Too Many L2s: A Case for Enshrined Monoliths (AI Agents! Memecoins!) | Arthur Breitman - Tezos

hey everyone welcome back to the build Squad stage I'm Jane and hope you all are enjoying your time in Denver very excited to announce our next spe speaker we have the co-founder of tasos his name is Arthur and he's going to talk about too many l2s so Arthur please join us on the stage handing this off to you thank you all right so yeah the title of this talk is too many l2s again for in Shrine monolith and AI agents and mem coin is is bait sorry uh so yeah it's going to be a more of a talk about uh how to uh uh how to properly scale a uh an ecosystem with uh with rollups so let's get started all right so ethereum has too many l2s uh you know I think it but I'm not the only one saying it I think a lot of people look at it this way and uh you know it's great when you have you know it's great that people are building l2s but if you have a proliferation of l2s what do you get out of it uh you get fragmentation so that means scatter liquidity because all of them are trying to spend millions and millions to bring liquidity to their own chain so um you also get limited interrupt because they can't really talk to each other that well you can Bridge assets but you can't really have Smart contract calls from one to another um and that's essentially is the little to no composability that you get um out of it so from a user perspective this is not great right the user does not want to use 20 different l2s um it's not a good user experience at all and there are solutions being proposed but I would say that all the solutions are essentially Band-Aids uh so so they can interact but natively and by that I mean they can move assets from one to another I think that's uh one of the big advantage of an optimistic rollup or rollup in general is that you essentially have a trustless bridge to the base chain by default um so they can do that uh however uh if in order to do it quickly they essentially need Bridges uh which adds a lot of fraction as those bridges are always custodial on some level you know it's like every single Bridge protocol out there saying that they're non-custodial scratch a little bit it's like yeah it's custodial like you know oh we're noncustodial you pick the addresses that control the bridge all right so you know someone's a custodian uh there's really no good trustus bridge that exists out there there are some bridges which use light client level security um there's the exception under the rule because they're kind of hard to maintain and even some of those are going to have special keys that let them be upgraded so you have a form of custom in all of those um and then you know they'll slap a token and slap some governance to kind of like claim that it's uh noncustodial but it is custodial all right so we don't really have a good way of uh and that and that's only for moving assets by the way right you know it's you're not even talking about doing uh doing calls here we're merely uh merely moving your assets from one chain to another and so ZK uh can be better so the nice thing with a validity rollup as opposed to an optimistic rollup that you don't have to wait long time to move your assets from one chain to another uh you can potentially do that very quickly uh so you know they'll give you a trustless and fast interrupt um rollup transfer so that's good maybe if you have some shared sequencers you can get a little bit you can squeeze a little bit of compossibility um out of it but uh they also add a lot of complexity to your implementation and they add a massive overhead so you have a you can look at it as a tril uh these are popular so what do you want in a rollup you want a rollup to be versatile or compatible in the sense that you want to be able to run smart contracts you want to be able to run compile codes you want to be able to run all of these comple complicated things uh and not necessarily just you know a very narrow set of instructions you want a reasonable cost uh and I'm here talking not not just about gas but like the actual like cost of running the computers making the proof and you want High to put and you can get uh basically two out of three uh you could serly limit the type of operations you allow and then maybe you'll get high at reasonable cost um you can have versatility and high throughput and massively scale in you know your your your proving setup in parallel but the the cost will be immense per transaction or U you can uh let go of high you'll be versatile have reasonable cost but your proofs are going to take forever so it's not a uh it's still not quite there and there are fundamental reasons why the cost of ZK proofs even if you have better circuits even if you have better techniques even if you use ASX there's still going to be a bit you know at least a few orders of magnitude overhead in compute cost compared to just running code so the underlying problem is that the fraction also comes from a lack of alignment fundamentally uh it can be a lack of alignment in terms of the culture like what is a vision what is a mission what is the EOS of this um in terms of um you know the technical viewpoints protocol developments governance research tooling none of that is aligned they all kind of like working in their own Direction and the economics right the individual projects have their own agenda they have their own tokens um and you know the whole ecosystem is not necessarily align around that so what I'd like to do is make the case for un Shrine monolith so what is that so first you know the monolith part is monolithic there is no fragmentation in an non-ri monolith you have full composability and the idea behind it is like you know when when people think monolith in general they don't really think rollup right they think of rollup as a way to scale uh horizontally you're going to have many rollup isn't you know isn't having many rollups the whole point and we just need to solve how they're going to interrupt with each other what people fail to see is that rollups can scale vertically rollups are a great way to scale vertically and the reason for that is the trust assumption for a rollup and for a main chain are different for a main chain everyone has to validate every transaction but everyone is also the set of people guaranteeing the Integrity of your chain now if you want your chain to be sensitive resistant to have all these goodp properties you need to have many many validators you want your validators to be very decentralized you want the cost of being a validator to be very cheap that's all good however uh you now