Are blobs good for Ethereum? | Devcon SEA

mik mik check hell hello thank you for coming to our panel today I think it's going to be very interesting um we're going to give the bear and bull cases for blobs I think on ethereum uh and as you can see the title is are blobs good for ethereum this has been a huge topic of discussion within the community for a while now um because no one really has the answer I think this is an open question and that's why we have this panel so I would like to introduce our panelists today we have uh Brad uh Tim and ancar they all work on different things within the system but I'd let them introduce themselves so we'll start with bread uh yeah hey everyone I am Ox Bread Guy uh former K now I'm part of the mega eth team uh I just got nerd sniped Honestly by blob specifically and just the value cruel method within our mechanisms within the uh modular ecosystem so a lot of long form post that's actually how I met Tim um that's actually how I met Anthony I just uh talked about it a lot maybe some of you guys have seen some of my research stuff uh and yeah hi I'm Tim Robinson I'm the head of research at Bard Capital um and I recently created The ethereum Blob simulator that you might have seen go viral on Twitter a few weeks ago um to figure out uh exactly how blogs react to supply and demand and um are they actually good for ethereum and uh I'm Anar diis I'm researcher at the eum foundation I'm very opinionated on this question because I'm one of the co-inventors of of blobs one of the co-authors of uh Protank sharding and now I'm also working among other things on um p does which is our next step in scaling The Blob throughput great thank you guys so the main question I see or I think a lot of the ecosystem sees when they go on Twitter go elsewhere about blobs is um you know should they be paying more to the network you know this fee value acral to the network has traditionally been you know pretty much zero lately over the last couple weeks uh we're at blob saturation so we actually are generating fees from it um but I know each of you has a different opinion on this and I want to start with bread because you've been very opinionated about this on Twitter so do you think that uh blobs should be uh free as a loss leader for the ethereum network um or do you think that we should be charging more for them regardless of saturation yeah okay so I my my journey through discovering and and watching blobs has been evolutionary I started initially on the okay ultrasound money I come from the school of bankless that's when I started watching stuff right that popularized through there um I saw blobs blobs come live fees go down what the hell is going on with ethereum and then why is it happening what's the future look like all of that uh so starting there looking directly at fee and and value cruel by by the way of burning ethereum making it deflationary uh and then seeing how it progressed to okay I have to accept uh ethereum as as money which I do now believe uh is the ultimate value prop for the ecosystem uh I have gone from oh my God this thing is never going to be uh value accretive in a burn perspective to okay it does not matter we need it for aics and E's money uh it's good as a loss leader we need we it's good psychologically to see l2s using this thing we can debate about we will debate about security properties and what that comes with but I think ultimately loss leader is the uh is is the way forward um yes so I like to answer these questions with data and it was bread guys thread that actually inspired me to create the uh ethereum blob simulator because a lot of people had opinions on well will ethereum ever be profitable again will we ever burn uh eth again um and like I thought well obviously we will but at what point will that happen um how much will transactions cost things like that um and so I I put together the simulator and um you can kind of see using it that um currently if we try and charge for blobs the the fees on l2s go up quite dramatically and that is a pretty bad idea we don't want to encourage them to move other to other sources of data availability um but long term um when like a few years maybe even a decade from now when ethereum has hundreds of rollups during tens of thousands of transactions it is actually possible to have uh blobs both keeping fees cheap and um burning eth and so uh I do believe we should be lost leader for now I think we should build up a bigger ecosystem I think that's much more important than trying to extract any kind of money now um but I wanted to give the the bull case that long term it's going to be okay right yeah um my perspective on that is that like I think given that our one has been so congested for normal transactions for such a long time we've kind of lost track of the fact that in a way the fact that we're even burning anything at all with 5059 is more a result of the embarrassment of physically just not having enough space for all transactions now I'm saying that but like it's not because we're doing a bad job or anything I think fundamentally we just there's not enough throughput on a single chain to process 7 billion users worth of transactions in the entire world so inevitably there will be congestion on the most popular L1 um and you need basically pricing is the fairest mechanism of allocating that to the people that have the highest payment willingness so basically you get some in inflow of of transactions fees fees that way and burning them makes a lot of sense there but it's not because we designed the chain to generate Revenue it's really just out of this embarrassment of phasically just not having enough room for everyone and and actually I don't know if how many of you all were maybe actually out of curiosity raise of hand who of you all send a transaction on ethereum before 2017 wow that's a lot of I I really expected like four hands and it's like 50 hands that's that's that's impressive but you might you might remember like before the first bubble it really was basically free to to to use a one um and the same we're in the same phase with blobs now now our goal is to scale as hard as possible on the blobs and I think we have a pretty good road map there but inevitably the same thing will happen right like we will not scale infinite infinitely and theand will go infinite right and so like at some point we will find some some point where basically like unfortunately again there will be some congestion fee hopefully if you amortize that on like a per L2 user basis it will be very cheap but there won't be a way around that but very specifically etherum is not a platform that specifically tries to like use its Monopoly position to extract more value than is necessary for the suggestion yeah yeah I I I think that's an important point with the how I mentioned the Lost leader and you guys hit on this as well is that uh we need to keep people in the ethereum ecosystem we want to keep them in the ethereum ecosystem and we want them using ethereum and without blobs I think that ethereum would be in a much worse position just personally and I think um that's pretty agreeable regardless of what you think about their value of crel properties and things like that but blobs aren't the only da layer I guess you could say in the ecosystem there is Celestia Ian da ail a bunch of others out there there's even data availability committees that people just spin up with a bunch of other people and put your da there so I think one thing that confuses a lot of