DAPPCON 2024: Scaling SecurelyThe Need To Prove Everything Everywhere All At Once- Georgios Gontikas

Transcript

Hi everyone so yeah the Topic of my talk Today is about scaling securely and showing how the need to everything Everywhere all at once Yeah let's start with the current state of block scalability landscape can 1.0 has focused Last cle of years both in Vertical scalability and horizontal scalability so horizontal is Bridges connecting different l1 Vertical is rollups layer tw layer threes um And yeah As we can see is basically scalability over everything so um The

vast majority of Bridges are secured by trusted third parties and the vast majority of rolls strong trust assp however the important part is that the p has grown till last year at some Point we have just a handful of rolls and the modular paradigm was not even a thing so was just ethereum a layer and today have aass growing of RS have atab layers roll service

frameworks Roll frameworks In General and then we can see that with tvl as a Proxy right has grown multiples Which shows the user Confidence gr security over seen the biggest Focus of the Industry so far Was You optimizing making everything scalable sacrificing essentially security and now I'm go through a story How is basically our Journey into building scalable Products in CR alliance diagnosis

ecosystem we decided to take the to pri security obviously in the beginning of sacrificing skillab And yeah We define a couple of values like a couple of Design principles How how we want to build our Products like What is going to guide our decisions design decisions when building Products and These are simple things security security depth first was Z clients we been

operating Z clients for interoperability like securing connection of Two ecosystems and an amazing te Trust of disadvantages that not many not the surface not many people have seen especially if not operating them actively so what we seen is First of all it's quite expensive to run multiple ones and since One of Our Kind of Design principles was Future proof and uh security in depth we wanted always to run multiple mones and that gets very expensive very fast um they

are slow obviously a lot slower than you know trusted Federation of validators There are still trust assumptions actually so Most of the implementation of the Z clients rely on uh the socalled s committee and not proving the full Consensus um and the most important part it's that they liveness concern so Although These Systems are kind of trustless There are specific or let say trust minimized There are specific centralized components that if they go down for whatever reason your whole system goes down so then we went and on and build a couple of Products uh to to solve this

initial issue so uh the the most important One being hashi which is using multiple sources of Truth um in a kind of multisig of Bridges way to secure uh to secure a connection between the blockchains uh then pand which is a very similar like produ BPL connec Dat avab layers cross safe which is using Super secure rails to secure Connections of smart accounts Across blockchains and fragola which is landing Focus Protocol kind of specific to facilitate the secure cross Chain landing of the stable

opa them again especially has that Super secure and proof but quite expensive right so we trying to find ways How do we improve Performance but without compromising security all What's the key hereof Don't think immediately zero zero are We can define proof something more General as I like to think of them as you know verifiable facts or As we WR here Portable verification of statements right and These Systems can help in a

couple of uh domains the horizontal scalability The Bridges That was our main Focus Vertical scalability obviously as an example Z rollups right are based on proofs but also Any Kind Of generic off Chain computation off Chain data like oracles or coprocessor is the a name that is being used Lately um so yeah we've seen that the Industry is actually going Towards The this Direction slowly so from the scalability 1.0 paradigm that we mentioned the beginning the scalability 2.0 which is um trust minimized right fault proof based optim

or zk proof based layer or multi proof based Bridges such as the hashi one we mention and also similar um trust Solutions for the data availability layers How do we go from 1.0 to 2.0 is this proof layer So conceptually it's a proof layer What does it consist of uh proving infrastructure Z kvm coprocessor and most importantly aggregation Which very powerful feature of proofs where you can basically maximize the cost reductions um however while working with this Systems we realize that this is not the end game

um yet There are couple of Challenges that um Yeah still need to be tackled mainly in three categories First of all uh proof Production How do you produce this proof uh this proofs um currently all the Systems are or Most of the Systems are super early stage involving rapidly and are very inefficient uh then the verification part like because proofs are produced of Chain um and then without verifying them on Chain they are basically useless you don't even need a blockchain So you need to verify them Somewhere um and then you come up since the Demand of this is

growing you you you come with the same problems of gas Spikes or capacity Limits right so we all know the capacity limitation of ethereum for example as a layer one for verification and last but not least the Architecture so the whole implementation of the system is by itself a big new attack Surface because it's Super Complex is very difficult to audit the teams that actually implementing them are very few and um Written in languages and similar stuff that is not like solidity for example that is common well used and known in the

blockchain ecosystem as an example Here is the the rise of the zvm right So there are three at least teams that have announced zvm last couple of months and made announcements I believe C of between each other one was two times faster than the previous One then another one came with an announcement couple of weeks later It's X times cheaper 20 times faster so and then we as as users of the system as Developers we we Faced with a couple of questions and concerns right like How can I um basically use them Which one

should I Choose can I basically switch from one to the other [Musik] Systems so yeah I mean the the main conclusion here Is that we we realized trying to build with system that there still kind of a Missing puz Missing Piece in this Puzzle um to achieve like Yeah I mean I didn't put the pie like completely top right illusion so just need to find bance and [Musik] infrastructure to achieve that um there

basically three key ingredients key points that we we need First of all is developer experience first right So this this is Focus for Developers so we need sdks that can support multiple proving backs and allow Developers to develop security in depth Systems uh then we need some sort of an engine Which will mainly aggregate uh all this kind of PR to get all these cost benefits we mentioned um and index them right and most importantly we want to provide kind of really good developer experience in

the form of a database like interface right So you You don't need as a developer to deal of any of this you know Low Level proof stuff you just need to query for specific facts verifiable verified facts that you need and get the answer for that um and a key part of this that it need to be fly Chain agnostic right um live in a multi Chain World here um Yeah this is how conceptually this would look like um so on the left Side I believe for you is also left Yeah is we have the off

Chain proving Universe Which includes Most of the stuff We just mentioned So like all these zk VMS that around or also mina or Neil foundation you know Or Technologies that are Let's say not that new like te that already kind of more Battle tested aggregation zk aggregation protocols coprocess even fully homomorphic description anything that is basically off Chain verifiable of Chain computation um then you have the unchange settlement universe uh that is Yeah you can like verification layer such as ethereum but Here we see the

need that this should be a set of Chain So it needs to be Chain agnostic need to be available Everywhere And yeah in the box In The Green Box is the applications the protocols that would use That on a high level Who would use that right um It would be applications that need some sort of access to some sort of Cross Chain State um such as cross Chain interoperability right or in the more General case um offloading a very expensive computation of Chain but having verifiable results on Chain and another use is privacy You can run code That is not public visible Chain you

have a proof you verifi it and use the Results the outputs on Chain Yeah here some more specific use cases to make this more Concrete what this cross change State could be right so Execution proofs from layer to right that something has happened onic lerab like proof of consens this is what Most of the zk Light clients are bas right now or storage or event proof like something like a fact verified fact that something specifically has happen on a

specific Chain you just need this result then the processors like any ofed computation another layer avs reusing ethereum security to secure your own applications protocols and These come in sort of commitments that can be also provable like verified facts fit so [Musik]

end Centric Future is proving everything So like Any Kind of Proof Type Any Kind Of system that needs to to offload off Chain computation Everywhere like Chain agnosticism needs to be available on any Chain you should not Pi specific Chains and all at once like aggregating everything and basically minimizing the cost and maximizing the efficiency of your system and Yeah that's it Thank you [Applåder]