you need to distribute all of that data uh all of that computation to all of these validators and that's very expensive with a rollup with an optimistic rollup you have a single honest party assumption which is great right you know it's it's a lot easier to convince yourself that you're going to have one honest person rather than convince yourself that you're going to have a majority of honest person I'm much more confortable with just needing one than a whole majority and so um you can have very very different options you don't need as many validators as you would uh for your L one and add turns let you scale vertically with without the centralization aspect that would come from trying to scale vertically a layer one directly so there's a lot of benefits that come from scaling via a rollup you want to scale your L1 well don't scale Z L1 directly make Z L1 minimum Mak the L1 focus on what it's really good at which is consensus and then scale vertically within a roll up and then there's the ensr parts so what does an enshrine mean um it means first that like the functionality of the rollup is implemented at the level of the protocol directly it's not a smart contract roll up as a smart contract can be very hard you know smart contct languages uh today they're limited uh it's it's It's tricky to get the complexity of something like a whole rollup specifically optimistic rollup it's a little better with zero knowledge uh proofs because with a validity roll up all the complexities in your prover whereas the part that's on Shain can be simple but nonetheless um it's quite difficult and and and the the idea of being enshrine and and canonical in some sense it promotes alignment you're not just you know some project running on top of the chain you are just an extension of the chain and that makes a big difference culturally and economically so that's the idea behind tesos x uh tesos X is a high throughput rollup being developed on top of tesos and the idea is um that it become the canonical rollup for all activity um to take place on it doesn't mean that you can't use another rollup if you want want but by default um this is you know the idea is to move the activity there um the goal is to have very high performance interrupt meaning support for mainstream programming languages uh be a python or JavaScript essentially you want to be able to take whatever will run normally on the machine and run that directly um and that comes with support for you know lowlevel support for risk 5 so you know no no no feeding with a custom VM or feeding with a uh custom programming language this is not we're not talking about also you know something that looks like JavaScript something that we're talking about something that really is Javascript where you can import libraries to all of that um follow exactly the type of development environments you have for any us development and high composability between all of that so you know where is whether you you're building a contract in JavaScript and you want to call someone who's buildt a contract in Python you should have interrupt the modic design um the idea is that there's modules but it's also monolithic scale so you have consensus with a lightweight L1 and T is very lightweight in terms of um the cost of running it publication on a data avability layer uh which is the P2P Network secured by Z layer one I works with data avability sharding essentially and execution you have a canonical rollup which is a big and Shrine monolithic L2 so technical benefits uh is andrine it's noncustodial it's secured and governed by L1 you know it is true today that a lot a lot of rollups are actually custodial like they might have a lot of code in there that could almost allow them to be non-custodial but if there's someone controlling the keys to the rollup if there someone that can upgrade the rollup with a piece of keys then essentially those people can control the rollup those people can pre you know can hack the bridge those people can steal assets from the rollup it's custodial so if you look at the rollup today like Bas is is 100% custodial right you know it might mean you know it might be nice it might be doing some nice things but it is 100% custodial no fragmentation full composability so you're not talking about having many different rollups where maybe you have Bridges no um you everything talks to each other you can support different execution environment in parallel and also the stut is not a concern because with vertical scaling you can get a ton of stut out there if you think about it you know and I'm not even talking about trying to have parel execution forget parallel execution imagine that you have a huge State you keep your state all in Ram you paralyze signature verification now if you paralyze signature verification and your state is in Ram a transaction is just moving in value from one place to another like you should be able to do millions of transactions single threaded like I'm not talking about verifying a million signatures that you need to paralyze but a transaction that should not take that long right so I I think there's a lot to be done with like multiple cord but even on a single core you can go really really far um and gain and and retain a lot more composability uh but there's a there's a non-technical benefit there a security benefit that comes with it which is social consensus a canonical rup is not just a smart contract it's an extension of the protocol and what comes with that is that every protocol today will Fork to protect itself right if you have a bug tomorrow in Bitcoin Bitcoin will Fork it has happened by the way if you have a bug in ethereum ethereum will Fork like that is true but it will not happen for a hack it will not happen for a bug in a smart contract and that's you know it might seem like it's not a technical difference but it's a massive difference it it changes everything if you're saying like no no no no the protocol which is a ground source of Truth essentially you can use the whole social consensus of a protocol you can use a whole weak subjectivity of the protocol to protect the rollup when you look at the complexity of the implementation of rollup today B A optimistic rollup or ZK knowledge that or valid rollups that makes a huge difference it is extremely important and you do not get that when you Outsource the development of your rollup to a VCB company um that's just not going to happen so eink is the first step that we've taken on tesos uh it's tag L is fast fair and nearly free nearly free because transactions are extremely cheap um it's fair because uh there is a uh uh there's