Unlocking New Possibilities with Stateless Architecture in Layer 2 - Leona Hioki

Transcript

hi everybody I'm Lea from in Max and so today I want to talk about statelessness uh Onan so yeah there are many words endless less right like pricess wiress uh borderless no it's something like free from something right so stateless is free from state but no what is state so basically State R is not about kind of no homeless or refugees it's like no it's more like a meaning of the state in the blockchain context and so in blockchain context

State what is a state so state is like you know condition or result of the database know this is a basically our state hard in the blockchain context right so what is stateless actually so if I if we deploy the stateless layer in serum it to be like like this no the stateful layer ofum is like isum layer one or the other uh no layer tools know they're quite solid state they have a quite solid statefulness and statelessness is more kind of light so there's a tradeoff

so stateless architecture and stateful architecture so stateless architecture has less functionality and more extreme scity and privacy at the same time this is their stateless and statefulness is more functionality and less scalability and less privacy now this is a b basic uh comparison of the these architecture so 99.9% of the block blockchain blockchain space are they stateful and quite few number of the asset system is stateless like inmax L Network and plasma and these kind of

stuff so inmax is like uh stateless plus ZK and ZK is used uh totally offchain and the very very limited data will be onchain committed and everything will be happen on offchain data preservation dkp computation dis these kind of stuff or yeah no all offchain and meaning that user side do a computation and this preservation and also it means that if there are 100 for example if there are million users million level paration and if they are billion users billion lb polarization this is a power of statelessness so if I talk about the

tradeoff I said that uh no it's functionality and scity and privacy but stateless trma is we can have more deeper perspective so okay stateless sounds totally fine right so it's even if it's a less functionality it's the extreme scalability and privacy and almost zero gas cost so we can keep that so what's the actual tradeoff so so many stateless or no system compromiser usability sometimes so for example lighning network is uh actually this is the most uh High highest dpsr no layer to are in this in this blockchain

industry right now but they have a significant problem are in uh their usability uh they have a channel capacity program where uh user need to fund a lot of fund fund a lot of money before they spend a small amount money and offline safety is like are also important because you don't want to run your node to pay money right so in light Network it happens so inmax is yeah almost there to be stateless it's a each data have five bytes uh only five byes of data consumption chain and you can include the unlimited number of transaction in that five for example uh

if you want to send a 100 transaction to 100 people uh it's safe five bites or without adding any data cost this is a significant uh result we can have in the research in the combination of stateless non zgp so yeah we are almost there but there's a TR know if we want to totally solve this H know we we need to sacrifice so we need to have a good tradeoff and I Tru believe that in Max 2 has a good tradeoff we sacrifice five ites but the other is perfect so I said so many things about statelessness and I I think that know if

I know show out show the statelessness you'll be fascinated by now if if I like know if I do like a dictatorship style like no I speak and people kind of no people in the state speak and know audience listen without doubt know it's a not good good conversation because know yeah it's trustless space and also people should doubt the new idea always and why so how do you doubt this system so I said that quite limited number of data is committed on on chain uh which is called Mark rout and everything like

a proof Distribution on the ZB computation regarding their uh that proof verification uh it's off chain what happen Happ s so if Brock producer distribut their mro proof to you guys and you can keep it and you can exit money uh to layer one as yeah no whenever you want so what happens so problem is like one the big problem in there no which is well discussed in plasma space in 2018 it's called B Block whle attack so if I distribute if the system is based on the premise that we I

distribute or block producer distributor no mro to you guys what if it it doesn't actually distribute right this is the first problem and second problem is how to prevent the double spend so let's focus on these points and first is that so if I don't distribute the proof of asset of yours and keep going their know blockchain uh let let blockchain keep going and people will be ignored of that our secureness uh in inse secureness we need to prevent that so this is the reason why we have a mutual

