Panel I The Next Step in Scaling Ethereum with Offchain Labs | Pragma New York 2023

foreign [Music] for today want to bring on somebody all of you have known for quite some time a lot of our New York natives here the co-founders of off-chain labs you may know about option Labs from the famous layer 2 called arbitrom we're going to talk about what it means to scale ethereum how that whole process has been over the last few years where we're going ahead and next in terms of the evolution of our infrastructure and everything in between from how everything started to what we're excited about right now so without further Ado please join me in welcoming Ed Harry and Stephen from off chain Labs team [Applause] oh pure silence I love this so uh I have a bunch of topics we want to cover I want to talk about sort of uh so the history of how everything started to uh kind of some things you've announced recently and uh I'm hoping this is uh more of a friendly chat between all of you as well because we want to understand all three of you kind of getting you on at the same time as uh uh I'm hoping we're gonna learn some new things so my first question is coming from Academia to where we are now how did you all meet how did this get started and what's the story sure uh yeah let me tell the origin story um so um well we met when I was a professor and Harry and Steven were PhD students at Princeton um and um you know we had a history there of doing blockchain related research um the original idea from for arbitrim uh started in 2014. um thinking about how to scale smart contracts before there were any smart contract blockchains and a bunch of undergraduates built a course project to build a sort of very early Proto version of arbitrim in the fall of 2014 and then after that I took this strange detour to work at the White House for two years came back to Princeton and after parachuting back into Academia and wondering what I was going to do with myself one day these two PhD students walked into my office and said hey remember that Arboretum thing let's work on that again and that was the Genesis of the it looks like we're uh what happened with the mic here we'll just check yeah okay so um these two walked into my office and said let's do arbitrum again and that was the Genesis of the arbitrim academic project well let me pass the Baton to um Harry or Stephen to continue this story okay mine is live again we'll uh just for the idea can we turn on all the mics just make sure is that better there we go okay good um yeah so we published uh the academic paper for arbitrim and fall in August of 2018 and that was at the music security conference and at that time we also looked at each other and said I think there's something really very important here and we want to build it ourselves so the way it sometimes works in Academia is you do something and someone else builds it but we're like and we all worked on many different academic projects in our days so we're like no this is the one that we actually want to go ahead and build and that's when we actually started the company that month in August of 2018 so we just turned five happy birthday and um we were building and building for for for uh you know the past five years and even before that and uh it was in August of 2021 so two years ago just over two years ago that the arbitrim network actually launched to the public and that was a culmination of years of research and building but and some of the actually the initial uh designs that went way back to Ed's 2014 days at Princeton and you can look on YouTube and find that original class video actually from fall semester of 2014 and the reason I like to mention that is because it actually every day um you know super cool so arbitrim has been uh definitely we've been thinking about it for a long time and in the academic literature actually it was um built as a very theoretical system like hey you can build this on top of some other blockchain but from the earliest days of the company even before the company founded it was clear that we were building this ethereum and that's where we put our Focus ever since and in the past few years since launch we've had a few different iterations so we launched arbitro Nitro a year ago and that um you know massive the increase the capacity and Lee is the lead uh brain behind that and he's here today somewhere so go find him and just recently we announced arbitrary stylus which is the ability now to not only launch evm uh contracts on the Chain but also to launch um Ross to CNC plus plus and Rachel is here as well you can find her and The Stylist team but anyway I'm sort of getting a little bit past the order we're gonna cover all those things I'm just very excited about those things so I I think there's a little bit more of kind of fun uh to map out kind of the history like slightly after that I think kind of it was I think early early 2019 I think kind of was when I think I I went full time and I was the first one to take the plunge into like okay this is this is it now uh and uh we had I think a good uh year and a half what a year year and a half but it was like just a like us we brought on like one or two other people tiny in-person office in in Princeton that was convenient to to get to from where we were we met co-working space then we had like our moved into like our own little dedicated office but like very tiny building away like in this kind of like hey we're all nerdy academics building something awesome that we love um and yeah yeah there's like we raised a seed and we had investors but like it was very much in this kind of like beautiful little like it wasn't a garage but it felt a little like it yeah um and uh eventually then we kind of like started to expand out a little more got up to like I think maybe like six or seven people we ended up deciding to to move into a wework in uh in New York and kind of had like a few months there because that was in January of 2020 or maybe like a month earlier um and then covet yeah and that was actually really uh covered was actually an important development for us as it turned out and the reason is that when kovid came so we were actually close to signing a lease in an office space in Manhattan and we were going to be a company with the centralized office in New York that was the plan but then covet happened and So that obviously wasn't going to happen fortunately we had not yet signed the lease promising to pay monthly monthly rent for for two years so we went virtual because of covid and then we started hiring people who were not in the New York area and pretty quickly we realized hey this is working a lot better this