people is why would people use or why would the rollups and layer twos use ethereum da over one of these other da layers because this the switch isn't simple like if Bas or or someone else or arbitrum wanted to switch to another da layer they technically could and the users wouldn't really feel it right um maybe they'd feel the cheap potentially cheaper fees but it would be the same experience to them but why are they still using blobs why are they still committed to using ethereum blobs even when they've started going up in in price so to to establish my position and I think actually why a lot of my posts started gaining prominence I am the chief Fodor of ethereum blobs I do not believe eth da deserves a premium I it deserves a premium I do not think teams will respect that premium on a long enough time Horizon and why I believe that is because a lot of the short-term incentives to switch from ethereum blobs to aan Da Celestia Avil is the users themselves do not understand like the the big thing is like oh I i' I've seen like torl I fight him on on Twitter pretty consistently he goes all the tvl is on rollups that use that use uh ethereum da and I go bro no one knows what the hell da even does for them as a user and that's all Downstream from a multisig that can snipe the stuff out the back door so like to to justify that like people are making decisions right now based on da when to the user You' called it out a little bit to use an alternative da layer well the che the teams that are businesses can lower their their costs as a business right it goes from tens of thousands of dollars down to potentially you know hundreds of dollars and the end users to get a cheaper median fee so like both of those incentives to me tell me that on a long enough time Horizon the users will complain and we we talk about capacity ethereum blobs will not be able to scale to the capacity of some of these chains I'm working for uh Mega eth right like we straight up could not even use ethereum da because it doesn't hit the capacity that we're aiming for so like not even an option and we're starting to see bass Bas is approaching the limit of what they're doing they're projecting like we want to scale one megag per week sometimes we'll go a little higher Jesse is on Twitter pleading like please ex scale blob space because if you don't scale blob space we are at a point where we're paying more and our user median fee is going is going up right like and we we just we can no longer on board the world everyone can't come on chain so I think like natural incentives is pushing people to Alternative da layers and even though it may be better on a longtime Horizon to use ethereum better security better whatever the short-term feedback loop is like its incentives to push people to Alternative layers yeah so in the short term I mostly agree with that assessment that like when lates are only at stage one or stage zero and they have multi six like it doesn't on on all the risks of ways you're going to lose your money like having an alternative D is very very low on that list um the main reason I think blobs make sense the long term is because you can have when you have full stage two ZK rollups um these can be um completely proven and secure by ethereum with no external dependencies in that like um when you want to update the state of this rollup you have U The Blob data you have the proof and you have the state they're all held by The theum Valor there's no one else giving some external data there's no Oracle there's no possibility of an oracle attack or any kind of external da attack um and um we when we if we want uh to be running the Global Financial system in the future and having billions or hundreds of billions of dollars on layer 2s um a lot of these uh institutions or people building large Financial products do not uh often don't want to make that trade-off of of uh security uh of having external dependencies for a small saving um but this also depends on how much blobs end up costing if uh ethereum doesn't actually increase the blob limit and it costs a ton of money to use ethereum um there is going to be very few people using daa so we we do have to scale to that point and be somewhat cost competitive and um I actually think of it similar to like a Content delivery Network or say S3 in the web 2 world in that there are cheaper cdns than like cloudfront um and there are cheaper uh ways to store files in S3 but a lot of people still use them because they are um kind of links to what they are so there's the security tradeoffs uh and it's it's also like the amount of effort it takes to switch off um if if you're any saving a tiny bit of money who cares if it costs if if it's costing your users a lot then there like Bas is now complaining that they're obviously very incentivized to start using an alternative da layer yeah um I think actually it's like a really really interesting question right like how the endgame for da will look like um there's a lot of people right now exploring how how it could look like even and the reason why ethereum even offers da at all right like a lot of people ask this question was it a good idea to even start like initially we didn't have a DA service and then like we created props and the question is why did we do that and it was specifically not because it was like oh that's an exciting New Market we can generate some extra Revenue but it was specifically like a consequence of the rollups road map right like we um initially had this like a different different plan with with the exe with a sharted execution and then it turned out no actually doing that in a distributed way with rollups exploring on their own makes makes a lot more sense and then the question was what do they need they don't need the the the distributed the shed execution but they do need the the data and so that's why we kind of um uh decided to build this this datail availability product and fundamentally like indeed like what is the difference like for the end user what is the difference whether you a is on ethereum where it's POS possibly at least reasonably expensive um or it's on Celestia or maybe even it's just some you put it you know you you publish it on AWS or some somewhere like you just push it somewhere and say hey it's available don't you see it um and it is a good question right like what actually is the longterm like how far can we even scale I think no matter how much we scale the there will always be a premium for like the ethereum da8 just because having security and settlement like the8 a security and settlement in one hand is the most secure way to go but of course if we have low throughput that means that that will only be ever be useful for like some special purpose l2s where maybe I don't know if you you know for your real estate transactions you know like you have a real estate registry and you only send one transaction every 10 years and that uses your like onchain da maybe you know that's a world where we we have low throughput high very expensive da if we can push to literally infinite throughput then everything will use etherum da that's probably not possible in a fully decentralized way so we probably going to go for some sort of middle ground right where like and the question then is what does that mean does that mean that a network like Bas just because that that that that example was given will still use unchain da I would assume yes I think we can scale to that level will Mega eth be able to use unchain da hopefully less less clear hopefully