DAPPCON 2024: Why ZK Is The Endgame - Alex Gluchowski

Transcript

hello everyone exciting to be here in Berlin I'm going to talk about ZK uh and I want to give you some like very high level intuition of why Z is the the ultimate endgame and then I I want to make it a bit more interactive and happy to dive deeper into technical questions but this is starting conceptually uh this year ethereum is going to be 9 years old very soon and if you remember how the web looked like when it was 9 years old it was unsafe very inconvenient slow you remember might might remember the the modem sounds and all of that is pretty familiar to how the blockchains web 3 world looks like today um and all of that gradually improved in in the early days of the web

with different inventions but there was one transformative thing there was one very specific specific idea and and technology that made web widespread that that enabled us to move our lives our lives online and uh to explain why we can remember the internet cannot run on a single server it's a collection of many systems many networks integrated together owned and operated by different entities um who you cannot possibly know so you can be in one place interacting with someone on the other part of the world maybe using servers in a different part of the world and to interoperate

the systems require a cryptographic protocol for ensuring the Integrity of communication and this came with https before https you could try to access your bank from an internet cafe or try to put your credit card online and buy something but that that was a dangerous undertaking and like a lot of people did not trust it for right reasons like there was a lot of fraud with credit cards like if you cannot know who is in the middle of your communication can they can be intercepting or messing with the uh communication it it's not really safe to like it you can't move your lives online uh and so like hgps enable

this Integrity of communication you could all of a sudden have like a point to point between any two points on This Global Network in all of the world talking to each other seamlessly trans like fully protected and the web three today is similarly in many regards unsafe it's very inconvenient and those two hang together if something is inconvenient people are making uh tradeoffs and and like it's also pretty slow we have very low throughput even with all rollups combined on ethereum even with with the blockchains the tradeoff decentralization for Speed we're nowhere near the demands that the internet of

value will put on the world where everything could be onchain and but all of that is improving however just like the internet cannot run on a single server the internet of value cannot run on a single monolithic system no matter how powerful so you it must be a collection of different systems for various reasons for technical in the first place just because it's physically impossible to do all of the computation required for blockchains in a single threaded environment or even multi environment with lots of interdependencies but also

for political reasons and and social reasons the web 3 is characterized by ownership and like different people will own different parts and like there will be a large collection of very decentralized systems um very heterogeneous systems and in the perfect multi-chain world we would be able to transact just like on the internet having a user on any of the chains any of the this networks um sending transactions and and and like value and messages back and forth from their chain to any other chain to any address on any of the of the other chains just like you can send an email from your address at your domain to any other address any

other domain and it's end to end encrypted uh and what stops us from having it today in the today's multichain is the lack of trust free interoperability so all of the bridging protocols today don't really work like Bridges when you think of a bridge it's something that connects two pieces of land um and you can use it to take like a truck load it with gold and like whatever Goods move it on the other side of the bridge this asset disappears on one side on one Island and appears on the other and you you transport it over this is not how the bridges work today the way they work is you park your asset on one

side of the bridge someone holds custody of it and then someone gives you some assets on the other side so maybe they hold the piles of of Supply on both sides or maybe they they issue a virtual asset which represents what's being parked on on the one side but this is not really a bridge like a true bridge will be trust free like it would not require any trust assumptions and it would not require any additional capital capital overhead you like you would be able to pay a flat fee irrespective of the amount of funds you're moving this is not what's happening today all the bridges like if you think of the bridges between the l1s or Bridges with between l2s on ethereum like bridges that facilitate faster withdrawals from optim

optic cups you have to pay a very significant percentage Because the actual bridging is not happening you have someone doing it for you and someone holding the the capital there to to facilitate this and they have the capital costs but uh yeah so like this approach has compromised either scalability if you have to verify everything that's happening of all of the systems or they they compromise trust and then you have compound Corruption of trust from one system to the other and you canot really scale you can build the internet but if we had one transformative

invention something like zero knowledge proofs we could have a cryptographic protocol that guarantees the Integrity of computation so if https provided the Integrity of communication the Zer knowledge proofs provide the Integrity of computation and we actually have it now like the what's happening is that uh the um the these Protocols are getting more and more mature from year to year and they enable us to build this world where many chains can directly transact to each other so uh to give you a technical overview like to dive a little deeper how it actually works um we have to consider

recursion like you are probably all uh familiar with the the concept we start with a single transaction we can generate a zero knowledge proof of validity of this transaction it's a small piece of data just maybe 1 kilobyte in size that guarantees that the transaction was done correctly and then we can put a lot of these proofs together and recursively combine them into a proof of proofs and then we combine the proofs uh together until we get a proof of a single block that can contain arbitrary many transactions can be thousands or even millions of transactions depending on how efficiently you can paralyze the block production and then this is a block of one single blockchain so you can take