this fully distributed thing is working a lot better than we it we thought it would and there's just a huge advantage of being able to hire people from not only all over the us but all over the world um and so we took that plunge becoming a fully distributed company and that's been huge I think we would not be where we are today without all of that talent that we've been able to bring in from all over um that really you know the ability to go from a few smart and hard-working people in a room to an operation that can actually deploy something like arbitrum and build the ecosystem right that requires just a ton of great talent and we were able to uh recruit that in part because we went fully distributed and I don't know if we would have if not for covid yeah I don't think we would have the fund funds facts maybe not so fun for Harry Harry got the raw into that deal because he relocated to New York for that New York office that never was so he basically moved to New York maybe three days later we exactly where everyone's we're going remote or going virtual and Harry basically had to sit at the cramp Department in New York for a year oh fun fun story for me I don't know about that he's laughing so we're all good oh it's all right I thought you know oh I like to say it's fun it's it's nice that you know ethereum is global um I wanted a much louder grown I feel like that really deserved you know that's what we call it Heath Global um well so I guess I'm curious about how the last three years have been or I guess a little bit over two uh you kind of launch mainnet it's been Monumental to kind of see what's happened with the success of how widely the the chain is uh use the chain is I mean it's still the biggest tvl out there if you check on l2b uh what was that experience like what kind of went well what didn't go well what are you skeptical of or don't agree with on those certain metrics from from that uh laugh I could I could have guessed uh how would you kind of retrospect on the last couple years well so look yeah so we launched this thing in the August 31st uh two years ago and um you know it start this was after about three months excuse me of a closed um developer only mainnet the idea is we wanted all the all of the application teams to be able to get their stuff set up and running so that when we threw open the doors of the mall all the storefronts would be occupied um so we opened it and you know we got some healthy um we got some healthy use and I think about a week Plus in we had like uh 100 million dollars of tvl or something and Stephen went offline for 24 hours and um when he came back he's like um you know how are we doing what's the tvl and uh the answer was uh 1.6 billion so we had like a billion and a half dollars of tvl come onto the chain all of a sudden over a 24-hour period um completely unexpectedly yeah so it was right around this time of year center of the Jewish High holidays and I get sad I go I went offline for a 24-hour period and then I just thought there was a mistake I'm like I don't know what's going on here RB Nyon for those who remember that was then uh that event um wow but it was it was a very interesting kind of like hello into the space just in terms of like you spend these like months working with teams and trying and getting tons of teams deployed on mainnet in the in this kind of relatively open developer period we had but the the first the first app the first application the first was that kind of got this huge amount of involvement was a meme oh yeah right but I think there is kind of a lesson there and it's sort of it's sort of proved out over time too is if we had at the beginning said okay we're gonna pick which four or five projects we want to really kind of work with and drive the success of um when we launch we would have probably guessed wrong that is um you know we're really happy in hindsight that we went with this Fair launch approach that was designed to kind of bring everyone on board and then let the market decide or the ability of these teams to deliver decide uh which ones were going to be big and I think it's part of the reason that arbitrum has been as successful as it has that we took we took that sort of big tent approach to working with applications but I also want to say that it's not that's not just a a sort of marketing or strategy choice because in order to make that work we needed to do a ton of really hard engineering work to make it so that it was pretty easy for a team to switch from say ethereum to arbitrum and because that switching cost was low teams that wanted to come on board did not need a month of special engineering support from our team right so in order to make that sort of uh adoption at scale possible we needed to do a lot of really hard detail engineering work um and I think that's part another part of why we've been able to be a successful as we have just just add a little fun piece of color onto this story it's it's now something that's kind of like a interesting exciting story to retell but at the time it was terrifying um i i people who know me can vouch me I tend to be like a pretty I have pretty solid nerves um it's basically the only time in my life that I think I've had like anything that gets it all close to the pan to a panic attack was when this like novel piece of technology that we had all sorts of warnings about of like this is novel this is experimental you know use at your own risk and kind of try to like educate people on we were not expecting and not really hoping for that kind of adoption and so I think that like it was a very interesting and I think that kind of like it's we had confidence in what we had built but there was also a level to which like after a week I became more confident that if there was going to be a major issue found probably people would have found it because the incentive was so huge but it kind of took that for me to sort of calm down a little bit that moment for me when there was a single transaction that bridged 500 million dollars just think about that for a second that's a lot of money so uh yeah that's uh that's a definitely a high stress environment um well uh one of the interesting things that kind of for some people who've been in this space for for longer than three years one of the interesting things is that we've also kind of seen a narrative change uh four years ago l2s were just like a witty something up there that can get the TPS up and that's kind of the metric that mattered now it's a mix of not only do we have better ways to think about what types of Roll-Ups there could be or what type of L2 definitions you want to kind of contort to uh we