Layer 2 Panel

Transcript

hi welcome everyone uh i'm pam from chain news the moderator of edcon's layer 2 panel thanks everyone to join us yeah um we already see the default summer last week last year the gas price 200 or thousands of qa and these days we saw the nav maniac push to very high qa these days and layer two adoption may take time so let's talk about this today um before we deep dive to the question i'd like to let everyone to brief introduction and your latest progress of player two so i think we can start with lui lu the ceo and founder of kyber network

hey guys uh yeah uh we're at kyber and we're building different you know liquidity protocols uh running on ethereum and expanding to other public blockchains at the moment uh internally we have been like i think doing a lot of experiments with other layer tools and uh also like you know having some sort of like prototype and all you know uh basically our own construction for uh for for our amm on layer two as well uh and hopefully we can share the result uh soon thanks for having me yeah thanks and next one the johannes cto of medics oh hi my name is joanne so i'm the cto and co-founder of metis uh so i hope you all had a chance to check

out the presentation delivered by our ceo elena metis aims to provide infrastructure for the western economy uh what my team and myself focus on of course is the technology part of it uh currently we are busy with uh launch of the public test magnet uh thanks for having me and looking forward to the discussion today yeah welcome young from metis okay um and next one john adler the co-founder of viewlabs hi everyone uh thanks for having me my name is john uh i'm the co-founder of your labs i do protocol research and specification and i review a bunch of prs there we have built fuel v1 uh which was the first optimistic roll up to launch on

main and ethereum and we are currently building fuel v2 which will support full ethereum style smart contracts in the ut excel data model yeah thanks uh last one luis guzman product manager and blockchain researcher at stockwear hi so yeah my name is louis i'm a product manager and research blockchain researcher at starkware uh starkware is a like a blockchain company focusing on uh zkp zero knowledge proofs and to bring scalability to ethereum using uh zika rollups uh we have been uh we exist for you for the last three years um we are in production with uh three four different companies today

respectively diversified immutable solar and dydx and dydx just uh to show that the altitude's already there dydx just said all 10 billion dollars of volume in the space for four months and so they're just expand by um reduced by factor 50 at least their consumption of gas from on their one before and after working bus yeah it was i see that the the idx yeah i also tried the idx these days and actually uh yeah other than with dydx i tried optimism and aztec and zkc previously but the transaction fee is not as low as

i expected we expect it and all everyone is saying the roll up can scaling a hundred times than the later one but after when i tried optimism last week it's around ten times less than the the later ones fees it's around several dollars to to provide liquidity to the unit swap v3 so i don't know if it's too low or too too high these days but i don't know it will yet have the potentials to lower the fees in the in the future or do we have other potential efficiency increase in the future that's uh that's my first question i think we can start with uh louis from

kyber your uh the only d5 protocol here yeah sure i think i think we've done like as i mentioned earlier we've done a few experiments with different protocols as well and i think you know different protocols we will have uh different kind kind of fees right like for example but i think most of them still need to submit the data the raw data to the layer one and i think that's basically i think accuracy is there is a fixed cost that everyone will have to pay at least right um so i think that one we cannot make it cheaper um and of course you know depending on on the approach and and the design of different solutions then you might have

to pay other fees uh like for example i think in most of the zip roll up solution you have to pay for the uh for for the cost to verify the proof right but to verify the secret proof on chain and that one is quite expensive as well uh so if you can verify a lot of generations within that proof then i think the cost is smaller right um otherwise you know uh uh it's getting more expensive uh and i think um and i think um like for the case of like uh optimism and and other uh of optimistic go up then there's potentially the course of uh uh you know submitting the proof and things like that when there's a dispute right which happens uh not not that frequent i

think most of the time people like know that that kind of cost um yeah so that's that's my percent yeah yeah so and how about three scaling solutions your your opinion of these questions and you know that maybe we can start with meteors all right so i can start first uh so again i'm gonna focus on the optimistic rolex which i'm more familiar with uh so i think uh lori if i pronounce correctly sorry if i was your name um is right so lot of cost is associated with the requirements of uh submitting the data to the mgna and that data availability is very important for optims to roll up because the value of