many blocks from potentially many different blockchains and you repeat the process you do the the recursive uh aggregation recursive composition of these proofs until you get one single proof of all these chains and you put it on a single settlement layer and today the largest settlement layer the most decentralized layer in the world which can verifies your knowledge proofs is ethereum and it's I don't know if if it's if it's ever going to change um but then what you can do is you can generate the proof of ethereum itself and this is getting feasible we will be able to do it in a very near term to generate the proof of the like full block basically

the entire layer one execution and then you what you can do is you can verify this proof on any smart smartphone in the world in less than one second and this will turn every single smartphone in the world into a full node of ethereum then so basically you don't trust anyone you don't trust ethereum validators you don't trust any other participants everyone in the world can verify the entirety of ethereum transactions from Genesis and all of the millions of transactions that happened on thousands of the ZK chains that recursively um produce proofs and settle on ethereum and this is what we would call a ZK

Singularity and this is happening very very very soon like this is I I believe that we will see it actually happening in 20 20124 and uh it's now time to build an ZK and I'm happy to answer questions thank you [Applause] hi okay perfect so do we have time we have some time for some questions we have a bit of time so I'll take a question down here hi thank you for the talk we're

here hi um so I just wanted to ask uh if we even can prove very efficiently isn't it the case that data availability and bandwidth becomes a problem like if I want to prove if I want to check all these proofs in my smartphone I still need to have the state and the state dips would this become bottleneck then that that's a fantastic question so first of all Zer knowledge proofs enable verifiability of all transactions uh and to have like fully trustless ownership of of of the assets yes the you need to solve data availability uh ethereum is going in the direction of data availability sampling so in some observable time you will be able to also verify at least Le part of ethereum full blocks from your um or

like not not the blocks the blobs that are used by ethereum l2s because everything is going to be in L2 from your phone so you will download just a piece of data and everyone will download some some random piece of the block so that together many users can reconstruct the block but that will apply only to the data availability that is handled through ethereum Network which I believe will be enough enough to cut the world demand for um store of value for for for like everyone in the world let let assume we have 10 billion people everyone will be able to have an account on ethereum which is not going to be super cheap you know like the transaction can cost like a few dollars

maybe for for this transaction maybe like up to $10 doesn't matter because you're going to use it only for the uh uh like a lot of assets that that you hold all the daily transactions are likely going to happen on validium and um validium can be fully paralyzed each of these chains can decide where to store where to hold data availability and also each of these chains will be a volution most likely so part of the some of the accounts on the Chain will be posted to ethereum network and verify it like randomly from the phones the other part most of the data will likely be posted to some other systems like this is for example what uh Lance is doing now with uh the uh lens Network built

with the ZK stack on on ZK sync they're using they're going for down the the validum route and uh um this makes perfect sense for a social network where you own the data you own your accounts the social graph is fully on chain every post isn't chain but it's not as critical if like all of this data becomes unavailable which could be a tail risk of a validium system thank you and we have another question here I'm uh wondering it's a thanks for this great presentation and it's a really impressive things about ZK Singularity however I wondering how you're going to solve the blockchain

trma problem currently it looks like that node operators use the maximum possible CPU power to generate blocks and it limits us to increase the block size or produce use more blocks and so on to compute this zero knowledge proof it requires much more computation power and in this case we should really pay something maybe we can produce blocks with larger delay or maybe we can reduce the size of the block to do this extra computation because currently it looks like we are on the edge maybe I'm not so specific but how you are going to do it on fly not to really make the etherum

working slow over in terms of generated the ZK Singularity uh that's a great question so what we need to like let's start with the um like reminding that the demand for computation is already immense in the world like everything you do on chain oh sorry everything you do online do Google search it requires computation you do like some data processing everything like video processing like and it it all happens in Cloud on demand so you always paying like when whenever you need some computation you do there like you do a query on chat GPT it actually costs something like 10 cents to generate this query and it's being done in the cloud just like elastically scaling and shrinking uh the same thing will happen

with the uh cryptographic demand for cryptographic computation if everything every asset in the world is moving on chain which I believe is happening in in in a very short like on on a like on the time scale of a decade everything like payments Trading stocks tokenization of all assets everything will be on chain like that that's inevitable that that's the way the world is going uh there will be huge demand for that so what we we like what the important parameter to look at is what is the amortised cost of zero knowledge proof generation per transaction and if we look at the costs today on ethereum al2s it's quite low it's like under 1 cent already today uh and we know that the there are a couple

of new um breakthroughs coming both on the proof system side that that will give us probably a 10x Improvement as well as on the hardware side which can give us another 10 to 100x Improvement so we're going from 1 cent to 0.1 0 1 00001 cent which is probably going to be more than enough for financial or quasi Financial transactions uh maybe not yet for for like I don't know like some super tiny micro payments but that that will happen down the line so what's important is like it's a very small cost per transaction and the proofs can be generated on any gpus and in the future we'll probably have the hardware accelerators right so uh we have enough