also have now nuances in well maybe everybody should have their own and we can kind of do spin-offs we can do very specific things and kind of looking at things like I mean you have the op stack the polygon cdk uh you have kind of the ZK development kit I forget the acronym that they use uh how do you kind of view all those developments kind of how does that cross compare to what you're trying to do and sort of what are your kind of thoughts on all those things so we have Arbors morbid which is you know the offering that allows you to build your own chain uh you can build an L3 completely permissionlessly and open and you take the code do what you want with it there's a set of customizations that come out of the box like your own fee token the gas token but then you can actually go in and make deep customizations I'm probably like I'll give a slightly controversial take uh maybe which is I I don't necessarily know that it makes sense for every project to have their own chain there's a lot of interest there and I think we'll find the right balance between them the analogy that I like to give is to TV channels in which it's clearly that wouldn't work if every every show was on the same channel there's just too much too much there and you need a differentiation but also clearly wouldn't make any sense if every show decided no we're not a TV show we're a channel um you'd end up with like empty programming most of the day everyone would have to go and negotiate their own their own deals with advertisers and Distributors and you know the system that we have I know this is data like who watches TV anymore but it's just an analogy the system that we have is you know kind of a decent one which is you have organization by uh type and interest so if you're into cartoons you have a Cartoon Network if you like nature you'll have the National Geographic Channel there's there's no exact science where you can have a cartoon about nature which could sort of go either way and I think that's what we'll see with roll up so you also could have some competition you have a few different news channels for example and I think we'll see verticals so you'll want to be in an environment where you're Loosely organize with others that are you know want to be in the same chain with you that will have benefits for you and your users for this discovery like oh I'm on this chain playing I'm on over to Nova playing this game let me check out this other game that's on the same chain so I'm not that uh personally bullish on the idea that every app will have its own chain I don't think that makes so much sense to me but I do think we'll see many chains uh pop up and it's sort of Open Season out on who those winners will be it's hard to know and I think a lot of people are angling to sort of build those for new verticals but also we don't even know what the verticals are yet I think those will pop up over the next number of years too yeah so with orbit the um really the the good reason to have your own chain if you do is because you need something different or custom that the big that the big public chain like arbitrum won doesn't provide right and so it's kind of about customizability in those um uh on those chains because I think a me too chain that works just the same but uh has a switching cost and less TBL and smaller user base uh doesn't really make sense right and so we're really thinking about giving people the most flexibility to figure out how they're going to launch their chain how they're going to configure it what kind of data availability what kind of sequencing um and um and so on so I mean that's really sort of from the standpoint of how we think about what orbit is providing it's really providing the that ability to have your own um to have your own chain to customize it the way you like and if you do this as an L3 chain it's fully permissionless this is actually pretty important too because there is not it's not a goal to have a single overarching governance framework that's going to or a single rule or law or principle that everyone has to abide by um it's if it's your chain it's your chain and you should be able to choose how it operates and you should be able to decide how it's governed for yourself what are some of the other kind of toggles that you can customize in this chain definition oh yeah um so um a good example is which field token to use um that um okay yeah good um a good example is which fee token to use um the arbitrome one chain and Nova use eth and there's a lot of advantages to using eth but some trades want to use their own chain as a fee token some might want to have a different economic model where say they have very low gas price or even don't charge for gas at all but then have some kind of say subscription model or in-chain purchases or in-app purchases or other kind of economic models some might want to tune the chain differently so they care a lot about they care more about cost and maybe not so much about super fast finality there's a whole set of of different trade-offs that people might make based on the needs of their particular uh their particular business approach I I mean I I had kind of just like a fun diversion from this which is very related if you want to do more features first I can just briefly I guess two more that come to mind are permissions so you can like uh um scope permissioning differently you can say it's a chain that's secured by ethereum but only these set of people or only I can launch contracts on the Chain so you can control not only um like your environment who's actually building in your environment but it can be open to users you can also permission on users this is something we're seeing from institutions they want to maybe have a kyc chain where actually you can only interact with the chain and again the cool thing here is you can have these permissions but it's still a public chain in the sense that it's uh secured by ethereum but actually that's another uh when it comes to public that's another area which you can innovate on is public versus private can you build something that's secured by ethereum but also has built-in privacy and the answer is yes these are some now now I got my my fun my fun little diversion here which is that again back down memory lane which is that the uh about around I guess around four years ago now um we were working on this originally like I was describing we were not imagining building a roll-up and we were not imagining building a uh a world with kind of like where there'd be like one big or a couple big arbitrum chains we were imagining something that looks a lot more like arbitrary Nova based on Arboretum any