us needs those data to verify whether uh there is some uh problem behaviors and and initiate a challenge process uh so that is almost a fixed cost uh so but i just want to first let's say one thing uh no matter what the cost is still significantly less than may not so that the value is definitely there even 10 times it's not a small feat right if you have a hundred transactions it's a lot of money that you saved on too uh so that's the first point i have uh but if you look back uh since last year i believe last year we were looking at the cost of layer 2. it is much higher than this one you're seeing there were actually a lot of technical technical announcements different teams and researchers have made

to optimize the data that especially the transaction data and pronouncing and they have maximized the the signatures they have optimized the compressions and all stuff and in the current it is already optimization and uh i think the researchers that haven't given up yet and there are still some active researchers going on uh to further optimize that data uh submission process and uh at team amity's uh so we actually have teams working on a alternative solution to try to solve that data available problems in a cheaper way so we're actually looking at uh utilizing a similar technology like ipfs the reason why we're looking at ibfs was that when we look at the venetian or

optimistic role lab and the the data access panel is actually has a recency bias versus i'm sorry recently pattern what i mean by resisting pattern is the the most recent transactions actually most demanded by the verifiers they want to look at the latest and freshest transactions to verify and that actually plays along very well with the ipf's pinning mechanism so so with those technology liking the similar exercise padding so that's why we are actively looking at the utilizing our vfss alternative data storage or data available uh availability providers for transaction data so it is actually going on right now at this and so once it's complete we expect a significant reduction of the drought cars

i can't wait to share more details in the future but i'm very confident that that's going to happen the third point i'll make is a lot of people are already starting to speculate whether all those layer two solutions will actually drive the durian price up is going down because uh we expect there'll be more transactions happening on layer two and eventually all the data will be uh submit to layer one and uh and drops the gas price up so uh as a a professional in this particular field working on those solutions i i do want to be very mindful about the amount of data we submit to the layer one when we're designing our solutions yeah thanks and how about john you how have you calculated the the fuels

version one um previously uh yes uh fuel uh each transaction should take around a few hundred gas uh it's extremely low overhead uh so to answer your question i think well i agree with the previous two panelists here which is that it's a combination of some fixed costs that are largely dependent on details around the implementation for instance you know you have starks you have to verify starks they're expensive if you have optimistic rollups you have to post blocks you have to post the data availability you have to post the signatures and stuff and all of this just adds costs you also have to do things like account for gas on layer two so depending on your

sequencer choice you might have to do certain things for instance optimism will actually just literally burn gas on ethereum in a loop they'll just burn user gas and they need this as a denial of service protection which is debatable whether or not that's good user experience to just burn gas for no reason in a for loop but you have to do certain things like this because of implementation choices uh and the second thing uh which the panelists have said is that it's a combination of uh poor or let's say immature optimizations we're still pretty young into how these things are implemented across all roll-ups and even side j and the stuff you know things like can you have a more efficient representation can you have better

leader selection that's less that uses less gas and stuff all these things will become more efficient better optimized in the near future uh so it's a combination of those things mostly i think why we've seen uh transaction fees on on layer two and other scaling stuff be higher than expected uh that being said even a small decrease in transaction cost even if it's like a factor of two which is it which it isn't it's something like what a factor of ten uh you know even if it was a small decrease it'll still be really good because you know a decrease is a decrease you you can't you can't say no to that uh but i think another point that there that we should bear in mind is one of the reasons that roll-ups are so

expensive is because they have to compete for data availability with you know defy transactions and arbitrage and mev and all that if we had a chain that was dedicated just to general purpose data availability then solutions like roll up could become substantially cheaper because data availability and execution are actually orthogonal things and they can and should be priced orthogonally for more information you can check out my talk at ecc it's a good one and yeah and last one how about luis your thoughts yeah sure so so as john perfectly said like the there is uh there is in the zika roll up so i'm