um will the long will like a gaming L2 be able to use unchain da my assumption is probably not like not not not L1 da right like they'll probably use some cheaper alternative and that is going to be an interesting um security trade-off one last thing I would want to mention there is there is this interesting wrinkle that we're moving into a ZK world right so um most l2s will over time turn ZK and in ZK you have this interesting design space that in principle you can have multiple sources of Da for a single L2 that's in principle not possible for an optimistic rollup because if you basically can mint infinite eth somewhere in the system you can basically flood the entire system with it but with ZK because it ensures Integrity even if you can't see the data you can use multiple sources of Da and then you have like some subset of the L2 where basically you you can in principle grief the people there and just not reveal what the current state of the system is but you can never steal from them and so in principle maybe the long long longterm endgame could look like some el2 basically only have a like a maximum security range of addresses that uses unchain da and then like basically free everywhere else but yeah like again it's a really interesting design space and all we can do from our side is just scale as much as we can and then let the ecosystem explore yeah I think it's interesting hearing all the different perspectives here because as you can see the the common theme of the panel today is that there are a lot of open questions to these things and we're working through them as an ecosystem um and I think uh we're all aligned on thinking that you know blobs I mean so far a good or like a net positive for ethereum but there is competition out there there are tradeoffs um and at the end of the day what users care about over anything is cheap fees fast transactions right uh and if the creators of these layer twos their ecosystems if the fees start going up because blobs aren't meeting that demand then we do risk losing these uh layer twos and Roll-Ups to the other uh data avability laters but that probably Segways nicely into eth value acrel when it comes to blobs because as Brad mentioned just before he's become a bit of an eth's money Advocate lately I think the longer that you spend in ethereum you're probably more likely to become this advocate uh just because I think you get you come to the conclusion that a lot of the value acral from fees and me and stuff like that tends to get captured by the uh people generating it or by like the apps generating it rather than it being bled to the L1 and I think on a long enough time frame this happens to every L1 uh personally but i' I'd like to get each of your thoughts on uh you know if if if e needs to keep scaling up blobs and needs to keep them basically Dirt Cheap I know Tim you made a simulator for this but do you ever imagine blobs being like a huge value cruel driver for eth or do you think it's complimentary in that it allows us to scale up eth as as money through these different uh layer twos I think most value is going to come through mimetics of it occasionally crusting this level of it being like the progression is going to be hit limit sit at limit for a little while they're going to reach maximum maximum efficiency to like compress as much value as they can subl liit it'll eventually Spike over we're going to have an air drop or something that's going to make stuff go crazy mimetics of that Burn's going to go crazy it's all back on the menu awesome like whatever and then ultimately these teams are going to go like go off like when I when I first started doing the research posts I admittedly I think I over indexed myself like I I was full on thinking like the lizard people are taking over like they're immediately going to switch over and they're turning the frogs gay like I was Charlie day doing crazy stuff and thinking L2 is immediately the second it's not free they're going to they're going to turn off and go where else I don't think that's going to be the case I think it's going to hit now now that I'm taking a step back it's going to hit this limit they will burn a considerable amount of eth because the pricing does ramp up very quickly that's going to happen we'll start seeing some weird stuff right they'll start adjusting they'll start like not posting people start asking questions it'll bring some edics back to the actual conversation that'll acove value just it mind space But ultimately I still think I still think teams will act in self-interest because it is advant advantageous frankly to do so for users for them like there's too many Natural Market Market incentives to move away from eth Da One it reaches the point point of consistent and consistent plus saturation that they'll migrate off to an Alda and then you have to go back to the process again base is gone right they eat like 40% of blob space they roll off because they made the decision it's better for them as a business fair then more people come in E this premium e is whatever maximally aligned fills back up spikes over the leader transitions off to another one and we constantly have this oscillation of like comes back in a Vogue goes away comes back in a Vogue goes away burns the on of e goes away uh yeah yeah um I think that makes a lot of sense that we'll have these um waves of um burn and uh basically free blobs as we keep scaling um I always liked um David Hoffman's 2009 post about ethereum being a triple point asset in that it has uh both it's both interest bearing it's um a store value and um it is used as a money um and if we do not have any burn then um it loses that leg of that um which is okay um but like long term I would like to see it uh basically be that triple point asset um and but I do think we'll go through these Cycles um of burning a not um and as long as we don't uh try and rent seek basically or like extract value from l2s and charge them too much um I think that'll be fine and like users uh will when they're already starting to use apps and using a chain then are going to go away because the fees went from like 2 cent to 10 cents um cuz like we saw people use ethereum and pay $50 for transfers a few years ago um and even now with you things going back to 10 cents people are still using the chains so it's not too big of a deal um but when it's uh a year straight of high fees that will be very bad yeah I mean I think it's really interesting right because like this is where the debate touches on the question of what even gives eth value and do we want to maximize uh Fe fee acrel um and I think that's very much an open question like I have opinions but I think it's a it's just because this entire space of of of blockchains of crypto of crypto assets is still so new and we just do not have a very good understanding of how these systems actually behave I think really we have to plan for all scenarios like I'm an ultrasound money guy I think ultimately the monetary premium is all that matters and I don't think the fee uh capture is important but at the same time that could be wrong like it I don't think I think it would be completely insane to be to have high confidence on this right like you have to be open to the possibility of that that's wrong and that actually you need the Fe acrel um and why do we even why do we care well we care because eth needs to have some minimum amount of value that staking is safe right like if if basically we end up having zero Fe Cru and that's what matters and