trust today with smaller groups of participants and like for example I think our go-to example at the time was a chain that's just one poker game that's created for the poker game and you play it and then the chain dissolves and much more kind of ephemeral chains and sort of at the time and this is like one of the things you get maybe when you have a bunch of academics this was not exactly something that had product Market fit oh yeah and so it's not an idea that yeah yeah but it's also the case that at that time that we didn't hadn't yet figured out what we understand today about how to build a chain that is ethereum compatible as an L2 right when we started out back in started the company back in 2018 I think nobody knew how to build an L2 that was performant and practical and ethereum compatible and we had to figure that out along the way and so one part I think we knew early on that something that had that generality and compatibility was something we wanted but early on because we didn't know how to build that we were focusing on something that ultimately is less useful but this may be coming around again because the ability to launch say an ephemeral chain to manage a particular poker game or like a duel in an online game still probably has some some some use eventually as long as it's not the main the main and only thing you can do another fun fact poker game might have seemed like an example Harry just threw out of there but you'd be surprised back in that Princeton office the amount of the number of hours we spend researching poker um well particular on um uh trustless poker yes yes it's actually Reliance of course for science yes non-trivial uh Hannah design actually for science no yeah there's a lot of academic literature on poker and um we used to be reading poker papers a lot um as I throw the example of what you can do on an Arbiter chain and uh just shows you know sort of how things have developed over time because back then there was actually there was a time where it wasn't clear that the adoption would be as quick as um sort of you know all in everyone doing it at the same time as it was and there were thoughts we didn't need to actually not only build the Corey stack we need to build some good examples showing people what you can do when there was a stop there would be this one-off environment like this so it wasn't like d5's moving wholesale it was like hey you can do a poker game on Layer Two and we were going to start build some of those early applications for our chain turns out the way history went we didn't need to do that unfortunately and others found the technology and were excited about it even before we had launched it so but we definitely spent a lot of time thinking about those early applications I bet those were the good days of State channels being the the top of uh where everything used to be yeah yeah I I think most people already have a good idea for optimistic Roll-Ups are but I want to just make sure we're we're not kind of assuming that so I'd like to kind of have you give like a 20-second version of optimistic Roll-Ups in the next question but but I wanted to pick that in before you answer the actual question which is kind of giving how optimistic Roll-Ups work um understanding how something gets finalized is important and decentralization and measuring that is equally important and I'd love to kind of understand from your perspective for arbitrum how do you think about sequencing and the guarantees I need to have in the space yeah um so this there's often misconceptions about this and there's a bunch of reasons for that one of them is that the way arbitrim does this is different from how other optimistic systems do um in designing arbitrim we made a pretty hard line that the sequencer should be minimally trusted that is the thing that the sequencer needs to do is to vouch for and record which transactions arrived in which order and nothing else so the sequencer is not trusted for any other purpose and that's actually really important so to give you an example the sequencer does not decide how much fee someone is going to pay right there are some other systems in which the sequencer just essentially takes from your account whatever it thinks is fair we think that design doesn't make sense and we don't think it is the right design for users from a trust standpoint right so for us it was always a requirement that the fees you pay are defined by the parameters of the transaction that you signed and so the sequencer doesn't set the fees it doesn't collect the fees it doesn't control what the fee amount is for any for any transaction also the sequencer is not a proposer again in some other designs the sequencer is a block proposer but it's not here because we don't want to have that that function centralized so sequencer does one and only one job what's the definition of a proposer just for the audience yeah so proposer basically so the sequencing is the function of saying which transactions have arrived at the chain the proposer is the one who who says which blocks get built as a result of that transaction right now one way to build an optimistic system is to say we have a single centralized proposer that announces blocks and then those blocks can be rejected by you know by the optimistic um fraud proof system if you have one um to um if it's not right but the problem with the centralized proposer is if the proposer is not proposing correct blocks then you can't make progress so in arbitrim there's a proposing is part of what validators do and the validator set is a broad set of currently 10 plus parties so um anyway so in the design we took a lot of pains to make the sequencer be minimally trusted for that purpose and to push all of this sort of truck to make all of the uh the other other important properties be part of what the fraud proof system is dealing with because the fraud proof system can prove the correct outcome of of what happened so that's an important piece of it that we did early on so that's kind of the role of the sequencer right and currently there is a centralized sequencer that is run by off-chain labs and of course it's under the control of the arbitrum Dao who runs that sequence answer and we started a recently announced a research collaboration with espresso systems to build a distributed version uh sequencing system and so that over time will become available as an option for any arbitrim chain why is it important to have to that be distributed well as I said before the sequencer is currently trusted for transaction ordering the sequencer can censor a transaction for a limited time not forever but you know you might