Refuting L2 Fud & State of Arbitrum | Steven Goldfeder

Is it true that maybe they're taking transactions off of Arbitrum? Yeah, probably, but that's okay. We're growing this pie together, and together we're all Ethereum.

Welcome to Banklist, where we explore the frontier of internet money and internet finance. And today we're back on the frontier of layer twos with Steven Goldfetter from Arbitrum. Lately, after one of our recent podcasts, a flurry of conversations has happened on Crypto Twitter and in various other crypto conversation channels about the Ethereum roadmap and the nature of the relationship between Ethereum Layer 2's and the Layer 1. People are kicking the tires of the Ethereum roll-up centric roadmap just to double check that this is the right choice. We're on the right path. Some hot takes were shared, some responses were made, and today on the show, Steven from Arbitrum will clarify some of the technical elements of that conversation. The decentralization of Arbitrum sequencers, whether Arbitrum is competitive or cooperative with the Ethereum Layer 1, and whether or not we should increase the capacity of the block sizes on Ethereum Layer 1. In addition to all of that conversation, there's also just a ton of cool new gadgets coming out of the Arbitrum universe. Stylus is allowing millions of Rust, C, and C devs to become smart contract developers, along with all the code that's already been written in those languages. Turns out there's a ton of cryptography that's written in Rust. And now with Arbitrum Stylus, that cryptography can become smart contracts, which I think is pretty cool. I hope you all enjoy this episode with Steven. It's pretty jam packed, so strap yourself in. But first, a moment to talk about some of these fantastic sponsors that make the show possible. Especially Fankless Nation, once again here with Steven Goldfetter, co founder of Arbitrum. Steven, welcome back to the podcast, my man.

Thanks for having me. Great to be back.

Interesting time to be in the layer two space. I know Arbitrum has been building a ton of stuff. We're gonna talk about bold, stylus, time boost, many of the cool uh gadgets and bells and whistles coming out of the Arbitrum ecosystem. Uh, but first, also, I think there's a big conversation going around in the Ethereum verse as to like a reunderstanding what's a layer two, uh, specifically in the context of how uh intimately attached a layer two is to the layer one. Uh Steven, I know you've been like kind of tapped into this conversation. I want to start things off there. How do you think about L1 to L2 relations? Uh are layer twos equivalent to the layer one? Where is the line between a layer two and a layer one? What's real Arbitron's relationship with Ethereum? Like, overall, can you start this conversation off for us? How do you think about this uh this relationship here?

Absolutely. I think that they're very symbiotic in that Arbitrum is Arbitrum today because it has a very powerful layer one, that layer one being Ethereum underneath it. Without the powerful layer one underneath it, Arbitrum wouldn't be able to have, say, its fast block times, its low fees, the high level of security that it has. And this is a very symbiotic relationship here. Arbitrum cannot exist without Ethereum. And frankly, Ethereum cannot reach its full potential without layer twos like Arbitrum. So there's a very strong symbiotic relationship between the two. And the two complement each other very well. You know, there's this uh it's like a tweet form. I think I was the first one to do it, but now I see it all the time. Like Arbitrum is Ethereum, base is Ethereum, and obviously it's it's making a point. Um, but I think it's a true point. It's it doesn't mean they're equivalent, right? If I say, hey, I live in New York and you live in California, we both live in the United States. And that's okay, right? There can be different uh islands or different states or different parts of the same larger being. And I think that's the case here. You have different layer twos, you have Ethereum layer one, and together they make up this larger um entity that is Ethereum, and they're all kind of critical to uh Ethereum being and reaching its full potential.

There's a one perspective out there that uh a dub for Arbitrum, a win for Arbitrum is a win for Ethereum. You know, Ethereum is one big ecosystem. Any dub that is earned inside of the ecosystem is a dub for the ecosystem as a whole. That's that's like the the layer two growing the pie perspective. Um maybe the opposite perspective is uh you know, each layer two is maximalist of itself first. You know, Arbitrum is about Arbitrum. Uh if Arbitrum doesn't win the users of the layer one, the assets of the layer one, the liquidity of the layer one, then another layer two will. How do you how do you think about this tension? Because I think like both can be true. Uh and so how how do you think about that that tension? Because all layer twos are in competition with each other. Uh, and I think one perspective is that if all layer twos are in competition with each other, it kind of converges as like trying to take away the uh the user's liquidity assets that exist on the layer one. How do you think about that?