then eth goes to zero the system doesn't work right so we need we need to make sure e eth has some minimum amount of value and that has to scale as the amount of economic activity on top scales like today's security level might not be good enough if the entire world economy runs on ethereum right maybe it is maybe it isn't but like it definitely can't if if it goes down by a factor of 100 now it's definitely not secure enough right so so we need to make sure that e has some value um I again I think there's a good chance the Monet that monetary premium is what we need if we need the fees I think it is possible that like a couple of years from now we might have to reassess I think most likely I talked about it earlier right like the there's fundamental kind of limit of resources that you can have in a decentralized L1 so most likely congestion will just forever give us enough fee Revenue as same on with blobs as well maybe at some point technology becomes so good right like too cheap to meter that actually we will have to consider having value capture mechanisms Beyond congestion I don't think that should be on the table anytime soon though I think there's something important to be said about fee generation as a proxy for activity and demand versus fee generation to the point of ultrasound money right like that's great for mics and I think the the ecosystem does broadly look at fee generation CU like look at salana right salana is generating a ton of fees now relative right uh to what they were previously and people see that and they go oh look there's demand here that's not doing it to the point of deflation of the asset which like unlocks another mimetic tier of whatever but I think so yes fee generation is is important mimetically like signaling socially like pulling people together it's like I think that is important I don't think like Le like looking specifically at the ultrasound like level as like look now all of a sudden this thing is worth worthwhile is a a thing to like hinge our entire the entire ethereum identity on yeah and and I think you all made made a lot of good points yeah and you discussed blobs as if they exist in a vacuum though that's one thing I noticed right because we don't just have blobs we have the normal L1 female Market as well and I think there is this obvious misconception in the ecosystem that the L1 isn't going to scale it it is to a point like not like these kind of Mega chains out there Mega l1's monolithic l1s whatever you want to call them um that's not ethereum's path but the L1 will scale but the L1 regardless of it's scaling or not is still generating fear revenue for eth and we've seen lately as the market heats up activity heats up it's starting to generate a lot of fear Revenue again and at times we are net deflationary and that's the thing you now have to add the normal L1 fale Market as well as the blob fee market and then you get you know the extra burn there so I think that I I like the the oscillation thing where you know you'll have um I guess Bull and Bear periods um and I think that's just true for pretty much every fee Market within the ecosystem um but I think when you actually look at the yeah the longer path like as as was mentioned and and where the value will actually sit and who wants to capture the value L1 tokens have yeah a very difficult time capturing value from me and fee Revenue uh versus capturing it from being a money and a store of value and that's I think the path that we should be pushing on as an ecosystem which I think we're we're we're all aligned on here but fear revenue is still obviously critically important as well um just shifting gears a little bit here to uh something um with regards to interoperability so that's been a huge topic of discussion within the layer 2 ecosystem um with the ethereum EOS system more broadly uh because of the fact that right now ethereum doesn't feel like a single unified ethereum it's a it's a it's a fragmented ecosystem a lot of work being done to bring it all together but there is one critical point here with with blobs is that when you have everything unified under one umbrella when you have the settlement and the uh the data on on the one chain here interoperability becomes a lot more seamless a lot easier and it's a lot easier to reason about and you can just basically say well I know how my data is secured I know when how long it's available for I know where it is I know my settlement guarantees whereas if you split that up you don't have to worry about two different things so you have to basically say well my data is over here on Celestia is the perform is something going to happen to my layer 2 if something happens to Celestia and then inability becomes harder because maybe the other players who are using EA don't want to be interoperable with you or don't want to play friendly with you because they're like well you don't you know you don't in our club you're don't in our cool kids club now so I'm wondering like could that be a driver the interoperability driver could that be a driver to keep these layer twos On eth Da especially if they want to become things like based rollups no no all right so this is where I tackled like this is I think another reason I'm I people a lot of this is an engineering problem to my perspective right so like yes okay so security yes I get it ethereum security deserves a premium awesome cool world war 97 doesn't matter it's secured I think because of the natural incentives teams can resolve for that while maintaining costs by leveraging like I I'll touch interrupt as well like by building in redundancy right so like oh ethereum is charging 20 times per kilobyte to charge a blob versus like I could post that same over to I da plus a veil Plus Celestia simultaneously and then like now I have that information built through redundancy like what's the likelihood of of Celestia going down I don't know probably pretty high relative to ethereum what's the likelihood that Celestia plus IA plus whatever is down like and then like you can build in some some safety there where like you need an N of and like to pull the same data from all three make sure it's the same cool it's good data whatever it's like I feel like that's a potential outcome that teams can build for and oh saves US money awesome like as a business that's great and then same thing for interop like that interop is the harder thing for me to attack but I think ultimately long-term Horizon my own head for where l2s resolve I've I'm kind of stfi Strat statisfying putting into different stratospheres right there's going to be a totally abstracted layer for people who are not onchain natives not people in this room right we're because we constantly want to be interacting with our chain directly with a wallet whatever the the interop situation is always going to suck right it's going to suck because we're we're so in tune with what's going on and then there's going to be like so for us there is no awesomeness and like yes interop is great still annoying I want a the Simplicity of a single state environment so like that's why salana is is gaining a lot of uh a lot of tracks right now CU like I don't have to worry about and I think that's going to ultimately happen at the L2 landscape where these Ultra high-capacity l2s just provide Simplicity through single state environments and then you have the total other Spectrum with like based app chains because they have access to L1 State and composability I think that's going to be a good selling point