going to talk about secure rollout specifically uh uh in the vk roll up like an optimistic roll up there is a lower bound which is for uh foreign so in the case of a zk roll up you you actually don't pay for the signature so you you you pay for the bare minimum of data there is no overhead because of the proof but for instance in the context of starkwear we we have now this uh technology called sharp where you or all the system all the the applications that run into our ecosystem are sharing proofs so for instance diversify used uh move to sharp and they are doing a lot less transaction than sorel they saw their their transaction cost per like their overhead proof divided by 70 and which didn't doesn't include yet the

ydx which is also five times more like five times more transactions than sore so if like at this point whenever we are whatever the application side that you get today you basically no longer pay uh the overhead um the the full price overhead depending on on related to to to any application even the smallest one can root and can run using but there is another point about about l2's chaos which is the on chain cost which is in the context of the zika roll-up you have the more design space when it comes to data availability and as we saw for instance when so in sovereign and diversity and immutable when it comes to um nfts sometimes

you there is no it doesn't make sense to spend one dollar into when you mean a five dollar asset and while we're waiting for you 2.0 to solve the divided solution with availability vk roll up enables you to choose as a user if you want to be on chain if you want to be off chain and if you are chain you can get to transaction costs as low as handle for reddit with the transaction costs at 17 gas so yeah really the dividend space is pretty big here when it comes to dk roller yeah i see like um like uh metro labs or star stock net also have the the on-chain

solution and off-chain solution which the user can choose right right let's go relation and uh it's a pretty pretty strong design space and even like for instance let's talk about uh darkness so in in many transactions to the ontario require on-chain data um data that could be considered what we call witness in the context of transaction for instance signature our witness meaning that if i once i consume them i don't really need them anymore but for instance things like approval on the rc20 is technically a witness you don't really need it to get self-conscious which is at the end of the day the end goal so i think this flexibility enables to to new design pattern that provide lower

transaction cost per users yeah sure um next one i want to talk about the the how to say that the narrow the scaling directions we see most of the layer two already or will implement evm or any kind of vm in the layer two but we also see some of the layer to support normal transaction in the very early beginning like zk zinc one or poly uh hermes or the current version of hermes like uh or aztec yeah so um do you think the pure transaction layer two have its own an opportunity or this very narrow

scaling direction will not exist in the future or and i i also have a question around the nft things right now because is it difficult to build an nft specific scaling solution right now yeah uh maybe luis yeah sure so um so just two two yeah two two so like so i'm going to generalize a tiny bit this this question because start where started with like what you call the transaction only network i will call single dap singing that roll up and start where has been having

single dab roll up for over a year now so diversify youtube also and uidx and so the the the we what we saw so far is that they provide scalability today like having those app on the network today is crazy like i just said like the idx reducing by 50 50 x that you get until right if you value your base 6x transaction cost now from our perspective the end game is definitely moving them to a general purpose network because we have many many team reaching out to us saying oh i want to build on top of sovereign i want to build on top of youtube i want that composability and when you build like single dab

application you can get that so there is very strong network effect from general purpose and now when it comes to nft specifically you're perfectly right like immutable x right now is building a nifty only uh roll up but those nst the end game for them too from our perspective at least it's to them to move them to start it so as i said various projects can come in and build new features on top of the infrastructure yeah uh and how about jump because i think your layer two or your version two just added the gen well we will add the general computation or

bm things so what's your opinion yeah uh so i'll similar to louie i'll i'll generalize the question uh to not just payments but also to just application specific rollups or application specific chains uh i'll also take a more optimistic point of view which is that i think there is room for them but you have to be careful uh because there's a trade-off you lose composability but you gain potentially a lot of scalability and a lot of flexibility uh and you can potentially get around this lack of this lack of composability uh if you don't need like atomic single transaction composability like you're