Yeah, I don't I don't see that tension there. I don't I don't think that's really true. Now it is the case that obviously the layer twos are somewhat competitive with each other. And, you know, I prefer that more uh activity happens at Arbitrum because we contribute to to Arbitrum technology, and someone building uh optimism or base is probably going to rather it has there. But actually, I think he, you know, a really good place to look back is Vitalik's 2020 roadmap, the eat-centric roadmap, this blog post that has become like, you know, this uh really important blueprint for the community. And we followed it quite well where where Vitalik actually says this as a feature. And he says, you know, a lot of people basically don't want to have a little level of influence over a large thing. They actually want to have, you know, a project or a sort of a piece that they control. And he uses the analogy of these little islands. And the fact is that these little islands may be in in competition, they may have, you know, different uh objectives and also um, you know, prefer themselves over their neighbor, but there's no, they're still part of a larger entity. They're still part of Ethereum. And that's basically, you know, how I see this today. So, yes, a a win for for Arbitrum is absolutely a win for Ethereum, and a win for Ethereum is absolutely a win for Arbitrum. And that doesn't mean it's not like, you know, contradicted by the facts that there are different layer twos. A win for base is also a win for Ethereum. And it may not be a win for Arbitrum, but actually if you view you zoom out and say, well, we are part of the same, you know, Ethereum ecosystem, I think that there is a viewpoint that says that too. That might be a little bit extreme, but still, a win for arbitram and a win for any layer two is absolutely a win for Ethereum because we have to zoom out and say, like, you know, two things. Number one, who is the competition here, right? So you have the same people that on the one hand are saying, hey, you know, L2s shouldn't, you know, give grants or shouldn't compete against in the verticals that they think should belong on the layer one, and then saying, hey, why is everyone going to Solana? Why is Solana seeing all this activity? How do we compete with them? It's like

There's your answer, right? The L2s embrace the L2 vision. Um, it's part of Ethereum, and L2s are the scalability uh layer of Ethereum, and that's where these apps can go and can thrive, and that's how we compete with other ecosystems. And the nice thing is, because we have Ethereum, we have the security as well. Because Ethereum is underneath us, we have a massive advantage over these other monolithic ecosystems that can offer the cheap fees, but can't offer that together with the same security and same decentralization that we have. So I think really it's important to embrace that vision. And it is true that at times, like, yes, it's possible, you know, a user might come uh from Ethereum to Arbitrum that would have otherwise stayed in Ethereum, you know, but it's like if you open up like a new highway, a new lane on the highway, it's true that a car that was in the original lane might go there. But the reason you're opening up the new lane is because you have a vision of there being enough capacity to saturate all of these lanes. And that's really the view that I have for Arbitrum and Ethereum generally, which is, you know, if we're stuck in time today and we can't imagine any growth, you might say, okay, we're all fighting over this same pie, but that's not really what's happening. We're all trying to grow the pie and we're trying to create the future capacity. And yeah, it's true that projects might move over to Arbitrum that would have been on Ethereum otherwise, but together we're creating so much space for projects to be in Ethereum. And I don't care if a project, you know, stays in Arbitrum or stays in Ethereum today. What I want is Ethereum to have the capacity for the next wave. And I know that if we don't open up the layer two lanes today, if you will, we won't have that capacity in the future. And we need to start that today and not worry about, oh no, are we, are we taking some some cars off of this lane? That's okay. We're creating more capacity. And the thesis here is that we're going to fill that capacity. And if you don't believe that and you think that we're just, you know, we're just fighting over what we have today, then I think that um, you know, you need to dream bigger and you need to really um have a much better vision of what we're gonna build because that's not too exciting if we're stuck where we are today.

Steven, there's a tweet here that I want to share that I want to get your take on. This is coming from a Zero X Bread, who I think has been really like nerd-sniped about this like layer one, layer two conversation. And he has this graphic about like use cases and where they exist in like the Ethereum sphere. And so he has like the Ethereum layer one as like illustrated as kind of covering the maximally decentralized, super secure end of the spectrum where you find sovereign ownership over assets, DeFi, and maybe you hold your NFTs on the layer one. And then you have the layer two, uh, which is like buying coffee with like stable coins or like game asset ownerships, maybe also some NFTs. And I think the the idea here is that like you'll allow the layer twos of Ethereum to do the low value, high frequency, high throughput activities. Uh, but then you like the use cases that are meant for like the super secure uh use cases like DeFi and sovereign ownership over bare assets is left to the to the layer one. Uh but uh he's illustrating attention here where actually layer twos on Ethereum can do the full spectrum. Layer twos can do both sovereign ownership over like, you know, bare assets. It is it can be, in the fullness of time, a place to store your CryptoPunk. Uh, but you can also buy your you know your daily coffee with USEC on Arbitrum as well, uh, which like kind of begs the question like, well, if the layer twos can do everything uh and there is this like inter layer two competition, what kind of is left for the layer one? Uh how do you think about this like tension between use cases on layer twos versus use cases on layer one?

