Ethereum Scalability Panel

It's a great opportunity for me to have this conversation, thank you Bankrat, I hope you guys are going to enjoy it, we're going to explore some of the more recent advances in how Ethereum researchers and leadership think about scaling Ethereum and I'm Bartek, I'm from L2Bit, I'm like a sceptic, like a blockchain sceptic I guess, and we both have been in this field for so many years now and frankly I've never got a chance to ask this question so before we go into something more serious and I want to keep it fairly technical for the rest of the interview, can you like kind of tell me how does it feel like to have one of the scaling features of Ethereum named after your name, the dark sharding, like where did the name come from, what's the story here? It certainly feels very strange and yeah, I tried for a long time to come up with a backronym for it so that it wouldn't officially be my name but unfortunately it didn't succeed and yeah, I don't know, it was more like an accident of history that some internal name from our calls became kind of the official way people started referring to it. It feels like this is maybe a mischievous way to keep it in the space, now you can't leave right? It's like you're Ethereum. So, you guys must have remembered the block size wars back in 2017 with Bitcoin that essentially resulted in the split in the community, so before I go into Ethereum scaling, were you a big blocker or a small blocker at the time?

Well, I wasn't actually really that deep in crypto or part of any community so I would not say that I had a strong opinion back during that time and yeah, I would probably also say that I'm kind of philosophically a small blocker but pragmatically a big blocker and to explain what that means, I respected how the Bitcoin community decided to enforce their will against the miners so I think that was a great milestone in blockchain history but on the other hand, I would probably have said pragmatically, also at certain points in the community, I was a bit bigger. It just seems to me a little bit, when I was following the discussions on raising the limit and this is like almost the core of our argument that you came as an extreme big blocker, almost right? And there was some pushback from the community, from some members of the community, however that was nothing like we witnessed in Bitcoin, it was short-lived and it seems like now the community seems to be rallying behind raising gas limits, like what's your take here? How do you see this? I mean the difference in the situation is that we actually have a story on how to really keep the values that we care about, which I think are really about verifiability and censorship resistance, which previously, like the story used to be, well you have to have like these small nodes that you can run at home and we can now achieve them without making these kind of like the center, like tenet of Ethereum, like I think we should really care about the properties of the system that we're building rather than like, I guess the inner workings.

So I've heard the following arguments from, let's call them small blockers, right? The main one essentially goes about the current gas price, this is very, very low, at least for Ethereum, it seems like the users have moved to L2s, which is what we kind of wanted in the original roll-up centric roadmap and now the ETH price has gone down, so maybe now everybody's concerned that the L2s have grown too big, right, and we don't have a good way of pricing this growth and every sort of existential threat that they might be too big and Ethereum too small, I guess, that they might break off eventually, right? So is it all about the ETH price, like if ETH was 10K, you know, that wouldn't be a concern or there's nothing else, because frankly, I'm not entirely clear like what we're trying to achieve here, are we trying to bring users back to L1, and I'm thinking short-term, right? Like let's put away the discussion of the KEDMs and whatnot, like short-term, are we trying to get all these users from L2 back to L1 to pump the price of ETH a little bit back up, or are we trying to get users from Solana, like what are we trying to do here? Yeah, right, so I guess like we have to see, I mean it's a complicated question, I think like it comes to like, it certainly, there certainly is like an economic question in the background, that like a, like it's, it does seem much harder, for example, to like maybe like five years ago we imagined, oh that like everyone wants to be on Ethereum and therefore like they pay whatever for the Ethereum data available, so it doesn't really matter whether they own L1 and L2, and that has just turned out not to be true.

And maybe we were a bit naive in that. So there is an economic question about this, but I do think that the economic question also ties back into the Ethereum project, even if you see it from a pure cultural and values aspect, because I do think that, for example, at least the situation where there's one breakout roll-up that basically captures a huge amount of economic value, while the Ethereum L1 doesn't, that would fundamentally change the relationship between the two and would actually lead to a world where we would have less ability to like pursue our values. So I do think it is good to keep the economic relationship in check as well, so I do think it's better to have like a strong L1, and I think this is right now the best way I see how we can achieve that. And how strong L1 will attack the ETH price? We'll see.