communication not oneway dictatorship power communication it's like block producer distribut proof on the client side Ling back on chain uh to to the B producer and this communication is compiled in zp circuit so they cannot Skip and at the same time how to spend divers spent how to prevent the divers spent though this is all about the five byes St consumption so to yeah I omit something but uh if I summarize that to preper double spend are no only five bytes is needed but that five bytes is needed know this is a no the reason why I had this kind of small luck of this statelessness in

maxu so proof update without without online requirement this is a quite important part so lightning network has a strong strong online requirement so if you don't run your note or or no online uh your fund can be compromised because know uh there are malicious activities or in the counterpart or in the channel or in Ling network but inmax or yeah ztp based system we don't have to care about that because uh everything uh it's proven in zkp of course and how to prevent double spend is like you

can you can uh prove that in zkp you can prove that uh after this block you didn't send the transaction right because the five bytes are data listes onchain uh consuming the onchain problem sorry on onch data so this is totally provable for everybody uh to to to uh prove that you didn't spend money so this is how to uh prevent ders spent so yeah this with this magic we could able our enable our people to know spend almost unlimited money unlimited number of money uh un liit unlimited number of transaction uh in constant fix five

bytes so one question there are many layer tools are in this space and I'm a big fan of the other layer tools no I was totally fascinated by no of course our arbitrum and optimism and no our you know the Z no they doing their no AI base or no lups and there are many functionality and I love that so how Okay so maybe you guys know many layer tools and if we have a 100 layer tools we have hundreds uh times scalability on etherum now this is the first question if we have a ZK sync stock

Weare drum and you know all of them are crazy smart built by crazy smart people but if we have a 100 H you do do we have a 100 times scalable serum actually answer is no because the block space is limited the most precious resource in this blockchain space is block space and block is there another type of block space uh which is no temporary shared by nodes no this is a block blob or EIP 4844 and this is a another type of block space but it's actually still block

space so people competing the block space of ser layer one uh from the beginning of their rollup and so State F uh layer to they are competing the same uh same block space and statelessness are it's we sacrifice functionalities know we cannot h no Unis swap V3 or V4 or no RV on on on stateless architecture but we can get away from this competition because the fixed size data will be on chain so if so the answer is if stateful layer tools or stateful blockchain answer is no and with status

blockchain we can have a 100 times scab ability if we have a 100 layer to with stat lenses this is a no biggest difference between the know the these two types of blockchains so okay and yeah we talk about the data no efficiency right so no I repeatedly said that no five bytes no fixed but yeah of course the biggest resource the most precious resource is block space data uh data know kind of consumption uh efficiency um but also their computation efficiency is important um and actually

inmax it's in inmax there users and client side do both their data preservation instead of noes or uh layer one and also computation on client side so how we could make their know verification extremely uh no efficient though this is the one of the another important part uh or inter bir so in interx basically there a communication between standard and recipient and they know convey the proof for each other and why the proof is right why is there given asset of no

given asset proof is right we can do like this so if you familiar with Mina maybe you are here know heard of this no it's a cyclic recursive ZB or suin zp so it's a change zkp uh sender make a zkp of their balance and the proves the correctness and the receiver receives that and the when he uh send money to another person uh when he becomes another sender he will no include that zkp in his no Circuit of chain and make another zkp and pass that to to another recipient and another recipient will do the same things for for another thing so

it's our end chain of DK uh off chain totally off chain no so and how to verify the end chain of zkp zkp is proving another zkp so this is a no end chain so pre if we prove the previous zkp you can prove the entire history of the zkp at the same time because PR zkp already proven the previous PR ZB and previous PR zp already proven the previous previous and it's like a know it's reaches the origin at the same time right so it's a mathematical architecture like once you verify the pr you can uh at the same

time you can uh instantly verify the all no history of zkp meaning that in this architecture it's meaning that and there holder history of utxo at the same time so this is a verification is extremely uh no efficient and con and the parallelizes are on each client side so that's the reason why if I said no that's the reason why I said if there million us a million level paration no data is on each client side or device or user and computation is efficient at the same time or it's paralized to each client side this is a no architecture yeah so yeah five byes

