to um play those dispute games and willing to attack the system oops wrong button um so over the past months with had l2b spent a significant amount of time analyzing those uh proof systems uh with a focus on optimistic crups um and we came up with another trilemma when it comes to developing uh fraud proof system um so projects basically have to chues between promptness uh safety and decentralization uh promptness deals with those uh delays so uh if they decide to prioritize um the ability to withdraw faster even if there is some kind of dispute game uh then they would prioritize promptness um safety deals with uh basically um how secure the system is and uh How likely it is that funds are uh stolen and decentralization um um deals with the amount of actors that can participate in those dispute games so if we are to rank those systems uh in this tremma uh okay before that I just want to mention that um this introduced because uh there is this attack called resource exhaustion attack um and um it allows a malicious actor to Outlast The Honest Defenders if they have more funds than them and if they can continue playing those dispute games longer uh that's why this trma uh exists and um with arbitrum classic that is currently on mainnet um the things that are prioritized are safety and decentralization because they have a low Bond requirement um and they um do not allow for parallel challenges to happen uh but this might lead to some delays so they're not prioritizing promptness uh same for cesi uh protocol uh then Bol uh as I mentioned will introduce uh larger Bond size uh which means that not everyone will be able to participate because not every will have this amount of funds uh so they would prioritize safety and promptness and op mayet uh their proof system that is currently on Main net unfortunately uh prior prioritizes promptness and decentralization uh but it's not secure enough and this is because right now um uh if a dispute game is ongoing uh there's a slight uh Advantage uh for the attacker uh so at the end if they have significant amount uh they can Outlast the Hest um actors um and you might be wondering then um is op Manet secure uh enough to use um it is uh because uh there are still those training quels uh so op Manet uh has a multis that includes the their security Council and the optimism Foundation that can upgrade the system at any time override the proof system so if something happens then you can trust uh The Entity to um uh to react quickly and uh do the upgrade uh but at the same time uh there is another risk here because um imagine that the security Council and the foundation are malicious or somehow all those Ms get compromised then um they might do a malicious upgrade uh and so what do you do then um that's why there is this exit window um exit window uh is basically the time you have to exit in case there is an unwanted upgrade uh um so to bridge back to the A1 um and you can see that op Manet doesn't have uh such an exit window right now so not great um arbitrum has one and the way we calculate the exit window is that we take not only the delay on upgrades into account but also um the delays for uh you being able to become a proposer or you being able to become a sequencer because it might be that uh there is a malicious upgrade then the centralized sequencer or uh the proposers uh do not in include your transactions on the A1 so you need to be able to do this by yourself somehow in the worst case scenario um so right now arbitrum one is the only system that you can exit in a completely permissionless uh and Stressless way uh this a tweet from yesterday because radiant uh Finance uh got exploited uh and we just wanted to show how important it is to um have uh larger thresholds uh for those multi that are used for the upgrades uh and the situation now you can see is uh not perfect those are the thresholds with the value that they are securing um when it comes to the sequencer failure uh right now most optimistic crups have some kind of mechanism uh for you to uh sequence the transactions by yourself so you canot become a sequencer if the sequencer is sensoring your transactions Z robs do not have this option because um there is a concern that uh some transactions if they're too big uh won't really work with the ZK proof so that's why uh um this is not allowed for now when it comes to the proposer failure um most of the optimistic uh rups do not have this option because again they don't have a fraud proof system so if anyone can become a proposer uh then they can uh propose an incorrect uh State rout and last part of The RK rette is data availability uh we have a dedicated P to this with all the details but it is important because for optimistic crups uh if you want to prove that a state is incorrect you need all the transaction data to be able to construct the correct state for zov this is not needed uh but uh it is still um uh needed if you have to uh exit the system in a trust this way so if you have to become a sequencer or a proposer you need to be able to reconstruct the state of the system from the data posted on the A1 uh and for this one uh to be able to rate the solutions here because over the past year apart from etherum launching blobs uh there was Celestia end da and so on uh we had to come up with another uh framework uh it is right now in our Forum so if you want to check it out uh you can do it and share your opinion before we uh publish it on the website with all the risk analysis for all da Solutions and what about L trees uh so this a case that we had to handle this year because um arbitrum orbit uh their strategy right now is to uh attract more projects building eles on top of arbitrum uh one so um for L3 uh the risks basically is stuck uh and we assume the worst risk for all of the properties because uh even if for example you have an exit window on Sano or on arbiton one if the other chain doesn't have one and both projects decide to cooperate to steer your funds uh then you won't be able to uh Escape uh so to be able to rank projects uh according to all those uh properties we had to come up with a ranking system called stages as you can see right now there are only two projects that have reached stage two four projects have reached stage one and 34 projects are still at stage zero and this is rups only so solutions that use ethereum uh for da um in the last year it was only Op mayet with the introduction of the for system that uh was able to move up in stage um so going back uh to this uh road map I want to uh briefly touch on the interoperability part uh so here there are uh there's lots of research being done uh not many solutions are live on mainnet yet uh right now it seems like uh all those ecosystems are aiming to uh allow for interop interoperability within their own ecosystem um and um for example optimism has their own Super chain erc20 standard but uh then it doesn't yet have the messaging protocol and the fra proof system doesn't work with it um zking and polygon have shared escroll so all the funds that are bridged through the different rups are uh pulled in one contract on Main net um and then uh it is the single source of Truth for the balances on all of the uh scaling Solutions um at the same time for optimistic props it seems like it's a bit more difficult because they cannot benefit from uh some of the research that's done uh so in their case shared sequencing is being explored so uh there can be sequencers that badge transactions from multiple rups in one batch um but it seems like uh so so far those sequencers would only work within the ecosystem because um they need uh the projects need to be using the same contract on Main net so it might be that in the short term projects have to choose between um building kazik R up that is interoperable with all other rups um but not the ones that are using any of the um doop stock orbit stock uh or building an optimistic C that's interoperable with all robs within the same ecosystem but not with robs outside of it so uh to conclude um it seems like uh ZK is the favorite approach uh in the research Community however if we look at the ecosystem metrics uh ZK UPS have um lower uh adoption rate uh interoperability is yet to be solved um the next major change uh when it comes to data availability and scaling is peer do uh so it will allow uh The Blob blobs limit and the blob targets to increase so hopefully more projects will start using ethereum for da um and there are still many trading uh training quills that need to be removed uh we as l2b are trying to uh push the projects to do this and will probably become way stricter when it comes to evaluating them uh in the next uh in the next year uh so yeah progress is being made but there's still a long way to go thank you [Music] n