It's hard to predict, I mean there are a lot of things going on there. So I think the question that everyone probably keeps asking, I mean this is by far the most different to that. Now that the focus seems to be on scaling L1, obviously everybody is wondering how does it really affect L2s, and maybe before you answer, because you get to answer this question probably every single day, I'll just throw two quotes. One is probably quite known from Anatoly from Salana, like he says, my job here is done, I don't have to make fun of the L2 roadmap anymore, but you know, I mean he's from Salana, right, so maybe he's not the guy to listen here, however, two days ago, Synthetix, which is one of the oldest DeFi protocols, and one of the earliest adopters of L2, because they were the first protocol that actually went live on Optimism testnet, if you guys don't know that, it's just like a, you know, Ethereum scaling history of sorts, they actually wrote in their blog, L2s are dead, the experiment is over, and Synthetix is going all in on L1, right, and I don't know, I mean that sounds to me like very harsh words, so I wonder, I mean maybe they were always extreme, right, like you know, maybe they were too early, and now they're like, I don't know, going too far? Yeah, I mean it's a complicated question, I mean like the very direct way to see it is just like the way we, and it wasn't necessarily that we had to do it this way, but the way we started doing roll-ups made them be in some ways like a kind of direct competition with Ethereum mainnet, and for like a long time we were like okay with it, because we were like there's no other way of doing it, and now the question becomes a bit more complex, but I don't think that has to mean that, it definitely doesn't have to mean that we have to be like enemies or so in any way, right, I think like, yes exactly, like there are many good ways in which we collaborate, I think for me the big difference is just that I would probably in some ways agree with Cain here, not in all ways, but I would agree with him that I would like to have his project on L1, and like there's nothing wrong with that, with us saying like we would like them to stay and focus on it, that doesn't mean they can't also be on L2, but like I guess like we will have a bit more of an active role of like also developing the L1, and I think it's, there are probably things in DeFi where we would see them like a little bit more like critically, and we were like yes, we want lots of liquidity on L1 and not all of it on L2s, and for other things like we would be less concerned and would be like oh yeah, like there are applications where we're quite happy if they continue being on L2s, their scale can never be provided by the L1, and so like they can continue being there.

Maybe the issue was that, you know, we had one tech for building L2s, but you know, different competing visions of what L2s were actually meant to represent, like sovereign zones with you know, really good internal interoperability, but not necessarily external, like you know, very similar to let's say commercial banks settling on the central bank, right? We don't use central banks to keep our accounts, you know, the actual commercial banks keep their accounts on, so Ethereum was supposed to be like the settlement layout, so that was one vision. The second vision is like this Model C, where you know, tens of thousands of roll-ups, you know, like little tiny building blocks, impossible for L2B to track, that's to me a completely different vision. The third one was that, you know, they should be like an extension of Ethereum, right? Like interoperable to the point that, you know, it should be all seen as one huge shard of chains, and like to me these visions were like different, and maybe there was no guidance from Ethereum Foundation, like what are we actually building here?

Well, I mean, the Ethereum Foundation has taken like in the past a very hands-off role in the whole thing, and I think it's shifting a little back, so it's probably not going to go all the way the other way now, but I would say like more closer to the ecosystem and more actively involved again. So what is your personal vision today for roll-ups specifically, like, and can you also comment on this, you know, this idea of the execute opcode and how does it actually fit into this vision? Oh, okay. I mean, two very different questions, but they're kind of related, right? One is more short-term, the other one is more like medium or long-term.

Right, right, right, right, yes. Yeah, I mean, I think like, I don't know if I have like a grand vision for this. For me, this is more like, I mean, it's like, it's not that I have a grand vision for other chains or for apps or whatever either, so like I think like it is a natural thing to exist and it's great, and like we can coexist and collaborate in many ways. There are probably some things where it would be like it's a more natural fit and there will be, I guess you can say, like less of a competitive dynamic if like the roll-ups go more into like the app-specific direction or into applications that are not tied to DeFi and so on. Yeah, that's how I think about it at the moment.

Okay, so let's switch gears a little bit and talk about scaling data, data availability, because this is how we started and, you know, having Dank shouting my name into it, you know, is a certain responsibility, and if anyone's interested in scaling VMs, EKVMs, that's going to be like a different conversation, right? So, data availability. Celestia has announced that, you know, they have on testnet 128 megabytes blocks, you know, they're trying to build to gigabyte blocks, and it seems like we're so much behind. What's your take? Like, you know, I mean, I remember your talk from Amsterdam a few years ago when you said that Dank shouting, you know, shouting generally of data should be the top priority.

I mean, you're certainly not happy with where we are, are you? Yes, I am. I mean, okay, so there's a complicated question. There are several layers to this. Like, one is that, yeah, I do believe that we have been not ambitious enough with Ethereum in the past, and we have to realize that we have to, like, have the somewhat consistent offering with the market, otherwise, like, the pure, yeah, extreme sort of cypherpunk concepts are not going to sell.

