The Future of zkEVM, Why it needs time to Evolve | KBW2023

this is yes this is this is working good afternoon everybody um thank you for the warm introduction and thank you very much to hashed all the audience and uh soon to be better introduced our two panelists on what hopefully is not only a very uh technically important inflection point uh the the ZK VM the future of which we have two very Stellar founder co-founders and and Pioneers to to scaling ethereum but I think one of the um most interesting of nent Technologies in terms of the future landscape and how it's going to affect the ethereum and the rest of blockchains as they uh has change the change the world and uh coordination mechanisms as we understand them now uh so the title of our talk is the future of Zed KVM and why it needs time to evolve we want to spend a little bit of time speaking about scroll we got Sandy Pang over on the the left and polygon Hermes and and Z KVM with David Schwarz here in the middle um some of the Milestones that they've done some of the progress that they're up to and of course importantly how they see in this what will be a multi-decade um Journey evolution of of the technology but just before that I want to allow them to introduce themselves uh Sandy feel free to go first sure it's a pleasure to be here and my name is Sandy I'm from scroll and um we're a project that's been building um the zkb and product for the last two years and um our our team is distributed and um we're trying to bring to everyone the best developer experience and try and bring um 100% kind of evm equivalent to developers so so that that there's a seamless migration process and uh we're currently on sapoia testet and uh we'll be coming to mayet very very soon feel free to go second thanks thanks for having us here uh I'm David Schwarz from polygon um I was one of the original co-founders of the Hermes Network team and we were acquired by polygon uh two years ago so um now my role is more open I'm just more involved into the engineering initiatives from polygon because now we have uh a set of uh proposals to the community uh related to the polygon 2.0 Vision uh that's a you know set of uh changes we want to do uh one of them is the launch of the polygon CBM that's the project I was contributing most uh but we have another initiative like uh upgrading the the polygon POs to CK and uh launch a new token and also improve the governance so there's a lot of things going on in polygon now related to this vision and uh yeah super exciting times for for everyone in the space related to CS awesome fantastic thanks for providing a little bit of color on what's been uh the progress and I guess top of mine for the for the perview so far snowballing um on Sandy speaking about the test net and Main net and David speaking about um polygon Z KVM as Main in this case definitely a big congratulations in order for the paraview uh these are enormous Milestones that have both recently been completed um so that we can take a holistic View and and dive into some of the moving Pieces perhaps if you could each tell us just some of the the main challenges and the main progress that have brought you to the point of launching sapolio testnet and polygon uh main net respectively okay so uh well um this project is as as Andy was saying is um Big Challenge it's a complex one um the ckvm as a technology two years ago was considered even impossible something not feable uh both scroll team and polygon and others now we are taking this path of providing uh full equivalence or uh the seamless experience for developers into the C rollups too uh which was a kind of a very complex uh project to do and to build so we were you know covering many phases of the project um this uh is a long path for us it took one year and a half to build this and going into maget but I can assure that it was very complex initially you need to figure out what's the solution to this problem which seems so difficult to solve and uh we even needed to create some programming languages because uh the evbm is not a friendly system in terms of CK CK for people that doesn't know is a cryptographic environment that enables to proof computation uh in this cryptographic model and creates a small proof that can be verified on chain this is the foundation of the CK scalability so the ebm was not friendly in this in these terms and we needed to to build many components um after this process of building the components then it comes the the model of uh assembling it together and make this something practical because also another vision of this technology is that it will be so expensive to be proven proofs will be costly and these kind of assumptions and we get into this optimization path and then we prove that this practical and it's very optimal and efficient also uh but then also you need to audit this because the the model of trust for secet rollups is related to this uh cryptographic environment to be considered secure so you need thirdparty Auditors to certify this and you need to provide a process to anyone to verify that the pro is there the Pro is correct and the pro is there so this is a long path we were tooking and uh it was successful like end of March this year with the main launch and we are super happy how this is going we have U uh 400,000 unique addresses using that uh 10 10,000 smart contracts deployed because it's a same experience as deploying on ethereum and uh 5 million plus transactions for us this initial phase was about proving that this Tech is so solid stable and we can just provide this as a mainstream service so we accelerated a lot the the generation of proofs or Pro is now uh created more than 700,000 proofs already so it's super solid and we are super proud of this but still there's a long way to go and we will discuss probably about this later yeah I think as David was sharing the zvm is a incredibly complicated engineering task and I often kind of um think about how amazing it is that you know both our teams and also a few others went from research to product