not going to flash loan nft so like do you really need single transaction atomic composability as opposed to just being able to query the data maybe not uh so solutions like cosmos is ibc could be very promising here for cross-chain communication protocols which would allow multiple application specific blockchains and roll-up service blockchains to communicate with each other so i definitely think think they're the space for them but that doesn't mean everything will be application specific there's pros and cons to each and i think there's a space for each if people are careful with what they build yeah and i have a question about you is is it hard to to implement the nft things in your version two

no it'll be very simple it'd be like implementing out these on ethereum fuel v2 will support full ethereum style smart contracts so whatever you see in the evm today and solidity or whatever a little bit of that will probably be easier because we're also implementing a high-level language that's based on rust called sway so it should be even easier to write secure efficient smart contracts than in solidity okay sure um next what about meetings uh yeah so for me this we uh we have a general purpose uh evm in place but i do believe one thing uh so everything we do here uh consists of some assertion of some degree of tradeoffs

should have some amount of security performance costs and degree of centralization composability that jumps out so there's there's always a balance we want to aim for um and uh uh immediately we we choose uh this breadth of application use cases this general ability general general the ability to handle general case uh over the other things specialization stuff uh but i believe all the different specialized chains they all have great and fantastic technologies actually i i believe they they can not only survive but i think they will excel in those specialization areas and their doctor needs in the market and this whole thing honestly this is a vast blue ocean

and there are i i believe there are enough opportunities for everyone to to survive and and throughout them so that'll be my point yeah sure and how about louis do you have anything to add yeah so uh at kyber we are building our own uh amm specific uh layer two right using zika web and i think what we realize is that we can do a lot of things to optimize the circuit right and hence optimize the the proof um but you know there's certainly like some limitations you know in terms of like composibility with other apps and things like that but you know we treated it as an experiment so i think for for us this is fine um i think one thing that we probably will see in the future is that

i think the guy the gas phase between apps in layer one will happen in layer two right uh and um sorry it will happen in the general purpose layer too um so yeah i so i'm not sure like you know whether uh or the general purpose layer two we have like any thought on that or they have you know thought on that at all uh yet right i think that that would be an interesting excuse to to see in the future yeah i think everyone agreed the end game is a general purpose and there too but everyone has any current solutions right now i think software is that is a very very best

to have dydx diversify or mutable yeah every kind of um app specific solution yeah next is about um uh oh sorry then uh before before can i address what we just threw out at us like are you the gas price for general purpose uh uh layer 2 solutions um uh i i think that's a very good question i believe a lot of people are saying hey if you have a layer 2 solution will that happen well will the gas price volatility be the same as the main x where you can see it constantly increase based on the demands of the system uh i would say it is definitely something that we absolutely considered when we designed the system uh at using metis uh we have the uh the sharding uh already planned in the design and that's part of the the

architecture of it uh so when the uh when the the train becomes busy the charting kicks in and uh we'll keep the gas price slow but it is something that we don't want to consider very heavily yeah um i don't see something specific specifically about this and actually i'm going to quote john from the fcc occurred document talk from fcc which is it's all about hardware requirement um so so the problem right now about ghastly is like people when when people talk about scalability they they if they compare for instance guarantee of ethereum versus scalarity or bitcoin they don't they don't

understand they don't compare the same thing bitcoin target machine is a raspberry pi when ethereum target machine is roughly a 40 machine on a ws and we are not talking about the same thing the same thing solana machine is expected to cost a thousand dollar a month so when we're talking about like you know what machine do i need to verify things and scalability it all comes down to what machine it's required the thing is and this is where uh vk roll up has a very strong advantage is that you cp roll up separate the the full mining like the people the sequencer nodes which are can have extremely strong machine because they're getting paid for

it and the average guy average joe machine the guy was going to verify without being a sequencer and what we care about for decentralization is the little guy being able to verify the big guy getting paid for it i mean fine you know we need this centralization so we don't need like a google size machine but if you require a machine that cost him like 5 000 a month he's spending more in gpus today electricity so it's not a big issue fire um um and and um so so as a as a consequence what i'm trying to say here is um you shouldn't expect

