New Ethereum talks, every Monday. The week's conference uploads by event, in your inbox.

Loading player…

The Superpowers of the ZK Stack - TOKEN2049 Singapore 2023

TOKEN2049Mon, Oct 9, 2023, 12:00 AM

Transcript

hello everyone um I'm gonna talk today about the superpowers of the ZK stack uh the ZK stack is the free modular framework for building hyper chains rescade powered zkdm power chains which you created at the matter Labs uh but I'm gonna talk about the super powers that a properly designed architecture based on zero knowledge proofs can offer you in general it's not specific to the zika Stacked by us but it it's it can be more broadly used but you have to design it from the beginning in the right way so um ZK stack is based on the source code open source 3D open source source code of the ziki sync era blockchain which is the permissionless zkvm-based L2 on ethereum and it's currently the most widely used the most actively Used Auto on ethereum so where above um arbitrum optimism like we're actually doing more transactions combined per month than all the optimistic roll ups together uh we have uh very significant fraction of the daily active users like I think 46 uh right now from from all the l2s we're third by tvl so it's a very properly bottle tested framework um and it's based on the philosophy of the uh the the easy kissing project which would have been very consistent for about for the last five years since the project was created uh we recently published a Manifesto called The zika Credo where we outlined the motivation and the very specific properties that we want the blockchains to preserve as we progress towards the the world where blockchains are basically in everything like it's the the extension of the internet with the value layer um so we we talk about very specific things that translate into very specific design choices made at Disney stack and if I had to use only one word to describe the superpowers combined of what zero knowledge proofs can give you and this presentation is going to be fairly technical but I really want you to get an intuition on why it matters the One World will be Limitless the ZK technology with ZK stack in particular is making ethereum Limitless um the first very important property that you get from the ZK is the Limitless verifiability so blockchains such as Bitcoin ethereum are valuable because they are trustless which in practice means that they are verifiable from the ground up you can take a look at the source code you can understand the math and cryptography behind those cryptographic functions and properties then you can run the full node of the the source code that you checked yourself and you can be sure that the state of the blockchain is valid and like every single piece of logic that was there from the smart contracts to the signatures to balance updates was done correctly you don't have to trust anyone you don't put your trust in authorities institutions uh other validators you you run it completely for yourself this is the property that is very easy to lose as we have to scale because ethereum today is only doing 20 transactions per second and um like it's a direct consequence of the broad decentralization of ethereum with hundreds of thousands of full node Valley like hundreds of thousands of validators plus probably millions of full nodes that verify the network today and if all of them have to re-execute every single transaction that happens for all the you know world's value interactions that would be impossible to scale so we have different approaches such as optimistic Roll-Ups uh side chains whatever in all those cases you give up some of the trustlessness you rely on some other people and some validators and some entities organizations to run the validation for you and you just say okay I'm gonna accept that and that does not scale with trust but it does with zero knowledge Pros what we can do is your knowledge proofs is we can use math basically like zero knowledge proofs is probably a misnomer we should rather have called them proofs of computational Integrity we can run arbitrary large computation we can do it in parallel on maybe on many different chains with many transactions millions of users transacting daily then we can generate the proof of small blocks on each of these chains then we can combine them together through recursive zero knowledge proofs produce the proofs of these proofs then produce the proofs of that proofs and so on until we get one final proof which we put on something like ethereum which is a broadly decentralized blockchain and then maybe we generate the proof of the execution of the ethereum Foo like block so you don't have to run the full node at all you only run a light client on your mobile device which attests to the Integrity of this entirety of these millions of transactions that happen you don't really care zero knowledge proves that succin's your knowledge proves snarks Starks is the only technology we have that that can enable that and from this superpower follow a couple of really interesting things so the next thing I'm going to talk about is the Limitless data because um on ethereum if we are speaking about growing like building the foundation of the the world's internet of value we have to be able to scale like the internet and that means you should be able to add more users more transactions like there should be no inherent cap like imagine if if the internet had only 1 000 servers that you can use to to run the the