within such a short space of time and another interesting thing that I've been kind of thinking about lately is how teams with slightly different approaches and using different tools have settled on a very similar looking type of product from the developer perspective um I think the the the polygons zkn team and us we started off from very different starting points in terms of the intent and also the the tools that we've been using to build this product but the end product that is shipping that's developer facing um is hopefully very very similar and I think the reason behind that is you know ethereum has such a wide array of network effects and and and and a large number of developers already so it makes a lot of sense to try and leverage some of the past experience and and and and learning so that um so we can kind of gain traction earlier without um W with a similar kind of um both testing environment tooling environment so on and so forth so um I think we launched our initial test net um around the same time probably about a year ago and uh kudos to the polygon team they've managed to launch mayet and we've been on test net um for just over a year now and um and and and and we have mainnet probably in Q4 later this year and in the last year or so I think we've learned a whole ton from the development process and um the the the amount of traction is has taken all of us by surprise um we have about 60,000 um smart contracts deployed and we have almost kind of 9 million unique um smart contract Wallets on Gorly and then now on sapoia and um and I think the the the the the process of adding a second layer to an existing vibrant ecosystem is a really interesting one I kind of often describe it as trying to kind of fix the plane while it's still kind of in the air and um and I'm sure the that the experience is similar like once you know once mainnet is launched there'll be a lot of improvements to be made and um and and and this is you know kind of probably like a decade long journey of improving proving optimiz uh efficiencies and also improving developer experience it's it's good to hear both of your Journeys in in that perspective I think there are like a facundity of of similarities the first and one I I share the the massive pain is is the the proof system and the design I spent four to five years now since since since 2017 right from the bulletproof days until very very recently um before spending a little more time on the investment side where I was designing implementing these proving systems with both both of U members of of your teams and even prior companies before before both the uh the particular protocols existed so I I definitely share and feel that pain and to David's point it's been really really surprising to see the evolution um of of of speed and capability on things like the polygon main with things like um Scrolls work and and codifying all of these circuits into ZK it was I think arguably um seen as like maybe not possible especially compared to the timeline of something like optimistic um VMS and optimistic um encoding sorry of of these rollups so I think you we all share in that pain and now something I've myself abstracted away from not working uh on the protocols specifically is the best way to design it for as you say what's important is developer experience whether it's from taking you know different abstractions from from bite code equivalence or creating languages knowing that solidity as as wonderful it is for making these fantastic tring complete uh arbitrary programs when putting those OP codes inside of a circuit or inside of a a very particular finite field is a complete nightmare but I would like to um talk about what you see now that you've got to an aifi level with the main and I I was speaking to David before and he was speaking about the the maintenance of of this plane as Sandy said that's in the air is a lot more intense than he was expecting and Sandy I guess yours is Off The Runway but but ready to hit the 10,000 ft what do you see as the immediate next challenge more in terms of the technical design and then we'll we'll delve into some of the the developer and ultimately user experience in a minute um ultimately we've never launched an L2 mayet before no one on the team has and um arguably not many people in the world have so it's quite a unique experience and we're also kind of learning on the Fly trying to preempt what will happen I think one thing that we're certain is that our learnings will accelerate and we will um get a lot more realistic user feedback um I think when people are not bridging real assets on chain they tend to be uh probably a little bit more tolerant of user experiences and and I think you know we're ready for the journey for things to get a little bit more real and um and I think we've been we've been spending a year trying to prepare both internally in terms of our mechanisms of how to deal with um reports and also how to build up a a kind of a people system as well as a a kind of a monitoring system on chain to make sure everything runs smoothly um it's I see the mainnet as potentially like a starting point um just like when ethereum launched mainnet was essentially the starting point that's when you get to really engage with the ecosystem meet your C meet your Builders meet your developers and find out what they need and um and and even now I think you know we've kind of built the beginning of a early ecosystem um mainly players who are established in the ethereum space already and uh the feedback we're getting is um incredibly um insightful at least like new to me um I think I'm learning every day about some of the the difficulties devs are facing and I think crypto being such a um a value driven um kind of use case it means when there's money to be made people are very very tolerant of us X problems and uh as more and more different use cases are being unlogged because layer 2's pricing enables that onboarding um we're we're seeing more and more um stringent requirements and also uh users are asking more