a high gas pricing at least into the key role because we can scale machine for sequencer while remaining low bandwidth and low cost for for for non-sequencing yeah uh john do you want to give me any feedback reply uh sure i mean regardless of how much throughput your chain supports eventually you'll end up hitting some sort of bottleneck somewhere right uh like even if you have you know logarithmic verification time that logarithmic isn't constant you know eventually you'll hit you'll hit a point where uh you'll hit a point where that logarithm is you know bigger than the

size of the earth or something if you have a computer that big you know if you're if you scale up that high so you know even if you increase throughput a huge amount uh you'll still end up with a fee market the question is how much good for the world can you have until you hit that point when you hit that threshold when you have to have an actual active fee market uh no starks and starks are good in this respect because they do provide some benefits in the sense that verification is cheaper than production and i do want to pile on on this point that a lot of people like i agree completely with louis here that a lot of people say or complain you know oh we need a really cheap validator no it's like no

that's not what you need you know validator no it's going to be expensive although block production notes can be expensive it's the user full nodes that must be cheap because we want to be able to keep the block producers accountable right we want to be able to know that they're doing things correctly and know when they're doing things incorrectly to either ignore them or penalize them in some way right it's like if you take an e2 node okay it cost a hundred thousand dollars for 32 eth at the current price of ether right and like you're gonna say well you need we need to constrain your validator nodes to run on a raspberry pi that's ridiculous like i'm sorry if you have a hundred thousand dollars to run one single validator node you can afford a laptop you can afford you know more powerful desktop you can afford a

slanted full mode even at that price right so it's a little bit ridiculous to say you know to optimize for the wrong the optimizing for the wrong thing if you say you need to be able to run a block producer and raspberry pi bank ball produces as expensive as you want is irrelevant as long as we can hold them accountable and in that regard starks and starks do have a do have some nice properties yeah um louise do it do you want to do anything do you have anything i mean like like i think we louie or luis luis so i mean oh i mean i i i couldn't agree more i mean i mean i agree i kind of

went a jump up a bit ahead when i said that there is no limit of of machines of course uh and there is some uh upper bound um the factory actually for for your knowledge john it actually the likelihood of us getting there basically filling the whole block with a proof i think we require like google size machine uh google site database uh data center to to to get to those type of proof in terms of generating the proof but the outside of the joke and a 100 percent agree aligning with what john said yeah sure uh so let's get to the third part the adoption things of player two as the optimism and arbitrage still not release

the full maintenance version but they are all talking about this month or next month and zika sync stock net or aztec not release the programmable version right now so and do you think the layer 2's adoption still takes much longer than we expected or maybe we need to wait one year to to let the adapts to move to layer two and users move to migrate to layer two it should take much longer time yeah and what's your thought yeah um maybe start with louis from from kyber i think user experience matters right

i think what we have seen at like for example uh bsc and also like the the pos of a polygon chain uh i really really showed that user they they just like move to uh uh you know the chain uh where they can do really fast and cheap connection and can can really enjoy a better experience i think i think player 2 is approached is approaching there currently i think it's still mainly for the technical users right um and and i think another reason is that i think the trade-off in different layer two solutions are not explicitly communicated

um like for example like you know what you need to submit to l1 and what what you're supposed to pay when is the generation and actually the final lies and things like that i think all these are quite complicated like for the for the normal users to understand i think it's gonna take some time and some education as well uh apart from you know the the technical requirement uh the technical development requirement right um yeah but i think is you know we are we are getting there i think with the development of a few projects that start where it's involved i think we we see great traction there uh hopefully you know with the with the other uh project like uh optimism and and a few others um we're gonna see you know much better adoption

yeah maybe you can you can start with uh liquidity mining to incentive users to move to layer two i think yeah i think that's you know there are going to be like several ways to incentivize to move there right i think people already did that um yeah yeah sure and how about luis from star wars yeah i mean like i can show you that you once you put the right incentive people move to l2s and i guess the simplest and easiest one is dydx yeah yeah sure i mean uh they they just uh passed like i said 10 billion they they are i think last yesterday they made 500 millions in in