something like ens chain I think that should be a base chain that'd be sweet um so I think like that's where it's going to go into that whole Middle Ground whether it's interoperable or not I think there's business incentives to like don't be interoperable and I think yes like cool you can do some uh some interrupt because we should have a shared da layer but I also think it's an engineering problem to where like what if you have a huge validator right you're pimp this thing out and I have like I'm running a side car for like I have a full node cool running a side car for Celestia I got Doom running in the back I got like I'm doing everything this thing is sick and now I can potentially have it's an engineering problem interrupt between between non like Celestia right Celestia and ethereum blobs and so like long enough time Horizon why isn't that a possibility that people end up doing it's like how is that a premium for ethereum in that World um sure um yeah I'm going to push back a little bit um in generally you're right um so one of the big so there uh for a lot of these interoperability Solutions you you rely on third parties or you rely on solvers or someone to do things for you and there are ways that rollups can basically interrupt with each other natively so there's a um a rollup Improvement proposal 7755 and that's where rollups can basically read what happened on other rollups based on the data posted to layer one um and if there are a bunch of rollups that uh want to work together natively you have uh you can basically move assets between using the system uh if they're like stage two ZK rollups they're basically as secure as ethereum um and um when uh this happens if you have like say an ecosystem of a few rollups they're all using on ethereum they're all ZK rollups they'll use this to interrupt someone wants to join them and and wants them to interrupt but they're like actually I'm using Celestia I'm using some other da solution um it kind of gives them pause because then when if that other rollup has some issue um it affects them as well like it affects um basically the assets that have been um bridged across or they don't know if if they're doing some kind of say mint Burn System um with assets between these other rollups they're interrupting with um you could have like a double spend problem um things like that like this is very very low risk um but it is a risk if you're it matters if you're moving millions or billions of dollars like it doesn't really matter that much if you're any of a few thousand and yeah yeah it's great that we found a topic of disagreement um yeah because so I'm I'm I'm with you I think it's a bit tricky right because you you could for example ask the question if you've been a user on one of those um l2s that have been settling on on L1 and and also have been using L1 da for a while you're probably very familiar with paying high fees right you might have paid a dollar in the past for the L2 transaction I certainly have um and you might ask what did you actually get for that right like you could have at the same time just went to one of the side chains we have also many um great side chains on ethereum and paid sub Cent and the question is did you get anything that was worth a dollar there and that is a tricky question because it's not obvious that you did it's not obvious that you did because often times what we're doing in ethereum and it's an opinionated choice is that we are designing for the eventual end State and then we are basically willing to pay some extra cost to basically shoot directly for it and go go to for it as as fast as possible because alternatively you could imagine to say well you're on ll2 for now it's a multi anyway might as well not use onchain da might as well not do a settlement right like let's let's bring the this on later once we remove the the U the multis but the problem is the problem is and that's maybe where the dollar was worth worth it after all right is that um if you do not build for the end game from the beginning it's going to be very very very hard to later on cost correct we've seen this a lot with with um side chains there's there have been quite a few side chains on ethereum that have expressed the intent intention to eventually move to unchain da none of them really have done that so far one or two I don't want them but like by by and large they've not actually done that so far because it turns out it's actually quite hard and same thing with like you know like the the um the different stages for for l2s that l2b tracks so well and like um actually if you don't for example say um a distributed um sequenc uh set right if you don't actually build for robust sequencing from the very start it turns out that even say even the the easier version of that which is just like a high throughput um onchain Inox queue right for sensorship resistance even just building that becomes a huge challenge for you because you just build the entire system with the assumption that you have one sequencer that is trusted right and so so basically at all steps of the way if you build for what's expedient today the price you pay is that you don't have you it's not clear that you have a path to where you want to go three four five years from now right and and we're doing this at the L1 often times like often times I think think of any of the conten topics on a one over the last 2 three years it basically always was an example of people being like why are you doing this thing that I don't like today and then we are saying well because we want to be at this point in 5 years and it is a trade-off right like the the um uh fractured L2 ecosystem right like why did we do this well we could have not not gone with the rollup C road map and maybe actually today we would be in a better position but I promise you 3 four five years from now we will be in a better position with the path we took today right and so I think this same will basically be true with sorry this was a very longwinded answer to go back to blobs now but basically the same is true in a way with blobs I actually do not think that L1 da matters so much today but at the same time if you think that because of that you can make the choice of not using it I think you really are making a very dangerous choice because I think over time it will matter more and more and more and more I think you did a great job explaining like these this differences in terms of bridging trust this bridging especially once we go to ZK and there's a lot of upside once it becomes the trust bottleneck which Today just isn't right but once it becomes a trust bottleneck it is it's going to be more and more important to be on the same da Network on and to have that be the same network that you also settle on and so the only way to guarantee you'll get there is by doing it today so I don't know this was maybe a bit of a weird answer but I really think that's like a tension that's at the core of a lot of these like long-term decisions in the entire ecosystem I think this is a common theme within the ethereum ecosystem that yeah a lot of people kind of I guess struggle to understand this path dependence that we talk about right because ethereum is such a large ecosystem it's so decentralized any decision that we make today does have a path dependence to it and if it's not the right decision today and then in 5 years we're like oh well we got to do something else now cuz we made the wrong decision beam chain yeah it becomes a whole thing right as we've seen with with