and more difficult questions about how do you kind of bridge assets from from say like eth on polygon to eth to another L2 and how does that come to scroll as well as you know how do we make the um make the L2 more on the on on the back end rather than something that you need to choose from essentially users don't really care about you know proving systems they don't really care about any of the infr layers that that that we care about as infra Builders and um and I think it's about kind of a managing our standard of building as well as meeting developer needs as well as user needs um but yeah it's you know we're just constantly learning yeah we we have a lot of challenges because once um you launch mainnet uh you have this uh interest from the community we have here in the room some partners that uh we were basically positioning the ckvm for defi because of the security levels we Pro we provide uh because in polygon we have also this POS service POS Network that's kind of the most adopted chain uh in the industry so we have this super low fees and one side and P and we were providing this ckvm that has higher fees because data availability is on chain uh the proof is not kind of a big component of the cost but the data availability is so we have several challenges one I would say is important optimization because uh many users are asking oh where the fees are in reducing in ckvm so there an upcoming change from ethereum the EAP 4844 the prot sharding is going to be very important for us because with the new this new data structure of allocation of data in layer one we'll be able to reduce drastically the Fe and the CBM um also we are optimizing provs every time the provs are faster and more lean uh as of today our provs require big bunch of memory and we are trying to reduce this and provide better uh capacity on on this side um also we have announced with polygon 2. z um the POS network will become also a layer two this means that we need to connect the posos to a CK rollup or ck Pro in somehow so the the POS in this sense will become the biggest Layer Two ever in terms of TBL activity asset all of this um and this connects with our vision of this polygon 2.0 ecosystem of chains where we can provide this unified liquidity based on CK proofs so we have currently a bridge that connects uh ckvm with ethereum and this bridge support multiple SEC rollups that are compatible but at some point we are just uh announcing and we are building this aggregation layer that would allow this ecosystem of CK uh chains to connect one each other and compose smart contracts so this is allowing um to to have this pool of unified liquidity because liquidity fragmentation is a problem in this ecosystem on player two so uh for us this is a very important strategic topic uh to overcome but uh it's another challenge and then we have this other security challenge because as I said before uh this um model of trust is becoming a solid PR uh we have a a b Bounty open we receiving very good feedback from the community on this side and we are just maturing a lot this CK progress but you see there are many things going on forward is that uh we also have the burden of maintaining U main net because um we have users we have uh we are deploying oracles we we are building all this ecosystem of tools to become a complete uh you know uh tool for D5 for example uh as s was saying our benefit is that we reuse ethereum tooling and ethereum ecosystem but still you need to complete this this chain needs to be uh every time more and more complete with more tools so the the experience for the D apps and developers are the same as ethereum in some way I think a lot of what you drew on there David is this uh monolithic challenge especially now you have polygon proof of stake becoming L2 and the polygon Z KVM of unification between the pair of those trying to support and I'm sure make new mechanism designed for cross chain liquidity and bolster up all of these different apps across those two as well as all of the the l2s that are that are growing into the space of ethereum and Sand's first answer is what I'd like to draw on just a little bit more which is the the developer experience this touches on a holistic sense I think almost every element of the design stack every aspect you have something which maybe many many would argue isn't quite Z KVM like like starkware then perhaps they would also do themselves won't got myself caught up on particular n cature here but those who've made their own VM style instruction and put it into a circuit and then decided that it was better to create their own um general purpose programming language some DSL that allows people to compile down to circuits and then right on the other end you have those who are trying to go as close to ethereum in the original evm specification as possible to make this bite code equivalence such that people or developers should I say who've who've designed apps in the past in solidity can just pick them up from ethereum and pop them into um a a a lower cost and and and lower transaction fee environments such as Z KVM and everywhere in between on that spectrum and we also have an uction of of discussions around things like um risk zero uh so so the risk five instruction sets and people who would sorry um specifications that design VMS specifically for binding static languages like rust and C++ and all of this is aimed at where the developer sits I feel like saying it so much on this stage I'm channeling my inner Steve Balmer running around and realizing that this in fact is the the the inflection point that changes what environments people work in how do you see that affecting your designs for the future and where do you see any changes or any particular large scale protocol changes with developer consideration being brought in in the near future um I think you've raised a very good question like the community is you know obviously our initial focus is on the existing ethereum community and uh but going forward the hope is obviously to make uiux so easy when