dater so basically are pro if you have a proof it's fine in this system so you can extend the proof right so the reason why one transaction have many many transaction at the same time and so we only commit the five by dater which is represent one transaction and beneath that uh tree you can know increase only limited number of transaction like no no Market extension so yeah this is a our no this is how we make a great scalability to accommodate all online citizens like a billions or something like that and

most significant part is like we can share their no it's a constant cost on chain so we can share their no uh gas cost with many people right so it's like so in state food system basically they have a know if many people come in their Network know more and more fluctuating gas Fe volatile gas fee and higher right but in stateless architecture surprisingly if many people come to this space we they can share their know the same bites and know they can know sprad their builds or know together so one and more people less and less uh gas we have no this is another good stuff and INX is a base R so definition

of the base R up is like no anyone can be a block producer instantly and no in Max Hazard that satisfies that requirements and so it should be more interoperable and no kind of holding the other many many great rer to slide know as you know and this is a arure of in Mar so it's like sounds sometimes it sounds like a lightning Network on Serum with with zkp and sometimes it's like a privacy layer and sometimes it's like a no inter operability layer and sometimes we can have a no conditional Smart Control payment so statelessness is quite mysterious and no when it comes to

another use case no it's shows her know another pH so that's the reason why I'm quite fascinated by this and so yeah and here's the academic no quite Academic Place uh so I want to talk about some academic stuff so impossibility of block stat blockchain has published two years ago and actually they what they stated is correct on the linear size Global state required uh basically they saying that stateless architecture stateless blockchain requires a frequent local proof update and as I discuss here no in this system many many local proof update right know

ztp on the distribution of Market proof uh and but actually it this is not harmful um so how to make a proof update local proof update harmless uh is all about how to make stateless architecture successful and the if we have more functionality like evm comp kind functionality we H harmful uh state of local proof update but now if we sacrifice that point as as well we can escape from there know this is a uh result we had in this research and actually inmax is almost uh no about to launch and so we

are no we are gathering The Miner right now and so we have a strong privacy as well and the Privacy mining is like uh no come to the system together and the greater make a great anonymity set and the kind of inen by that so this is a know how our privacy mining works on the know yeah so I don't I don't push to buy anything I I don't know I don't are know I don't ask any money it's it's like a mining and you know if you make if you join the Privacy together we can have a greater privacy set an so this is a know how we incentivize know our

blockchain system as well so this is a uh no one of the invitation and no don't take so much time for this and no it's uh sometimes it's stock car it's no it's mining so it's a CLI so it's if you stock car know it takes a lot of time so if you are in lost let me know so thanks so much and yeah I appreciate your know yeah patience thank you thank you any question from the yeah hey thanks for your talk um so you

mentioned that one big Layer Two role of vision do you have any existing collaboration with any ler to or did you ask for feedback or yeah M so uh yeah one big L up it's really great and actually know if we have a 100 is fine because they have different functionality and I think it's quite beautiful but now we need to have a kind of more interoperability and strong composability um base Roll Up is working for that direction uh base roll up will make many many layer does uh the one big layer too um so I think that uh this is one of the best way on the espresso or

share sequencing uh it's one of the idea to to uh go that are no Direction so I'm a big supporter of Base rollup so because it's more natural and just sharing the sequencer it makes her no composability it's quite beautiful so I truly believe that in in coming two years no this is the most uh important movement in serum ecosystem but yeah in short term to make our uh composability like super chain is doing well and the their super chain year C20 will be alive or in couple months and the I'm ex

expecting that so much and maybe there there seamless uh swap between unj and know opj or base and I think that is good idea but yeah in more trustless and extreme idea Bas up will be the best idea in my opinion and we are part of that and we we kind of transfer layer obious ser thank you a lot Leona another round of applause [Applause] [Music]