like volume in the thing like the same day um and and then the the the digitization the token provide definitely is a very strong incentive to to to to move um now um like the if the question is like the adoption of l2s is going to be longer than expected i mean l2 is already there like i said uidx is extremely successful uh we see the success of sidechain already um and uh sidechain as like we have been talking to various projects and one of the main issues they have with sidechain it's not decentralization is is that because you use a tech that is you know

a bit consensus based you you your bridge has has very big risk so you i expect to see a lot of those existing projects big project that goes to sidechain to move to an l2 to get a more better proven security on their bridge so either to adoption to longer than expected i would say it's already here okay yeah but but i just yeah sorry sorry to jump in but maybe it's better to uh you know if you have some some numbers on like the number of users and and things like a number of transactions and like that on uh on a few platforms that you have i think that would be interesting yeah all right uh dydx i think it's important the

public i think the number is public so i can give it to you i think it's yesterday was the uh last week they did 600 000 transaction um uh which represented something like if you put back that to like on ethereum would be like 12 like starcats deployment in the last week in terms of transactions represented 12 of the ethereum bandwidth like if those transactions were to be on ethereum they would be 12 of the of the entire block space in fear um and so so i mean it's already yeah i pointed that and this is growing uh last the week before it was seven percent like so is picking up the immediately speaking of diversified speaking like we

and from our perspective we see like just a growth coming every day it's pretty crazy yeah yeah i know but but when i start this question i just want to highlight the roll-up part you said ui dx is it's a different story i see yeah um yeah and i think next uh yeah how about you yeah sure all right so my answer will depend on the definition of altered adoption as you can already hear uh i think louie and luis that they have different answers because they're definitely different definition of relative adoption uh i'm gonna throw something controversial too to me i really believe the overall blockchain

adoption is still pretty low compared to what is happening out there in the the other side of the world um as i mentioned earlier this is a huge blue ocean and there's still many opportunities for everyone to take off i hope one takeaway that this audience has um is the awareness the technology is almost there i tell you the technology is almost there um and we're ready to dramatically expand the use cases for watching we just need the uh the applications uh we really think i only think we are at an inflection point where we can see we can witness a explosion of blockchain uh based on application allegedly is definitely one of the major catalysts for this golden area of blockchain you can see a lot of delays from optism from arbitron um because this technology is indeed very complicated uh

there are many moving parts uh in the system i do want to for the audience of the complications and challenges we're overcoming and we're still facing but it is definitely a fascinating technology and and uh i absolutely believe the future and i think similar to leaves i think the adoption is near i think i don't think it just needs a very good application to um to let people uh to this technology and get the general public to adopt blockchain as i say in mainstream technology and very confident about it and how about john your expectation of your your field version two's adoption uh sure so can i get some clarification

clarification on the question is it are you asking how long will adoption how long would adoption take after these chains were deployed or are you more asking why did it take so long for these chains to be deployed the first part yeah first okay so after they're deployed how long will it take to get adoption yeah sure uh okay so in that case i would say uh for all these chains adoption will come very quickly once they're deployed uh and we've seen this um optimism for instance right optimisms ux uh both in developer and user side was quite poor this whole time and this isn't like a slight to them because guess what ethereum's ux is also pretty pretty bad right uh and it's not it's not necessarily getting better like you you

have 1559 introducing two fee settings now instead of just one uh so it's you know the ethereum's ux is also pretty pretty atrocious so the uh you know this isn't like a slight optimus in particular but it's like you all all ux around there too is terrible right uh and it can get better but the thing is despite the fact that optimism had no composability because they had no applications other than one uh and the developer and uh user ux was terrible uh which again are not slides uh despite all this they still have 100 million dollars locked because of one little thing which is called printing money and giving it to users through liquidity mining uh you know polygon not to single them out but you know they're they're a side chain uh at least