it comes to on boarding that we can onboard more devs in in in in in who are previously working on other use cases and um one uh particular bottleneck that we're we're getting feedback on this is probably to do with the stage of our development is uh is the bridging experience and then the risk associated with that that is something that we're spending quite a lot of kind of resource and attention um kind of working on and then the other um the other aspect is account abstraction um I think just statistically when we look look at new users on boarding uh a lot of the users uh something around so so so one one one interview I did last week with a with a wallet U Builder is that something like 30% of new users are lost at the um at at the P code stage so at a point when someone's expected to kind of write down their um cryptographic key uh passwords uh that's when they lose interest and move on to doing something else and um so unless the draw of the crypto EOS system is so large um this is a major bottleneck that we need to change and um and and that the the the the the the the current kind of infr layer solution to that is obviously account abstraction it could be built in with the wallet and it works very well with um zkn projects such as scroll and polygon and um and I think there's a whole ton of um adjacent research along that direction that needs to be done before we're at a stage where um kind of non-financial use cases can really be talked about and I think um just just back kind of backing up a little bit the the the whole effort that went into building a ZK evm with fraud proofs is to provide very very high level of security and then the tradeoff is that the scalability will have to um so the kind of the gas optimization circuit optimization along as Prov optimization that will be done gradually over the next you know three to five years and um and it's going to kind of decrease in in a kind of steady state but initially I think our goal is to make sure the onboarding experience is incredibly smooth and that's why we're we're Bas basically working on a on a g fork and so that you know we're using the most tested execution environment for existing Builders um but I think um account abstraction privacy there's a whole range of different tools that are like zkm adjacent that needs to be resolved and and and that requires a a a kind of an industrywide Collective effort and and there are just a bunch of problems that cannot be solved by scroll team like independently so to solve the the cross-chain bridging problem that requires kind of bridges it requires kind of maybe working with the polygon team and working with other l2s and and collaborate on on on forming a solution but the end result of that is hopefully that we can bring on a lot more users and and you know the whole saying that you know Rising tide lives all boat kind of scenario so that's an area that we're where we're spending a lot of effort and focus on and I think when it comes to um uh protocol like protocol design at at at at the starting point everyone will look very similar but depending on where the feedback and where the interest and users come from um it is quite possible that the the infra like on the ZK EVMS will start to diverge as time go on so I'm I'm curious to see whether int years time the scroll zkm and polygon zkm will be will still be a very similar kind of product from a from a Dev perspective um yeah I think you know these are just open questions that we're kind of thinking about yeah I agree a lot with Sunday I think uh well at polygon we have a a clear Mission that's U you know on boarding uh everyone possible to web three so uh massive adoption is our uh goal uh we have a lot of development of the community also in Korea we have we are discussing with Partners every time so we receive a lot of feedback from them so all these efforts we are doing in the technical side are basically to on board more the apps and to have better ux so this is what we need to solve initially all the experience uh for for the D apps and users to web3 because the scalability is about this and then we have many other challenges of course the the beautiful thing about EVMS um in the format of CK is that we can extend the the evm of ethereum we can do more and as s was saying we there's privacy features or there's other ideas we can extend to different languages we can we can just figure out how we can attract more developers to this kind of U blockchain scale level networks um there's a lot of research also in the new CK formats n CK tooling this is also something we need to adapt and we need to uh continue including as we go uh but for now there's this um cycle of maturity there's different in my perspective there's different waves of CK research and Innovation that are coming that are amazing and where polygon also will try to be involved and we we are doing that but there are different cycles of maturity and our mission now is to apply to our available protocols the best the state- ofth art that can be applicable practical and can be provide the providing the best experience to developers and users so I think for now uh let's be realistic and try to solve the problem the original problem because uh I think my perspective is not yet done so but we need to keep an eye on this Evolution absolutely I I think the the underpinning theme of both your answers is the time scale and the perspective there is very very important this is um as Sandy said before a 10-year joury as as as David said a lot of Milestones a lot of questions really unanswered and unknown you can only do what is in front of you and work um in the uh I guess directionally the area which you think is correct and onto David's last point and hopefully also drawing on the the same sort of spectrum idea that I I started the last question with you're talking about the extension of the the evm we've been talking a lot about the future and I said before a decade and I think I think multi-decade future for some of these changes um and how those changes be