a um sequencer that's uh that's elected but can be replaced and we're working on code to uh to have uh to Blind the sequencer from the transaction not um not yet in place but upcoming and fast uh because we have subsec latency on confirming uh transaction it's evm compatible so drop in replacement for any uh ethereum or L2 that you might want to use subse latency minimal fees uh it's un Shrine it's completely governed by ZL1 it's an extension of tesos uh and yeah and there's a token on this link and a token is called TZ there's no reason to have a separate token for it um you know the Playbook today is like you launch an L2 maybe you sprinkle a little bit of Technology on it you know like what does a Playbook look like okay you launch an L2 you're going to keep 90% of the token for yourself you're not going to pretend it for yourself no no no no you're going to have a pie chart and you're going to divide the pie chart in 20 pieces it will be like well that piece is for liquidity provider that piece is for Community incentive that piece is for liquidity incentive that piece is for team that piece is for investors you know you got it but like no matter how you look at it about 90% is held you know by insiders uh and then you use that to create massive incentives to bring back tvl you'll bring some tvl for a while tvl will go out what else do you do you either say like you're going to be very high throughput so you have a little bit of like technology here and there or you try to add some yield in some form but eventually you know it's it's a bit all of the same and you know when uh those tokens end up being unlocked uh you're left with no tvl and nothing uh so no this is just an extension of tesos it can make for a less flashy launch but it's definitely more bust next steps uh so etherlink Plus data avability because we're launching data avability on uh on on tesos so the data viability charting which is if you're from the ethereum world you're probably familiar with d charting so very similar to that uh using zero knowledge um uh uh proofs around uh you know sharding your uh uh uh your data with error correcting codes uh so you can publish on a data vpt layer which gives you IM more throughput and execute J link but we have more coming and so this is all going to be part of tesos sex the idea is to like merge all this uh so first you know you fan out and then you fan in and as part of this fan out we have Justice or gstz which is a JavaScript compatible rollup uh real JavaScript you know you can use node environments you can use uh JavaScript libraries all of that is uh if you're web developer this is exactly the type of environment you're used to Tes link which is a melson rollup so this is for all of the builders who built over the years on tesos uh it's an environment that's familiar to them so every application on tesos can upgrade uh to Tas link and the idea is that Tas link will run aside uh from etherlink so you'll have instant transfers between the two so it's two environments but one rollup uh and more and so later all of that migrates into a canonical rollup so uh if I have you know if I I had to have an advice for the eum ecosystem uh you know was just my two sense I would say align on a canonical e ethereum rollup just like build one and I think this is something where the uh serum Foundation can probably send a strong signal you know even though it's not you know it doesn't have any special right it can be used as shelling point so just basically designate one and say this is a network extension of ethereum it's not you know some rollup that someone built that's a one uh and then Dr from the best of optimistic design don't wait for ZK I am so excited for validity rollups uh I I think I you know my my idea on scaling in 2017 I wrote a blog post about how you know sharding was not the right way to scale that you could scale with zero knowledge proofs uh that has become one of the dominant Paradigm out there uh since then and nevertheless You know despite you know this being my opinion at one point I say today if you really want to be able to run anything you want on a rollup don't wait for validity run with optimistic rollups especially if they're enshrined you have very very good guarantees with them because in practice you're not going to be running those reputation games and so and and and now I promised this was my my bait AI Ag and m coins so I'll I'll give my few sents on this uh let's see well first of all you know a mo modern AI models kind of suck at being atic like they really um they're very smart they know a lot of things they're now very very good with reasoning um agentic behavior is kind of like the weakest point at the moment uh and so yeah people are making a big deal out of using AI agents in crypto uh I I think it's mostly hype at the minimum if you're going to do it at least you know run it on some kind of uh on on some kind of like trusted environment so that at least it has like it's meaningful independent I think it's you know it's fun but like to see it as a new paradigm it's just basically like essentially it's trying to create hype around like well everyone can see this technology is important it's not really related to what we're doing because fundamentally these things are accounting systems uh but people are trying to make it so that say oh buy a token and you get exposure to this new technology that's like emerging uh but it's not really that so I'm not saying they're never useful but they're definitely way overhyped compared to like their actual connection to what it is that we're building uh and mem coins um yeah again I think that a lot of coins are MIM coins by definition um fundamentally like if you're a store of value you're a MIM coin uh but by that what I would say is like it comes from uh it's also a shing points people just describe to agree to acquire value to a thing sometimes it's sticky sometimes it's not obviously a lot of the m co are launching today they're not they don't intend to be sticky they intend to be mostly a roller coaster uh there's not much to it right it's a gambling application and gambling has always been one of the most popular use case in crypto and so I think it'll continue to be a popular use case but if you are thinking that they're about building communities or anything like that uh you are really fooling yourself so you know sure make them they're fun and people like to gamble uh but don't pretend that they're more than that all right uh that concludes my presentation thank you so much for coming uh and check out tesos X and also check out ether link bye-bye [Applause]