with ethereum many times so yeah and and we don't know right as an I think you mentioned earlier we don't know what the right decision is we're making bets we don't have the 100% right answer uh but we're just trying to you know research the right answer then do some implementations do some experimentation around it and if we have to change paths do it quickly right because if we don't there's that path dependence Network effects get built and then we're stuck and we this is across all of ethereum it happens across all of ethereum now the maybe the the last question which I think is going to be another another big question uh that a lot of people ask is okay we have blobs you know l2s are scaling it's all well and good we've got a high priority placed on blobs as as part of like upgrading them with uh increasing their their counts and their targets and Maxes and also pias is a huge upgrade for it but should we also be scaling L1 execution that's a huge question and there is debate about how much we should scale it versus uh how much we should be scaling blobs uh I would love to get each of your opinions on the priority of scaling L1 execution versus blobs do you think blobs should be should remain the the first priority and we do L1 execution slowly over time or do you think more emphasis and more priority should be put on L1 execution scaling I I I like John charo's position of like L1 execution scaling is not isolated or or separate from scale l2s right there are like you look at the eclipse team you look at like a lot of these teams talk about what we need with L1 execution in order to increase our performance single slot finality like the interoperability we've been alluding to with shared blobs also comes from increased execution I also think there's a little bit of mics that's like of using eth directly I think the moneyness of eth kind of Fades the further away you get from main chain like it still exists because you have a lot of l2s but those l2s have optionality right they can switch from the gas token to whatever and then like you have abstraction on top of that it's whatever so I think eth should should continue I think ultimately the way value occurs in a modular ecosystem is you have to continue to force or scale your execution environment your state to be the hot State up to your level of comfort with like preserving the properties that you want we obviously decentralization is a huge thing that that ethereum prioritizes that's great have that discussion I don't think I don't think you should you should stop that too soon and you should attract a subset of use cases to be preserved directly from onchain users on your chain we talk about high value D5 stuff he's talking about billions of dollars moving around like you want like tokenization of treasuries like you want that on ethereum like you want some of that so I think I think ethereum should continue to do this stuff and by bringing value to your state specifically whether or not it's made for an L2 specifically will make those other l2s want to tie into ethereum and we Main maintain them wanting to tie of your network CU if ethereum doesn't feel like it's valuable State and for whatever reason onchain users would ever continue to to migrate to salana and salana becomes the the like their tvl moves over there then ultim ult like ultimately they they start to get on par with ethereum well the L2 can kind of switch over and then like they want to be attached to the best uh settlement layer I know John will kill me for saying settlement layer but like yeah I think they I think the priority should be on L1 execution and L1 execution is not uh indep dependent of scaling road map through through l2s yes yeah um so it's very important also for if L2 goes down um and you want to get funds out of there or if L2 starts censoring and you want to push transactions through currently use L1 for that um and if we don't scale it then that gets very expensive we had a time when the um arbitrum sequencer went down for a few hours and L1 was just clogged with transactions of people trying to push things through so we do need to scale there the problem is that like we can only really scale like 10x um maybe more if we zfi everything um so it's not a good long-term plan if we want to actually stay decentralized and have home stakers um and so scale it a bit um but uh longterm blobs are where most transactions L2 is where most transactions are going to happen yeah and to double to basically double click on that a little bit I think one thing that not always is quite as or focused on in this discussion of like you know ethereum versus Solana and actually I like salana you know this is not supposed to fate them or anything but like you can't just take the L1 and scale it to infinite users maybe buing some very fancy long-term ZK technology but like specifically because of the fact that you want everyone to completely trustless be able to verify the current state of the world chain right and so for that you basically just have to make a choice like what is the minimum requirements of a full note and the nice thing that you then get with the roll cic road map is with with this I would say it's I call it the execution compression trick of basically the L2 being able to they they they they able to run it like in principle a single L2 could run it like gigas or like even I don't know teras or something and then you compress that all down into like a very tiny easy to execute proof on the L1 and then you get the best of both worlds you get like super high throughput but then still the ability to verify with very little compute and so fundamentally there's a limit to how much the L1 can scale now we are not at limit so I'm not this is not supposed to be like an easy out here I actually I will say my personal opinion is that we deprioritized L1 um scaling a little bit too much over the last 2 three years um I think that has good reasons we like first the merge and then there was a lot of like merge cleanup and then also like it was just not so clear with the rollup setic road map what exactly the L1 is supposed to provide to the l2s so I think we are now we now starting to have a pretty good picture so I think now is about time to like commit and really build the L1 into the best L1 it can be for this world that we are starting to see more clearly but they are also real trade-offs so the one the one point that I would make is we should not just scale in a dumb way because again we're always trying to shoot for five years in the future if we just grow the state I think we've seen some other early clones of the ethereum chain that just ran recklessly with State growth and then they struggled a lot because of it we should not make the same mistake we should take the five year outcome into account there are good ways to scale I would call it Bally SC smart not just by boot Force I don't think the gas limit is the right mechanism I think we have a lot of things now in the pipeline that can actually give us a lot of extra throughput there is compute repricing like compute is just too expensive on ethereum there's no reason for it we have the throughput like we can just literally make all compute five times cheap on ethereum today and it would not be an issue like we should just do that I mean there's some nuances it's not like a oneline change but we can do that um storage we should be better at pricing exactly the right behavior that we want for example um if the same state is used multiple times to the