On data availability, it's especially hard, because, so with Celestia, I would ask them a question back, like, how are you going to get sampling for, like, a gigabyte per second? Because if you have a few thousand sampling nodes, this means you're talking about many megabytes per second of sampling. Yeah, but Ethereum will get more nodes, right? More nodes, yes, of course. Like, if your vision is to have millions of nodes, that's great.

But, I mean, that is also a very, very hard thing to achieve. Like, I, so I think, yes, there are many questions about this to be answered as well. I don't, I, and I think this is a real question that everyone as a data layer, for example, will have to ask themselves where they see themselves on the spectrum. And I actually would say I don't have the answer even for Ethereum right now, because traditionally, yes, we have done the very conservative route and said, okay, like, we only want to do data availability if it has these very high security guarantees. But then the question is, like, is that actually valuable to anyone, what we're building?

Like, this is unclear at the moment. Yeah, and the conservative almost feels to me like this is a big bet from Ethereum that the people will care, right? And they will actually pay for it. And they will prefer to use Ethereum as a settlement layer because, you know, it's like decentralized after Bitcoin, but you can't do much on Bitcoin, right? So, you know, like institutions should, like, use Ethereum almost as a default.

However, this is not what I see, you know? I mean, Ethereum is dominating, yes. But I think the issue is more political rather than, you know, tech-oriented. And I'll be personally very concerned, you know, if it's like more lobbying for, let's say, Solana or for something else. I mean, how do we convince, you know, the big stakeholders here, and by them I mean institutions, that Ethereum is the best place for, let's say, RWAs, for example, right?

Because it's like the most decentralized. Do they even care? And if they don't, then, again, what are we building? Well, I mean, I think, yes, this is like kind of where you have to find the middle ground where what you're offering actually works. And I do think we can build a lot of things there, but I think it is hard to find that ground.

And I think, like, on data availability, for example, we still have to find it. We still have to find what is actually the right offering. We have started. I think what we have now is definitely not good enough. Like, we need to build a lot more.

But what exactly? That is, I think, a big question for the coming years. I think, like, surprisingly, on the L1 scaling, it was a little bit easier to actually figure out, because I think we can see where the technology is going and what we can probably do in a few years without, like, too much compromise, but also what is a good enough offering. Like, for example, I am quite confident that 100x scaled L1 will do a lot for users and will be very valuable for the world. But on the data question, I think there's a lot to figure out right now.

So when I, like, look at Celestia, EigenDA, and Ethereum, I mean, I see Celestia trying to be as much decentralized as possible with the novel P2P network built, you know, for supporting those MAMOV blocks, as they call it. On the other hand, EigenDA is, like, you know, cranking the volume as much as they can by being centralized and saying that it doesn't really matter so much because, you know, they comes from. So it's a very, very different, like, set of trade-offs, right? And it's like you said, I mean, do we know exactly what trade-offs are we pursuing with In other words, can Ethereum support as much, you know, block space as Celestia? Like, technically, is there any reason why Ethereum cannot in the future, you know, support that much DA?

Or that would, like, go against some values of whatever they are? Well, I mean, like, technically, like, again, the question is, like, what do you, what trade-offs are you willing to make? For example, like, yeah, right now, the most fundamental, ultimate limit is how many nodes are sampling. And I guess we can try to get a lot more sampling nodes, but that will also be very hard work. And I think, like, we can do things there.

And I think, like, what we're doing now, also with ScaleDL1, we'll push in the right direction because we are adding much lighter nodes to the whole system as well, so we can have more of these. But I think, at least from what I've seen so far, pushing to millions of nodes is going to be very tough. And then how does this sort of compare to EigenDA, where, you know, you use, you don't really sample, and I think you just, like, disperse, you know, chunks, and that's it, and there's no peer-to-peer network at all. No? No peer-to-peer network, right?

Right, right, right. Well, I think, actually, the peer-to-peer network is not really the critical part in making it, like, verifiable by a large number of nodes. But the sampling is. I do think they want to do that, and I do think they have... Yeah, it's a plan, but, yeah, yes.

Okay. I was asked for five minutes for the audience, so ask questions from Danekard, but before we do it, you know, I started on the light one. I'm going to end on the light one, because we could go, like, really for hours, you know, going deeper and deeper, and this is always a great opportunity, you know, to be able to chat. But my final question to you is, like, why did you stop blogging? You know, I really appreciated your blogs, and I think, you know, they were great explainers for the community of hard concepts, right?