worry that the sequencer will will front run or reorder transactions in an exploitive way um and currently that is based on trust right the Dow chooses one sequencer and then has to trust that sequencer to not do these extractive um uh these extractive actions but when you go to decentralized sequencing distributed sequencing now there's a committee of sequencers also chosen by the Dow um but you only have to trust that some majority of those is honest and just to Echo what I'd said I think that there is like this misconception out there when people say decentralize the sequencer sometimes people think that the sequencer can at will post arbitrary transactions or take money from your account in some systems that actually is the case that's what a centralized sequencer candidate that gets to what Ed said about being a proposer and you you know if the sequencer if there are no fraud proofs and are no validators where the sequencer says basically is Law and no one can challenge that but because there's a live validation and proving system the sequencer is minimally trusted so the sequencer can't take money from your account if it tries to two you know the validators will stop that they you know they can't do that the sequencer can't post a bad transaction that will be included in the chain there's a system of already a distributed validation system that will protect that that's not I'm not trying to minimize the importance of decentralized the sequence server but you know that's I think what ad was getting I'm just focusing it's about this problem of reordering or delaying transactions it can't infinitely censor because you can always post a transaction on ethereum but that has its own cost and delay what it can do is it can reorder transactions and it can delay transactions and today there's a centralized guarantee which is um we are not doing that today the arbitrary sequencer is processing things in a first come first serve matter currently um and it's enforcing that policy in a centralized guarantee in a centralized way but the question is can you decentralize The Ordering of this of the sequencer such that you have a decentralized or distributed property that's guaranteed that hey if this consensus property does the right thing if the majority of these users of these participants do the right thing then the sequencer not only isn't but cannot do this maliciously ordering and that sort of again not to minimize the problem but just to focus on what the problem is it's not quite this large problem that others think it is again in the case of arbitrary specifically that's not the way the sequencer is designed I want to just circle back around and finish that out with kind of like what is the 20 seconds of like kind of how this works at a high level just because then hopefully this will help some of what Stephen and Ed said click into place if you're not already kind of familiar which is basically you're using metamask that's a good place to start you submit a transaction that transaction with arbitrum goes to an entity called the sequencer the sequencer then very quickly confirms the order that it's sequencing those transactions in I mean that's where the name comes from it's sequencing your transactions um at that point it says hey this is the order um if it's malicious it could change things it's just a promise that it's making it's not enforced by ethereum that's not kind of the real ethereum security of a roll-up what happens next is periodically on arbitrum one it'll be every couple minutes depending on how much traffic it posts a batch of those transactions to ethereum that batch just contains the signed transactions so it's just the same transactions that would look be in an ethereum block get posted directly onto ethereum that's just the inputs that's just kind of specifying what your transactions are the same way like I said would be in a block after that's posted the order of transactions set so anybody looking at the chain knows what the result will be you read those transactions you execute them you know exactly what the result is anybody off chain at that point knows what it is ethereum has no clue so you can't withdraw your funds at that point because how is ethereum supposed to know what the result is of executing those transactions the whole the whole thing is we're not forcing ethereum to execute them that's how we're making things cheaper at that point a validator comes in or a proposer and says hey if you executed those transactions this is what the result would be and it just makes a claim about that and at that point again the clean you can check off chain very easily if that claim is true it's trivial to check it's just that again not ethereum we can't and this is where the the way the challenge period comes in where any other validator can see if that's incorrect and if they it is if it is incorrect they could initiate a fraud Challenge and there's an interactive mechanism for ethereum to actually resolve that challenge and figure out who's correct and who isn't correct and make sure that ethereum actually has has proven knowledge of what this is which is how you can take funds on arbitrum and withdraw them back to ethereum securely I don't know maybe he also quickly worth just understanding then what is arbitrine bold and and kind of what's the Nuance here yeah so it puts bold is the um Next Generation Um challenge protocol or fraud proof protocol that is um that is currently in the in the pipeline we've published the code for bold as along with the paper explaining the algorithm there's been a security audit of the code which is is published and it's on the pipeline to uh to get to uh being Deployable so what it this is the part of the protocol where um validators are able to excuse me are able to um are able to work together to establish what is that to checkpoint to ethereum what is the correct outcome of execution so a way to think about this is a validator can make a claim about what the correct outcome is and stake on that claim and if nobody makes a conflicting claim after about a week the uh that first claim will be confirmed and it will be canonical and official and um and recorded on ethereum if somebody else makes a conflicting claim now you can think of the think of it like a fork of the chain so there's a fork and we need to have a fork Choice rule to decide which one of those branches is the correct one so unlike ethereum where it's a consensus protocol and basically it's kind of voting among the ethereum validators to decide which branch to take with an optimistic roll-up like arbitrum with fraud proofs the way this works is that it is actually