block that should be cheaper because that is actually cheaper for the note that executes it um a lot of these kind of like smart waste of scale we are working on statelessness right statelessness is this big thing that I don't think to my to my knowledge like very few ecosystems right now do that very few chains do that which would be a huge unlock because then the people that want to verify this Global chain trustless Le do not have to have that state that keeps growing so that would allow us to potentially there are many other bottlenecks then but potentially scale State um access State growth by by by a big factor as well so my message would be yes we should scale the L1 but please do it in a smart way don't just run with the gas limit yeah yeah I think so I I I think we're all in agreement that we should scale both the l1s and the l2s but it's yeah that that tension about what priorities we should do and how long it's going to take and not to sacrifice certain parts of the ecosystem that that always exists uh so we've got a couple minutes left here I just wanted to quickly give Tim a chance to talk about his blob simulator tool that he built because this is a really good tool cuz we talk a lot about blobs but you finally put some data to it and you gave people a tool to be able to actually simulate what it looks like in the future as we scale blobs up as more rollups come online so I just wanted to give you a second to be able to to shill if you will what you built thank you um yeah so I love bringing data to these debates I don't like it when people just argue with hypotheticals um so I created this blob simulator it's open source on GitHub MIT licensed I'd love for people to create reports with it um play around report any bugs um maybe add add things like how much uh execution gas will be used by rollups things like that would be awesome um and uh if you have any ideas on like um more data that you would like to see that you could bring to the argument um create an issue on the GitHub as well or reach out to me on Twitter um and we can try and get that in there so we can actually have an idea of how things will look uh a decade in the future and figure out where we're going can I make a feature request for for a toggle where just like at this point the L2 off and uses a different uh data layer that's I went with that yeah sure great great well this has been great uh thank you very much to all the uh panelists uh for for joining me today um one actually last thing I should do because I don't know it feels like we're all kind of a bit too bullish uh right now we need to be a little bit more bearish I think I'd love just a quick rapidy from each of you your your most kind of like bearish case for ethereum blobs like what does the future look like if ethereum blobs kind of failed honestly so the reason I I think I've ultimately embraced the eth's money meme is because once you've accepted that world you realize it does not really matter right like it's available to whatever as long as ethereum can prove itself to be available whether or not teams use blobs like yeah it kind of matters but I don't think ultimately it matters on the surface and you ultimately just want these l2s whether it's just posting like proofs or if they're doing proofs plus da plus whatever you just want money in the hands of people and for them to denominate their assets in that so therefore you want to export it doesn't really matter if they use blobs a ton we're going to bicker about it because we're nerds and that's what we do but like to the if for the scale that we want ethereum to be it does not matter uh longterm frankly yeah um the most be case would be if we try and rent seek or try and make as much money as possible now and think really short term um and don't increase the blob count um that would be a terrible idea I think yeah I mean from the inside I would say I don't think there's a big risk that we'll do it for R seeking purposes but I do think we could do it by not realizing how much of a priority it needs to be because it is a real engineering challenge to scale blobs uh there is work ongoing I'm very optimistic that it will get us far but it requires continuous attention and focus and so the failure case I think would be that if over time the ecosystem loses track of that focus and we we ship very lackluster throughput increases and over time that's just not enough for Al and they are forced to leave and I don't think anyone could blame them and basically we end up in just a world that overall is worse for users because they have less security yeah great so let's keep scaling blobs let's make sure we don't rent seek um and let's make eth money thanks everyone absolutely let's give our panelist and moderator a big round of applause thank you gentlemen all right now let's go straight to our Q&A we've got lots of questions coming in now we do have a lot of questions and uh if it's okay with with the gentleman on stage just share the questions among yourselves and let's keep the answers short sweet simple and straight to the point okay first question up on the screen blobs are great and so is Proto Den sharding when D sharding question mark yeah I guess I'll take the road map questions um yeah so basically I think D Shing was the initial way we described the end game for like where we want to go um then we realized that the best way to get there is actually iteratively so we will launch what's called padas uh in one of the next two upcoming hard works on ethereum and then that approach will probably iteratively get us more and more towards dun shotting I'm not sure we will ever never have a hard work that then is dun shotting I think instead your mental model should be ethereum will keep scaling the a throughput every hard for I I'm I'm trying to establish the meme 3x throughput increase every year for the next couple years all right thank you for the answer great question as well with six votes we have another question do you guys think that things additional sorry additional things like blobs will make it harder for everyday people to run nodes from home what do you think about that so the the meme answer is no because no one are going to use the things like no one's going to post data to them so you don't have to download so you're fine it's good no shot sweet you want to add something to that go ahead it will make it harder and we have to be very mindful about that but I hopefully we have good Solutions all right great discussion okay let's move on to the next question as well with four votes if Bob based fees go up on ethereum is it viable for elts to batch and post non-urgent transaction blobs separately optimally whenever the blob based fees go low I actually know of a protocol that's actually specialized in this specifically working with L2 team so that they can smart post all of their data and because that doesn't immediately impact users teams are opting to use it right they're saying like oh we'll delay this batch because like oh it's a little hot right now or could get a little hot and then we're going to post it a little bit later and that's ultimately what's building into my bare case of this stuff is that like L2 teams are like some are showing to really care and like f you we're using eth eth is money cool other teams are saying like yeah yeah cool we hang out for a little bit or like your scroll and scroll just said like we're not going to post because it's hot right now so like sorry guys you won't know because you're end users but