Like, people are so confused about, you know, like, what does it really mean, you know, the censorship resistance, the 51% attacks, and whatnot, not to mention the famous Casey G. explainers, and everybody that I talk to say, like, yeah, I mean, I've read, you know, Danekard's posts, and finally I got it, you know? And I checked, and it's been three years, Danekard, like... Yeah. Well, actually, I don't know.

I, yes, I had... So, I'm, like, a predatory community coach, you know? Like, please blog more, you know? Then I think my suggestion would be, if there's something where you would really like an explainer, then hit me up, and I will think if that's something. I mean, it used to be more natural, and I think, like, I guess for me right now, we're in a very different stage in some ways, but it has become harder, so there were fewer topics where it felt naturally, but if there are things that pop up where you think, this would be a great topic, then, yeah, just hit me up, and I will think if I can write a blog about it.

Okay, let me give it to the audience. We probably have time for, like, maybe two questions, right? Yeah, we have about five minutes. Yeah. Okay.

Any questions for Bartek and Danekard? Hi. So, I have this question... Let me hear it. Hi.

So, I have a question about why focus on scaling DA, right? Because we see, at least last I checked, that the blog price is essentially zero, so it seems like the demand is met. Is it the fact that we're anticipating more demand? And essentially, how it plays out with the focus on L2s being more application-specific, will they then use more DA? What's the vision?

So, I don't think that present mainnet demand is actually a very good way to measure this right now, because we do know that there is a substantial amount of latent demand that's not coming on Ethereum blocks, because, like, for example, L2s like Sallow, simply, like, already when they made the decision, they looked at it, and were like, well, there's no way we can go there. Polygon, which might now have changed strategy, I'm not sure, but we were also talking to them, like, a couple of years ago, and they were like, well, we can't, like, put it on blocks, right? So, there's definitely a question of, like, which projects can even commit to being a roll-up, given the current throughput, and it's not a lot, right? Many of them can't. It's a very big commitment, because you're basically saying, I can never have more than these transactions, and, like, by scaling it more, we are losing that.

What we can also, like, achieve is simply, like, I mean, it's a similar question to scaling the L1, right? Like, if I want to build a project, and that's especially a big project, like a roll-up, I'm going to do a lot of BD, I want to know that there is capacity for future growth. So, I think, like, right now, the answer is, like, what we have right now is simply not good enough. Like, it's almost meaningless if you want to build a big project. Like, you simply cannot commit to build it on ECMDA at the current level.

So, I think we definitely have to go away from there, no matter what. So, for me, if I can, like, add one thing, you know, if I see, let's say, on-chain games using LDA, I'm not concerned. If I see apps like Celo using LDA, I am concerned. You know, there's really no reason why they should use LDA, and I guess I'm not going to satisfy them if they can use, you know, just for the security, right? I mean, we're talking people's funds, and we're talking projects with very significant DL are using extremely insecure LDA bridges today, to the point that, you know, at L2B, we're actually making this effort of trying to, like, push them really hard, you know, to use a good DA bridge, but we found that this is actually hard.

It's a hard task to convince people that this is important, because then, you know, questions come up, like, who's running this DA bridge? You know, who's the relayer there, right? Who can upgrade? You know, what is the upgrade delay of this bridge? I mean, all sorts of, you know, like, Pandora box of questions that no one—and who's paying for this, right?

Another question for the panel? If not, then we'll wrap up here. Can I have one more? Can I have one more? Yeah, of course.

Please. Like, what are the main points of disagreement with you and Vitalik? Ooh. Because, like, the reason I'm asking this is that, you know, I heard rumors that they are some. Rumors of?

That they are some. There are some points that you disagree with, Vitalik. No, just kidding. But, you know. We had—well, I don't know.

I think, like, once we discuss, we can usually, like, come to an agreement. Like, we had recently some discussions. They're public as well. You can see it, like, about provers. Like, do we need to be able to run provers at home?

And, like, I said, well, it's, like, a nice long-term goal, but it's really, like, not the thing we should be most concerned about right now, also given that, like, the technology will advance massively. And I think, like, I had more discussions recently with him, and he was like, oh, actually, like, he's come around to that, and I think it makes much more sense to just check that there are lots of alternatives to get the GPUs that you need rather than insisting on it being at home. So, I don't know. Like, yeah. I don't know if there are really very big disagreements right now.

That's good to hear. I mean, we don't want these guys to really disagree, I guess. Okay. Thank you so much. Thank you, everybody.

I hope that it was useful. Please, a round of applause for Bart Shatner. Thank you.