the proof of stakes i chain to the side chain just before mihailo comes and yells at me uh for call out for calling polygon the side chain uh they didn't really have too much adoption going on when they were trying to do their magic chain and stuff and then i think if remember all avi came on board and they did liquidity mining and just gave people a bunch of money and now they're one of the biggest you know scaling projects on ethereum or in the blockchain space in general uh you know tokens can be used to build communities for instance and you know i think many layer twos are either going to use their own tokens or they're going to do liquidity mining from projects that build on top of them for example you know ave launching on polygon for instance optimism doesn't

have a token yet publicly but they do have synthetics that was the one application with no composability because there's no other applications uh and that one application did liquidity mining affirmative well on optimism and they managed to draw you know 100 million total value locked in there and i expect other applications that launch or that relaunch on layer 2 will do similar things to that because i mean otherwise no one would use those applications right so it's it's in this it's for the good of the applications too but they would do such things uh so i think it will be very easy once these layer twos are actually launched if they support liquidity mining either on their own or through applications that build on top of them i think it'll be very easy to get adoption the problem

is if there are two launches without any sort of incentives uh then then then you have then then that might take a bit longer because then you're relying purely on technicals and as we know in the space no one's in there for the attack we're all in it for a number go up right yeah sure that's that's why i want to talk about the the fourth question because uh last layer two talking about their consensus i think most are there to still not release their consensus of the validator or sequencers and i remember most of them starts with one operating node and because their lack of consensus do you think this become will this become a very big problem of the layer 2

and have you heard or do you know anything about the progress of the design the layer 2 because how you design the layer 2 consensus um yeah i think um how about you sure i can start off uh the nice thing about rollups in particular is that since they inherit security from the main chain you don't need a full-blown say bft style consensus protocol you don't need two-thirds of some stakers voting on some block because you don't need that security right the point of two-thirds of stakers voting on a block is if they vote on an incorrect block or an incorrect fork or they try to reorder the chain you can slash at least one third of that stake right that's accountable safety but you

don't need this because you you have the base blockchain you have ethereum or whatever chain you use for data availability it could be celestia for instance you have that chain after the time stamping mechanism once you have a timestamping server then you know you you don't need the block producers of the roll up to timestamp things right they they're just there to create blocks essentially to order transactions not to timestamp them uh which you know are orthogonal things uh so you know the the way you do block production and layer two the way you do this consensus in there too can be much uh it can be much simplified than what you have on there one so i've been calling on this process merge consensus and you can see various posts and stuff i wrote about merge consensus uh

but the now the implementation details or like the how you do this specifically every single project i've seen does it differently right like arbitrarium has this inbox model fuel v1 has a priority aggregator they don't have individual transactions but anyone can be a block producer just one has priority optimism has uh see single sequencer now i don't know what their plans are in the future uh but either way every every chain every every raw project has a different way of doing this and i'll say that it's let's say it's still an open problem how to do this like if there's a way to do this without major trade-offs i think it's still an open problem every choice that every chain has made has some downsides and it has some upsides

uh for instance optimism because of their particular choice they need this notion of nuisance gas which is like i said before they just burn layer one ethereum gas in the loop they're just like waste money literally they just burn money they just throw it they just throw away money and they need this as a denial of service vector prevention uh and this is because of the particular choice of leader selection protocol if they did something different maybe they wouldn't need this i don't think arbitrary needs it uh so there's upsides and there's downsides to every single choice and there's so far we haven't seen one leader selection protocol that is just better in every that is strictly better we it's still an open question yeah sure um how about from kyber what's

your perspective yeah i think um you know the centralized operator on layer 2 will make uh you know all the other uh all the attacks uh regarding like centralized miners on layer one uh words right like for example like they can order the generation as they like they can even decide to exclude like you know transaction from from the users that they want at school and and then things like that right so i think um i think sooner or later probably most of the layer two will happen to find a way to decentralize the operators right um i think um uh yeah and and and like like even for kyber in in in our experiment we are also like working uh we also building a full flash like bft