transactions like sure we could optimize data load we could optimize the algorithms to fit more and more um uh your throughput onto this thousand servers but that would put an inherent cap like you would not be able to onboard like billions of users you would not be able to run like video streaming and so on and this is what we have on ethereum today uh we have like it's a it's the most broadly decentralized smart contract platform and it has this roll-up Center crowd map with layer tools permissionlessly innovating bringing in new ways to scale it but all of them are limited by the single um bottleneck which is called Data availability right now on ethereum you have something like on the order of 150 kilobytes per second data throughput that all the roll ups together compete for and utilize and like if one roll up uses half of it you only have the half lap for all the other ones which is nothing like it translates to maybe a few hundred transactions per second across all of these roll ups and if you have 10 major ones you got only like 20 per roll up it's like it's incompatible with bringing billions of people to to blockchains to get into the mainstream adoption the eap4844 which is coming soon to ethereum mainnet is only gonna alleviate this slightly even though it's only reserved for Roll-Ups uh it's not going to provide a significant boost so we're still going to be in in in these limits and even if we Implement some more advanced forms of sharding on ethereum within the couple of years that's coming again we're gonna be limited you know like the the peer-to-peer network of ethereum cannot process infinitely many transactions but you can totally do it with your knowledge Pros so one way you can do it is instead of publishing the transaction inputs like all of the actual like call data of the transactions you could publish the outputs and this enables you to compress the data and use it much more efficiently uh similarly to how like one one truck and carry one container but the train can carry like the flexible stretchable amount of containers with transaction outputs and we call them the state Delta or the state divs uh you can do let's say 1 000 Oracle transactions and then since they all write in the same data storage slot you will only publish one single update at the end of this batch or or big block and so it compresses super linearly like you can do hundreds thousands of transactions of certain type and you only pay fixed cost so amortized per transaction is going to be very very cheap and you can do them a lot more frequently and this enables things such as um fast oracles like Oracle dates every second and all the subsequent Arbitrage transactions following the circle updates can be also very efficient making the entire defy protocols very efficient then uh it enables practical privacy because privacy is very data consuming in order to run private transactions like anything that touches privacy again zero knowledge proves is the only technology we have today and so you have to publish the zero knowledge proofs on your blockchain and verify them this verification is could be pretty efficient on on all roll ups but if you have to publish the data for uh the the inputs of these transactions you quickly going to consume all of the ethereum block data uh because the the zero knowledge proofs are large they are at least one kilobyte large with the state Delta approach with which ZK sinks ZK stack uh Embraces you don't have to publish that you only verify the proof you discard the input data and you you publish the results making transactions that previous preserve privacy essentially as cheap as normal transactions like blurring the line and finally you can do things like um Native account abstraction which ZK Stacks is also supporting also as cheap as normal transactions because account abstraction requires you to publish some significant non-trivial amounts of data for biometric signatures that maybe you're using your mobile phone to to sign it and not not you know like crypto native wallets or specific pre and post conditions that provide more security for your transaction so all of that requires a lot of data and it's going to be cheaper on the ZK stack but that's not what gives you the Limitless ability to process more data the the the the real benefit you get from the zero knowledge Pros is that you can extend the data availability from these robot of ethereum to beyond that um using external off-chain data availability techniques you can do like private chains with um Enterprise data availability committees you can use decentralized data availability systems like zke border and you can combine it seamlessly with the roll up with ethereum data availability building a system like uh what we call volition SDK Porter is is the zika stacks architecture for implementing volition and all of these accounts can interact seamlessly with each other all of them every single transaction is verified by ethereum full nodes and mathematically guaranteed to be correct by using zero knowledge proofs the rollup accounts use 100 secured by ethereum data availability so they they are as secure as ethereum giving that there are no box in the protocol implementation the extended data availability has slightly less security guarantees but you we still cannot mess with the data or with the validity of this data and this gives you basically Infinity many users you can just extend it with more and more data capacities as you need and the transactions