a kind of a dispute on the merits about who's right so one party one Staker or one party who's correct is able to force their version of the the of the the chain to be confirmed and all the other ones to be pruned off um and the way that and so the protocol for doing that that is the thing that allows validators to um hook to a stake on claims about the uh about the correct outcome and then any disagreements about them to be resolved that's the challenge protocol so bold is a new version of The Challenge protocol which has a bunch of really important technical properties which I'm not going to go into right now but the upshot of it is that bald makes it safe to make valid the validator role be permissionless so yeah so let me talk about what that means um So currently the gear it's currently there's a set of around 10 um uh around uh 10 um validators that's a permission set and includes a bunch of large and well-known entities in the space and all you need is for one of them any one valid validator can force the correct outcome to be uh to be accepted even if all of the others are are are trying to to lie and cheat so when you make validation permissionless now we have what I think is probably the best possible security guarantee for the chain is that you'll get the right result if any one party is honest if any one validator is honest and you can be a validator that means you yourself with your laptop in your you know in your little apartment if that's how you live can force the entire arbitrum chain to get the correct result if you're hearing in 2020 you can you can be a validator and still make sure nothing goes wrong because you just need one honest just need one honest validator right and it's something when it's permissionless anyone can be a validator yeah so that's the key thing that bold makes possible that's awesome it is uh so this is uh just a research proposal or does that get rolled out what's the there's well there is code for it um it will be um and because it is one of the core security functions of the chain and because because it's replacing a fully functioning and battle tested fraud proof protocol we're going to be cautious about and we think the Dow should be cautious about when to roll it out and so there'll be a process right it's been through security audit the code is published the paper is published and there will be a test net with that uses bold and then presumably after that bold will get the uh with the assuming the Dow decides to the um a primary arbitrum test net will be upgraded to do bold and then again assuming the Dow decides to it would then arbitrum 1 and Nova would then get upgraded to bold so that's a process um and the next step of it is uh is a a public bold test net but if you want to see the code want to understand the protocol by reading a paper I want to read the security audit by trail of bits that's all available on on our website this is one of those cool things and like I think when Ed was describing this there is the kind of the Clause of like assuming the Dow approves it and the kind of it points at this really cool thing about where we're at now which is like Dows are are big and messy and complicated which is great um because these are our big technical changes that are critically important and you don't want it to be easy to to deploy changes like that I mean this is kind of the beauty of where where ethereum development has really thrived where we're kind of you have a lot of voices and a lot of thoughts and you kind of need to need to form consensus over over kind of wide swaths of uh of stakeholders in order to really make changes happen which is just kind of critical to avoid sort of you know capture and to maintain decentralization and so I think that kind of the the neat thing about sort of where where we're at working on stuff like this with with off chain is sort of we're we're building the tech and we're building it in kind of the best way we know how and and I think kind of the the cool thing is that orbit chains for example can can make their own choices can even before the arbitrom down like if they choose to adopt uh something like bold for arbitrum one anybody else launching an arbitrum chain can just make their own choice um and and sort of seeing how all of that evolves from just kind of like implementing this Tech and then how it actually sort of like and there's into the world and gets used is is pretty interesting and I think it's also worth like just clarifying when we say assuming the Dow approves that's not a formality like you know um and Harry and I like have the keys and we'll listen to what the Dow says that will do it like that's not at all how it works that's not the case to be clear yes I don't have the key to the chain the three of us together don't have the Keys No One in our company does the only way we can upgrade the chain is is we can we can't actually we can propose software to the Dow as can by the way anyone here go ahead and build software and propose to the Dow and the Dow can vote and hopefully will vote on actually executing that that change and it's self-executing what that means is the downloads vote is not a signal to some multi-sig and that's how it's actually some Dows work where there's a non-chain vote and there's a multi-sig that's supposed to listen to the vote that's not the case here but Dao actually votes on what the software is and that vote actually executes that change change so it's not like us we don't have any capability to do that we you know no matter what you do we can't change that the only entity that can change that is the Dao and it's really cool because that means that anyone who um you know is part of the down part of the community can participate in that process either in the discussion on the forums or the voting and really can build a future and if there are developers here again you can go to the Dao and and build software changes yourself and present things and you're in no more better position than we are at Option Labs we just have a lot of expertise in building the software I want to I want to touch on one other technical thing before we move on to uh kind of one of my last a couple questions but um bear bear with me on the silver simplification but option labs and arbitrung was not really started as a way to scale ethereum it was a way to scale smart contracts so so from kind of my perspective like you've always sort of talked about and innovated on how you can actually have better execution and better run times for smart contracts so I'm kind of curious about like what are kind of your different views here on how people should think about the