um you know consensus for our uh for layer 2 operators um and and the reason why we can do this is because we uh we basically like focus on one particular use case right so we can optimize the circuit and things like that to make it quite simple uh i'm not sure like whether like it's gonna like make things a much more complicated for archer for other like general purpose uh layer two um but you know for for kyber specific usa uh we we do see the advantage of uh you know focusing on one single application there um yep yeah thanks um how about you guys i i was not saying uh you may sound

familiar the answer depends on your definition of consensus um so so in my opinion um uh earlier two solutions familiar with the underlying technology they have consensus mechanism it's just not the traditional way of consensus um uh using options overlap as example the the data access data's consensus what i mean by data consensus is that they actually submit all those data and that consent is happening on layout so that's happening all the time every time that what happens but the execution consensus is not happening on the uh it's not happening all the time it's on demand by the challengers but what i think the execution consensus is if you're given a initial state they have a bunch of transaction a batch of transactions and whether the end state

is actually the result of the execution of that budget transaction that's the execution consent consensus that's not happening on layer one all the time but that's optimistically deferred to the top challenges on the pay-per-view basis so so on that note um i have a similar opinion uh that i do think from security point of view it's not end of the world to have one operator of the internet because the consensus actually are inherited from the layer one i mean that um but uh uh i mean if we do admit this we do value decentralization uh to guarantee business can do it continually i think that's probably the bigger problem um as i mentioned um um when we design the technologies they're

always trade-offs uh and we're constantly making trade-offs all the time actually and i i absolutely understood why um all the later two solutions are designing uh making certain design decisions especially in the short term um so our goal uh immediately is to allow anyone to operate a node and try to participate into this uh metis multichain ecosystem so as of yesterday we have actually enabled the peer nodes yama network and that's a important step towards that mission i'm very happy about it and we made it uh yesterday and so we're really looking forward to talk more about it uh in the coming weeks uh at the same time we also have teams

working on a specific validation system called rangers uh that is again not the traditional consensus uh that we usually talk about the block consensus it's more of a transaction consensus uh focus on the execution consensus uh we actually that system employs a set of randomly selected uh participants provided transactions and studios all the time and get incentives along with it and interestingly um it is actually quite a similar mechanism the vitalik uh talked about on the first day of the conference purely a coincidence by the way so so that that was quite fascinating when i was uh watching that particular talk but

um but again again uh every system that we put in any additional system uh there are traders for example the um the violation system we just talked about the rangers uh it will actually drive the uh the transaction cost up uh at nature because we're gonna give incentives to the batteries of course it's gonna come somewhere um but then this class will be uh hopefully upset offset by the uh the use of ipfs which i mentioned in the first question that should dramatically reduce the transition cost because i see here i'm constantly making these uh decisions to to of different trade-offs and hopefully we can strike the perfect balance and that will go yeah thanks i think uh the staff will

start that using different design is it correct lewis so so let's uh i mean uh roughly speaking start net has the same you know same core design the indirect uh when it comes to to uh to um to to design space meaning like a single sequencer and we are going to start with a single sequence i mean we are starting with a single sequencer circle being in testnet uh but we are in universe we are looking for decentralization as john say uh the leader election meaning who is organizing the current batch is still an open question um um so so um we

right now there is a we we are still looking at the various you know approach to all of this um we don't for instance one one of the main main problem you have is like for it's enforced transactions first transaction is is one of the most important features of optimistic product in the zika roll up you don't you don't really need for the support transaction to get the self custom and so here we have uh um a question that does impact strongly how you do leader election and so all of this is still under on the still open question right now we're focusing on getting a single sequencer right as you can try it on testnet and whenever we nail that down properly we will expand to towards decentralization

toward bringing the censorship resistance that we we are looking uh looking for for the for the out yeah thanks and i think our time is running out um so i'm calling out uh closing it and thanks everyone for for joining us at carl's lair to discussion and thanks john yuan luis and louis from from your team yeah so i'll hand it back to yeah thanks and heading back to the the ad con yeah thanks bye everyone