brought about I I won't recap but from the last talk there was fantastic answers from Lisa and Charles around ways to upgrade the protocol many of these are exciting unknowns to be spoken about but one thing that's um I guess uh underpinning something prior to evm is is ethereum itself and in that I think there lies an awful lot of values and it exists is a decentralized community it's a permissionless open environment where everybody gets to come and everybody not just being the people in this room not just being the people who who solve the research Problems But ultimately the millions the the billions however people want to to quantify and we should we should ultimately strive for everybody but in ensuring that you Foster and keep all of ethereum's values of openness of decentralization of security of low barriers to entry all the positivities that that that drove its creation and evolution and um hopefully ultimately design design choices and extension of evm what are some of the unique milestones and unique decisions you see in front of yourselves as your as your protocols evolve to ensure that you keep those values and keep those values um I don't want to say enshrined but held within uh held within the the the Z KVM as it evolves um so scroll's Mission from the very beginning is um where we're an open source project we want to build with the community and uh and and you know we we we actually kind of very actively track open- source contribution to our public repo and uh and we devote a reasonable amount of resources while building the core product to engage more Builders and contribution I think that is a that is a difficult thing to do and uh the the the results are not immediately felt and part part of that uh part of that effort is to kind of direct builders um particularly kind of ZK Engineers to to use the public Source repo to build other things that are uh either related or adjacent or if you want want to run the the same thing that's that's great because it means you know everyone's stack gets more tested and uh gets more eyes on it and and that is a a part of the Scrolls mission in the long run and um another part of the mission is about being kind of Community First and I think that sometimes comes into that comes in as a trade-off from if efficiency and execution speed so um if you want to go fast go alone if you want to go far go together and you know these mantras kind of ring true when we make decisions all the time about how to um how to do certain setups and how to arrange things and and how to coordinate group learning even kind of internally but um you know but it's a sliding scale and it's never kind of quite on one end or the other um I think when it comes to these kind of tradeoff we always try to steer a little bit bit uh more towards the you know Community contribution um side of things uh but you know it's it's just something that that we're keeping you know top of mind while we're also trying to build a you know um state-of-the-art product and high value ad like you know proving system to to the market but um but I think you know the the the the process of um providing um very price efficient proofs um to an ethereum equivalent ecosystem thus require multiple players and uh contributors from different kind of verticals of the stack so uh One Vision that we've had from the beginning and and that's based on uh my co-founder yes research is that we believe Hardware will add another three to five x to proving efficiency and that's not something that we have the capacity to do internally and we made that decision very early on so you know we try to kind of share research and encourage third party participants to do that and similarly with account abstraction and and privacy I think these will make you know 10x if not more contributions in terms of the user onboarding experience uh and it's also unlocking you know institutional engagement or social use cases and and these things cannot be done and I don't think it should be done at the protocol level so even though we're we're very early um we we're very mindful of how to sufficiently engage um and um and act um as a Collective yeah in order to kind of deliver what the core focus is which is to deliver like the best developer experience possible yeah I think scroll did a a great team a great uh work the by this uh model of contributions from the community uh in in our case we were more aggressive in terms of uh deadlines as she said there's a trade-off here in terms of how much how many contributors are or coordination probably in and uh we were more focused on on shipping as soon as possible to prove that this was kind of practical and uh we had an internal team that's very very strong team and we were focusing on build build aggressively uh after this uh Pro was buil and the main net is there or before in the testet phase we were opening more more and more to the Auditors many of them coming from theum foundation and uh this Tech specialist uh and now we are just getting this effort of opening more to the community in terms of contributions third parties so we want to become more and more a neutral protocol this agregation layer I was talking about this is the vision it doesn't need to be fully polygon it's kind of a a neutral layer we want to to build to aggregate different CK rollups but uh the values uh we don't have much time now but this is a long discussion we want to respect all of them uh to become decentralized and as the pr plan we're discussing um the POS is kind of decentralized but we we launch a new token to do is taking on different chains on these multiple chain Visions so all of these we want to to include and respect because uh we are fully aligned with ethereum in the sense well those are two uh answers so enchanting we are well over time thank you very much feel free to callor any one of us if you have any more questions around the development of Z KVM please have a round of applause for our two panelists come Sam me down thank you all right perfect perfect um thank you very much