Yeah, so I I think this chart, you know, this how it should be here, like

That's their vision, and that's that's a beautiful vision. It's not my vision. Like, I don't think that there's this arbitrary line where we have to draw that says, hey, L2s shouldn't touch DeFi, for example, which is what what this line shows. And by the way, it's not Vitalik's vision either, right? So if you go back again to that 2020 roadmap and you look in the comments there, he actually makes a comment where he says, I think that he says DeFi will be the one that drives this all, because he says the DeFi apps are going to want to remain competitive for users and other fees, and they're going to go on and and drive this layer two narrative. So that's a wonderful vision that this person has. Um it's not my vision. It's not Vitalik's vision. Why is it not my vision? I don't know why you draw that arbitrary line and say that L2s can't do DeFi. DeFi users also want cheap transactions. And I think you actually, like, let's go into DeFi for a second, right? Let's take, you know, a simple DeFi protocol. Let's take Uniswap or any, you know, um AMM and actually compartmentalize their users and not just put it in this one thing that says DeFi, but says, who are these users? Well, there are the whales there that are trading, like, I don't know, a million dollars at a time. And then there's a user that's trading $10 at a time. And to put those people in the same bucket and say that, hey, you you both need to use Ethereum, the fees are okay for you, is ignoring like probably the larger vertical, at least by people, by population, right? So I do think that there are users and are use cases out there that say, hey, I'm transacting at a certain volume and I have a certain comfort level that I actually want to keep on layer one. But to say that, like, okay, your DeFi user who really just wants to trade, you know, $3 worth of a coin, um, like, who are we to say that they're not important? In fact, they are important. And in fact, they're probably, in sum, a majority of the users out there. And those are important use cases too. Take another example, take derivatives, right? Derivatives didn't exist on Ethereum before Arbitrum came along because the transactions were just too expensive. They they just wasn't able to support that, right? The you know, GMX thrived on Arbitrum first. Uniswap and the like, you know, Ave were on Ethereum and on Arbitrum and other layer twos, but the derivatives actually rose with layer twos because we were enabling a sector of DeFi that literally wasn't possible. You know, fast forward to today, uh, Renegade Finance went live last week. This is a product that offers dark pools on Ethereum, uh, on Arbitrum, and it's made possible by Arbitrum stylists, right? So you can't do this on Ethereum. So who's to say that, like, you know, even if you had this notion that there are certain verticals, I think that the buckets that we drew there are just arbitrary and wrong and cut in the wrong place. You can talk about value. Like CryptoPunks, I don't see a good reason to bring your CryptoPunk to Arbitrum, to be honest. Like, why would you? It's it's doing well on Ethereum. It doesn't have, you know, really anything that you can do with a utility, like in a gaming world today. Maybe one day there will be and you can mirror it, but that's not doesn't exist today. So I'm not here telling you like bring your cryptopunk theorem. It's perfectly well on Ethereum. It's a status symbol, and that's where it should be. Uh, but at the same time, I think we have to be a lot more nuanced about who the users are and who we're pricing out and who we're actually including by having layer twos. And then the choice becomes, by the way, okay, do we want to include these people in layer twos?

Right. The people aren't saying, so so to take one step back, people aren't saying, oh, Steven or Brett or whoever it is said that like I shouldn't launch my DeFi app on Arbitrum. Okay, fine. Pay high fees on Ethereum. That's what Steven and Brett say. No, they're going to say, all right, let's go to Solana. Let's go somewhere else where we're going to get these fees. And then they're going to have a worse experience, a less secure experience, and they're not going to have that Ethereum experience. So, right, we're not we don't actually have that level of control that we think we have. What we want to do is say there's a need in the market. There are these users, there are these $3 DeFi traders. There are these users that want these products. How can we give them the Ethereum experience? How can we give them the best possible and the most secure blockchain experience at their level? And I think layer twos are the answer to that question.