evolution of smart contracts and nothing or The Stylist comes in into play here too uh what's kind of thinking behind that why do you think it's important or what that unlocks or kind of keeps the same but you have some better guarantees what's kind of the the way to frame all of that from your perspective I think the way to frame all that is yeah as you said arbitrim as a project is a Content existed before ethereum but it was very clear to us early on as it is today that ethereum is where it's at for smart contracts from a technical uh standpoint decentralization security but also from from a community from the room here that we have today all the people building innovating that's happening on ethereum but I think we also have to step back and realize that the world of ethereum today is relative totally small if you look at the numbers out there they're about roughly 25 000 again estimated to be solidity developers uh three million rust developers 12 million cc plus plus developers and I believe in ethereum it's Global right I believe that ethereum should be globally I don't think that that should require us to say okay all these millions of developers need to learn all the tooling that we know how today we have something very very special here but the rest of the world is catching on and we should bend our systems so that they can easily develop on those and that's when you know I think evm is critically important I think we have a massively important Community around the evm and that's the people that's the code and that's not going anywhere so I don't I wouldn't recommend that Arboretum or any other chain get rid of evm support arbitrary has the best evm compatible support that's possible today um deep deep down you know uses the same gef node and that's here to stay but we should also say hey is there a way we can do more right evm shouldn't be the ceiling it's not the ultimate goal resource the floor we need evm but hey are there ways we can make a forward insist other developers and today you have other chains that have done things like you know have Salon um Solana has rust support for example and I know they've done a great job in onboarding Rust developers but the question that we asked is can we in one environment have have solidity and rust and C and C plus plus working alongside each other so you have the Best in Class evm experience but you can also bring in C libraries if you want your application or others can and you can interact with them in one chain and you don't even have to engage with like it's not like oh you're in the rust zone I'm in the solidity zone no that's one zone I don't even know what language your your content was written in but I can talk to it and that's what arbitrary Stylus is and again it's about inclusivity it's about we have something great with evm we're going to keep that about how we do evm plus we call it can we actually expand this to expand the reach to more developers to make it even easier building on ethereum and and the answer is firmly yes we can yeah so uh let me just put in a plug for trying Stylus is one of those I keep telling everyone this but I think it's true it's one of those tools that you can fully appreciate until you try it um because um the the team developing stylist did an incredible job I'm led by Rachel who's here in the room um and um basically it's it's one of those tools that when you when you want to do something you think I hope it works this way and um stylist is one of those tools where everything pretty much worked the way you hope it does um and sometimes it does the thing that you thought probably wasn't possible but you just wished for It's actually kind of an incredible development that's rare just to kind of zoom out slightly on this just because I think Stephen and Ed already did a great job talking about stylus like one of the one of the cool things that l2s get us for ethereum is basically kind of this open space to innovate that that ethereum has like is is slowing down and it needs to slow down the rate at which it's breaking things because fundamentally like you need a strong layer that that everything else can build on top of and l2s and like in particular I think I mean I think Arboretum one has made similar decisions to kind of innovate relatively carefully um and I think Stylus is is kind of very careful Innovation that that makes sure to not kind of disrupt anything that exists just to add but roll ups also provide this kind of open slate of you can do anything um I think we just saw like uh it was yesterday before uh now it's like a roll up on top of ethereum that uses the Solana virtual machine that like all of this and especially with arbitrary of orbit change the ability to kind of just arbitrarily innovate and have like chains that are secure by by ethereum consensus but can do all of this experimentation and playing around in the execution layer is just such a cool ability and the really cool thing is the you know the successful experiments there can actually be potentially brought back to ethereum so I think this would be a long road to get there but I'm actually you know pretty hopeful that um maybe ethereum would consider um adding sort of style of support eventually as well and to those of you that sounds like a foreign concept if you're here for long enough there used to be a project called ewas and there was a lot of interest around this yeah it you know it didn't actually leave for different reasons it was early uh wasms developed a lot in the years and maybe now is the right time and arbitrary is excited to lead the way but we're also excited you know an offshine lapse to contribute back to ethereum that we build prism as well and uh we're excited to contribute in many ways to ethereum so my hope is this is super successful in ethereum and our goal isn't to make to lock it into the Arboretum ecosystem is to make it accessible to whoever wants it um no thank you for the this is actually extremely helpful because one of the things that we've seen just kind of from my personal perspective the last five years is getting people to learn a brand new language is not easy because their Primitives are not exactly the same they're sort of equivalent but they're not exactly mappable but more more importantly having a rich ecosystem matters being able to use dependencies that you rely on being able to get support on those things you just plug into things that we know have worked is a massive unlock and this type of thing just enables a lot more of that so getting those 12 million more developers in is uh it's a pretty big win yeah and and um there's another way in which that large developer