we'll continue to do our thing but at the same time they can't wait forever like if you're in 2021 and you're waiting for gas to go down to 10U you're waiting for a year and um they have to be secured by ethereum at some point yeah sure exactly thank you very much let's move on we've got a lot of questions let's see how many we can get through what are your thoughts on eclipse or other execution engines settling on ethereum doesn't this solve a lot of ethereum's scaling issues yeah I mean they they don't use ethereum blobs so like yes they do they do settle there and I do like if you believe eth is money in these environments Eclipse Mega eth like if you believe that in these environments they're going to use ethereum in the hands of users as gas as denomination of assets of all that stuff then like solving the issues for scaling is like how do you make value go to my token like got my bags are going down I want my bags to go up how does value acrw like the the next tier of eth making new highs is people perceiving it as money and wanting to have access to cool and needing eth in their pockets to do that cool stuff and yes eclipse is doing that Omega eth will be doing that like Bas is doing that right they're making people need eth to do the thing and like that's you just need more and more of that it actually does not matter about fees directly but also at the same time I will say I do like that we're seeing a lot of experimentation with different execution systems um like svm based ones I think I don't think the evm is the bottle make in terms of throughput I think you can also scale the AVM quite far but I I I actually am very curious to see you know once we get to the efficient Frontier of very high throughput l2s which system actually wins out is e money a very good question to ask audience is e money yes thank you well I I'd like to hope all of you are internally agreeing thank you very much because you got to build stuff to make e needed all right Builders May avoid including blobs due to the higher latency they cause do we have a mechanism to incentivize them we do today and it's uh the tip so um the one thing we don't really have today is that uh because for Simplicity reasons we didn't give the blobs a separate tip so it's just the normal transaction tip which is a little bit annoying because it doesn't scale with the amount of blobs but so you have to like have some extra logic around that right now I don't think clients for home builders expose that and I think that would be a very easy fix right so basically give uh home builders the ability to specifically set a minimum tip below which they just don't include blobs and I think that would actually be the market driven solution because it causes extra effort so you might as well get extra minimum payment otherwise you just don't include them very good anyone else want to add no all good I would also appreciate tips thank you very much you can put in my hair I appreciate it uh anyway let's go to a few more questions we got about three and a half minutes on the clock our home stakers the bottleneck to L1 scaling what do you guys think yes yes and that's a good thing I mean like yes well that's that question answered well done anything else you want to add to that yeah but that doesn't I mean it's great that we have them like that's they a good bottleneck to have so that's that's specifically the bottleneck that we need to basically work around and make sure that we will forever scale within the envelope that makes it possible not by the way stakers I think people always too Focus too much on stakers and don't focus enough on full node Runners other than stakers in my vision for the long term everyone runs a full Noe or at least something close to full Noe even on their phones and so in the long run it must be easier not harder to run a full Noe for ethereum good point everyone in the future should be able to run a full Noe I think that's the pathway at the right way sorry you wanted to add something um they are not a um a blocker for uh L2 scaling though with um with 2dp with um blobs uh basically blobs are shattered amongst all the all the stakers and so if we have 10,000 nodes that's um the blobs can be shouted across Lum so even if we have 128 blobs per block you're not downloading 128 blobs every block you're downloading like a handful of blobs and everyone's downloading a handful and you're checking each other um and so we can scale l2s and blobs with home stickers excellent thank you we got two and a half minutes let's keep going does blobs or do oh excuse me it moved up where can I find blobs who's indexing all this data yeah so this is great actually so as people know blobs is actually uh they can be pruned from Full noes so that after a certain period no one has to keep them U but there are several ecosystem uh participants that are have committed to long-term indexing of all this information archival stuff so like you have lob scan which is a separate scanner just for that you have Dune you have ether scan all these people are preserving this data so yes it will home stakers I know you've already publicly committed to like storing all this information uh so there's multiple places I think I actually have a repository where uh all that can be seen I'll actually make that public if people are interested but it's blob scan is is the primary place for me all right blob scan very good do blobs help solo stakers save hard dis drive space um yes and no on one hand it's just another thing that adds extra storage requirements but it does take a lot of pressure of of call data which call data you have to store forever because it's part of the history we are going to change that in the future it's called Erp 4444 but for now you have to store history forever whereas blobs you can throw away relatively soon that's by Design um I think part of it is just qual is just too cheap today um we're going to make it more expensive but in the past we couldn't make it more expensive because we did didn't have alternative ways to do to use da now we have them so overall the load based from history on on stakers will go down all right we got a question about da since you mentioned it da is a commodity what characteristics can be added to e da that make it special in some way it's not [Laughter] special I'm just keeping it cheap and then because it's on ethereum you don't have any external dependencies that makes it special um yeah I think keep in cheap for now and over time it will be more and more important that you actually use L1 da in the meantime maybe we can mint nfts for the people I don't know all right I think we got 30 seconds let's see if we can fit one last question the biggest and the longest question how's BG feeling about blobs burn e doing 25 million a year run rate in the past 24 hours more generally which kinds of l2s will be willing to pay how much for blobs before migrating to alt da to save I wish I understood the question but I'm sure you guys do maximally aligned l2s right like it's the people that are that are fear the the repercussions from the social layer are the ones that are going to stay there um that's a good question and it would be good to find that out and add that to the simulator of when they would leave at what price all right ladies and gentlemen we've just finished our Q&A time let's give the gentlemen a big big round of applause for the amazing session thank you guys we welcome you down from the stage really appreciate your time all right that was what