are going to be very cheap and and they are fundamentally Limitless right there is no um cap on how many users and transactions you can process you probably see out of the box without changing the protocols you can stay in the ZK VM evm compatible Paradigm and uh have private Enterprises Banks um companies or or even Dallas around this privately secured chains out of the box very easily and this brings us to the third very important property that is your knowledge probes give you that is the ability to trust other systems machine to machine without relying on human and or like validating intermediaries because if we like if you have a chain that uses your knowledge Pros for validity and you have some other chain that uses exactly the same circuits exactly the same protocol then any message that comes from that chain can be verified trustlessly you can trust it as as much as you trust your own code and if both chains settle on ethereum or on on some other layer 1 you can have access like you have the shared finality in shared settlement so you can tap into the data or storage of the other chain and um this basically enables direct access to users and liquidity of that chain and here is why it matters the future is multi-chain I think we're we're like by far it's a broad consensus that we cannot run all of the world's value transactions on this single monolithic system it doesn't matter how powerful how how much throughput you have on the system could be something like Solana with I don't know thousands of TPS it could be something even more centralized with like really high throughput it's gonna be inherently limited and we need Limitless system for the internal value so with multiple chains some of them are going to be general purpose you can think of an analogy of the case blockchain as a city like Singapore Hong Kong New York would be General general purpose cities right you can live there you can have businesses you can go to work have some entertainment other blockchains are going to be application specific so there you can think of some exchange like dydfs or or like uni swap specific chain or um something for a social media platform that requires its own chain they are more like you know like Disneyland like a special city which is dedicated to a certain purpose where you don't have residents you only have visitors maybe some of these visitors come there for a long period of time so you fly to Hong Kong to Disneyland you spend an entire day there or a couple of days and then you fly back right so like since you spend a lot of time there you can't afford this expensive travel cost there and back but there are other application specific use cases like a shopping mall which would be you know like you would probably not fly to a shopping mall you needed to be much closer and much like a closer proximity to the place where your user residence reside and we actually have this properties with with different blockchains we have we have different degrees of proximity so think of Roll-Ups on ethereum we can always Bridge from one roll up to another through layer one fully trustlessly um it's gonna cost you a little bit more than a roll-up transaction because you have to go through layer one but at least you don't have to like trust other validators trust some bridges and we we know that bridges are prone to a lot of hugs we've seen like billions of dollars being hugged because of the basically essentially the multi-signature of those bridges you don't rely on cryptography you rely on some people and people can be corrupt and you don't want that this is this defies the purpose of why we will build blockchains but then you have other blockchain systems that are outside of ethereum that are even even further from from us because now like there is no way to transact to layer one you have to only go through those bridges they bring risks they bring Capital efficiencies and they bring more delays so what if we could combine like connect rollups together even more seamlessly where you can call you can send a message from any address on the roll up to any other address just like you send an email from any like I can send it from Alex matterlabs.com to your name gmail.com you don't care what the domains are you don't care about the cost the cost is always like off one just one click and you send it and it's cryptographically authenticated so you can check the signature the the encryption and you know that you're talking to the right person and we actually can build that with their knowledge pros this architecture is called the hyper Bridges and this is the key to me it's the the most important property of the ZK stack the ability to connect to to get plugged into a network of this hyper genes where you have basically direct native seamless access to all of the other hyper genes sending liquidity in just a few seconds with the finality of their knowledge proves achieving now like basically sub at some point we'll get to sub-second finality not not yet now it takes a few minutes but it's probably okay but it requires zero Capital it requires zero trust because you only trust math uh irrespective of who runs all those other chains and it is Limitless it can grow as you need as your demand for from the user's goals and uh yeah let's go Limitless together um I hope you now have more appreciation for the ZK technology and check out the easy key stack for more thank you [Applause] foreign

Automatic transcript — names and jargon may be misspelled.