community in these languages really helps and that is that the software development tools tools for testing and debugging and and all of that are much more mature in um for these more popular languages and if you want to ask like why why is it that Stylus is 10x or more more efficient for computation than evm there's a bunch of reasons but one of them is that it gets to use say compilers if you're using the rust compiler the rust compiler is better at optimizing your your code than the solidity compiler which is what you'd expect given that rust has so many more users and therefore so much more effort and Engineering has been put into optimizing all of that and so you know it's not just that we bring in all these developers it's that we bring in these much more mature and battle tested and functional tools absolutely um is there something that you disagree with each other on I wore a polo shirt Harry a T-shirt and add a button down so I'm gonna coordinate that I thought in a different day it might have been different although interview will probably be in the T-shirt every day yes I I'm the only one who's not human apparently judging from my profile picture I think the thing that I'm most impressed with is uh I don't know this was coordinated you just ended up sitting in the same spots as your photos that was something else there we go not my first panel yeah um a lot of a lot of success in this field is about attention to detail I absolutely agree I still want an answer though well yeah let me say that 10 cents a detail yeah no no let me say this um I think a really important part of being successful and having a good experience as a startup founder is who your co-founders are and the ability to work together and the ability to complement each other and sort of uh take advantages of the strengths of each of you um and I think we've done a pretty good job in hindsight not a perfect job uh we're always learning but I think we've done a pretty good job of that of taking advantage of the things that each of us brings to the table and we certainly have had some vigorous internal debates um I wouldn't call them fights but I would call them vigorous debates about what we should do and we end up I think making better decisions because of it I'll give you a fun one um there we go it was so um Cornell uh has this like um blockchain um boot camp every year and I think it was 2018 and [Music] um anyway I went and someone like was really actually was Nikola Greco from philocoin was like a pushing man like hey there's this thing called ZK roll up you guys need to call this thing a roll up call this thing a roll up and like we had like a vigorous internal debate or like you know Larry Davis style I don't think so over time but over time we came around to that but um yeah it's kind of funny because like the ecosystem is sort of the people for most of us here yes um well yeah so that was an interesting one because it wasn't clear what a roll-up what roll-up was going to be and in fact the thing that people call the roll up today is not really what that term means something different today than it did back then um yeah and one other just on this point is sort of tangential but we were raising funding for the company for the company initially in like 2018. we got into this like the following conversation all the time so are you uh a state Channel or are you a plasma like uh neither and but no no no you're a state child neuroplasma and eventually I actually gave a talk uh where I said arbitrary of advanced State channels where you sort of just like reframed everything was day challenge because no one else plus exactly there you go um so it's interesting actually the way um it doesn't really answer your question it's like so I'm squirting here I'm satisfied but okay good I'll stop no but the vocabulary develops in in a very interesting way in this space and the other interesting point here is people get very caught in sort of their their memes and like things you're supposed to believe one of them which I disagree with currently is ZK is the end game and anyone who wants to I don't have time here because you don't have a few hours if anyone who wants to engage with me in that I'm happy to um but if you look back actually just a few years ago and I mentioned this when you raise money plasma was the end game and the many of the same smart people and they legitimately are very smart people um saying that then or saying the ZK thing now and and you know there's vigorous debate on this point like this and you know to make our Point here like you should have your opinion and if you want my opinion I'll be happy to give it to you I guess I already have but the point is actually um Builders out there it's it's uh you know it's it's super interesting because don't like get caught by the memes and like say like or don't be locked in by the walls at other people's people have built because if you do we'd never have rollups because we'd all still be building plasma because we were told that that is the end game you know um there are a lot of smart people out there but smart people make mistakes the smart people's thinking evolves so if you have a really good idea and you're convinced it's a good idea keep building it and keep trying and uh the narrative will come around to you if it's a good idea yeah so um just more on this sort of ZK is the end game thing um watch the off chain Labs a medium account for a post written by me that will be going up uh probably within the next 24 hours on exactly that top you heard it here first it might be up already actually I was just gonna add on one of them I think what in terms of like in terms of arguments one of the fun things I think is like all of us kind of coming from an academic environment is just like you're very used to like arguing in front of whiteboards uh it's definitely a very important part of Academia and so kind of like before we started getting into sort of like the business world I think we already had a good amount of experience with like butting heads in a productive way um yeah beating yeah one of the first things we bought actually as a company was a ginormous whiteboard massive got to get it through a window on the second floor because it wouldn't fit in the elevator actually now of course we're fully distributed so we only get to use whiteboards together occasionally but that same whiteboard actually I learned not that long ago still exists in a storage unit somewhere this is true well Stephen Harriet thank you so much for joining us and sharing your thoughts this has been excellent and uh looking forward to that blog post thank you [Applause]