supply of gpus in the world to handle that with all the surge in demand for AI based compute the the supply of computation providers will only keep increasing and the the the ZK proving will massively contribute to that like vitalic has had a had a post I think two days ago like reinforcing this idea like in in 10 years from now everything is going to be ZK every like every single L2 on ethereum will have to either Embrace ZK and upgrade to it in in one way or another or or we just become irrelevant and I just think it's it's relatively hard for existing systems to just Embrace something like most likely they will have to build new versions

from scratch embracing some ZK T and then ask the users to migrate rather than trying to upgrade things in place but eventually everything isk all the Assets in the world are on chain and everything is being proven in the cloud okay so we have time for a few more questions and I see you up there um hi a quick question here um in the world you here up here in the world you describe what's the where is CK snc actually um playing the role so maybe this last sentence you just mentioned oh that's that's an amazing question so like what what I believe if if you ask me about like how the the ecosystem will look like to to give my prediction I think that uh we will have just a few

ecosystems forming that can implement this Vision like this vision is not possible to like to build on top of existing heterogeneous Chains It's not possible to say like oh we're going to build a layer that connects all of the existing rollups or like existing external chains that is not how it's like that won't work because you won't be able to implement the this seamless bridging between the the different systems so what will happen instead there will be ecosystems that understand this vision and that will be working towards it and they will be like and the zyn is one of them and I think we will be the first one to actually make this

reality with a do like several dozens of chains already building um ziki chains in this Zing Network like chain of chains the network of networks um the upgrade that is happening like the next upgrade the version 24 which is happening on on on on chain on on ziki syn will enable the shared bridge and will enable those chains to build directly on this shared Bridge so like all of the assets will be still in single contract on ethereum and all of the movements of the assets will happen basically just in the Merle Merle trees you will not need to transfer anything on ethereum the reason you cannot

connect external chains with a separate layer is because you have a native asset such as eth you might have other native assets on those chains and they are currently residing in smart contracts on L1 and you cannot just say oh the asset is on this smart contract on L1 on optimism but somehow I'm going to use it on arbitrum and it's going to still be a native asset like you need to figure out a way to move it from this layer one contract lock there to to to this other one and that that that's just not going to work like you need to build it from scratch and the system will grow from from scratch so what I believe is ZK sync will lead the way and there will be maybe one or two more these Peak

ecosystems of chains that will then like basically compete in a good way and and try to improve the these Technologies and and this is what will comprise the ethereum ecosystem ethereum will be like a big network of networks of of the ZK chains okay yes we have ask sure I'm just curious you about you know three or four big players stock developing you know recursive improv technology polygon has their layer um very similar Technologies do you synergies do does the ecosystem need

three or four of these Technologies to dve so I I'll repeat the question the question is like there are multiple different players already working on similar ideas uh if I see them uh uh if like what's what's the question do they complement each other do they complement no like I think they they all all of them are working towards their own version of this universe like of of their own like ecosystems with interconnected chains I actually don't know that too many are working on this like there are only a few and what I think is happening like I I believe the you know like the back to to this discussion with vitalic two days ago on

Twitter the um a superpower of ethereum is that it's a permissionless network where teams can build an innovate freely without asking anyone for for uh consent or without asking for funding from this one specific organization from EF or from whoever and it leads to Innovation and and some like friendly competition between the teams like race to to be the first one to build something great this is amazing this makes ethereum extremely powerful and this is why the L2 Center crowd map worked and the original idea of sharding which was like centralized let's have like Central planning waterfall model like this this is how

it's going to work that did not work as well as as this like this decentralized world and and for the same reasons I believe that the the ecosystems will compete and the network effects of these ZK ecosystems are much greater than the network effects of a single L2 so the L2 list as like what we see now on L2 bid is not going to exist in this form anymore there will be like a few um like ecosystems with like very stable very mature code BAS base that is like heavily audited uh very well like crowd crowd Source secured Etc and that is actually trustless so the individual

teams launching l2s cannot screw users what's happening today is like you can launch an L2 not have fraud proes have like a 105 multisig uh controlling the upgrades and and call yourself an L2 on ethereum like that's nonsense really like this is not how it how it should be it should be the other way around like you have a few big established tax stacks and then anyone can indeed permission launch their own L2 but um the team you don't have to trust that particular team the trust should be solely with ethereum open source code and math and this is how it's uh the ethereum future is going to look like thank you thank

you so I have two small announcements to make um the first is if you're here for the block Scout talk it has been moved to 6:40 in the evening session and um the second announcement is that we'll be taking a short 10-minute break and we'll be back for our next talk which is the fir side chat um in just over 10 minutes at 3:25 thank you