I do really like this idea that DeFi is actually strictly better on Ethereum layer twos than it is on the Ethereum layer one. I think the evidence that uh derivatives taking off on layer twos is a great example of that. And also just like the congregation of like the $5 DeFi trader and the $5 million whale speculator being in the same spot is actually generates welfare for everyone. There's just efficiency when everyone's in the same place. And that's going to require a platform that can handle more people doing more transactions on a daily basis. Uh, and so like uh the natural evolution of DeFi moving on to layer twos, I actually think is a pretty strong point.

Yeah, and by the way, there's you know another thing which is related. There was an interesting paper from Uniswap by uh I think it was by Austin Adams on their team that showed the um the effects on LP efficiency um when you have like faster block times. So Ethereum block times, and I know the shortest will come up, are are obviously significantly slower, but even amongst L layer twos, Arbitrum has you know 250 millisecond block times on Arbitrum one and the

Fifty milliseconds.

Yes.

So four blocks a second.

Exactly. Yes. And the arbitrage opportunities, you know, the arbitrage are the efficiency, the capital efficiency leading to like faster arbitrage are just, you know, much more. It means LPs do better, you know, they they collect more fees, um, and it just leads to more efficient DeFi as well. So there are like a host of reasons. And and and again, you know, it I think the false, you know, uh, the false sort of the economy that people put here is like, okay, just do it on on layer one, or like, oh, maybe Arbitrum will just do this all in abandoned Ethereum. And the answer is those two things go together. Arbitrum can't. Arbitrum could not offer what it's offering with the security that it's offering if it wasn't for Ethereum providing a critical service. And I think the, you know, the good analogy here is the internet, right? So you have this, you know, TCIP, this stacked architecture. And the internet is probably the most successful protocol ever of all time. And, you know, the what one what one is this stacked architecture. And it's not, you're saying like

You know, oh, we should do everything in this level in a second. No, it's the idea is that different layers of the stack are benefited by layers underneath them. And that's exactly what we're building here. It's uh in Ethereum and an arbitrum, which is Ethereum is critical. It's a critical layer of the stack, just like TCIP is a critical layer of the stack. When I'm ordering food on my phone in Uber, I don't necessarily think about TCIP, but we wouldn't be able to do that if it wasn't there. And I think that's important as well that people sometimes get like, you know, we're we're we're zooming, zooming, zooming in and saying, how is it possible that Ethereum isn't everything? And the answer is Ethereum doesn't have to be everything. Like, look at the internet. Ethereum is is making everything possible at its layer of the stack, and it's okay to build on top of that as well.

I think uh I remember asking, I think I wrote this tweet is like it does MEV increase or decrease with block speeds? And uh Tim Ruffgarden, I think, responded to it showing with some like very technical paper that's like well beyond my capacity for understanding, but the TLDR is the faster the the block times are, the less total MEV there is. Uh and so it's it's not just like obviously there's also less MEV per block because the blocks are sooner, but faster block times also just decrease the total aggregate amount of MEV, uh, which would go so far to say that just like the faster you can speed up a block time, uh, the less total MEV, the less total value there is extracted there. And I think 250 milliseconds blocks is like the fastest blocks that exist. I don't think there's a faster blockchain.

So the only faster ones are actually, so to cannibalize ourselves, the only faster ones, so the arbitrum orbit orbit stack is actually able to go even faster. It can go up to 100 milliseconds. And a bunch of the public chains run at 250, but a bunch of arbitram orbit stacks actually push in and go to 100. So yes, other than that, it's it's you know, I think Solana is the next closest one, and it's at 400 milliseconds.

Okay. Okay, so then I think that brings us to the question if DeFi is strictly better and it defyes just move into layer twos, uh, then the decentralization or security of layer twos also becomes a high priority of a conversation point. Um can you kind of just like frame frame this conversation for us? The decentralization of transaction ordering, decentralizing the sequencer of Arbitrum. This conversation is a little bit of a meme. If I say decentralized the sequencer, that that's kind of a meme statement. Maybe you can kind of like unpack that question for me and talk about like on the grand list of priorities for Arbitrum, where is that?

Yeah. So the answer is decentralized in the sequencer is important. It's a priority. And it's actually something which, you know, we at Off Chain Labs are collaborating together with a team called Espresso Systems on building a decentralized version of the sequencer. And it's uh going really well. But let's like zoom out a little bit and talk about what is decentralized in the sequencer, because as you said, it is definitely a meme. And

Pretty loaded.

yeah. And like people like really don't know what it means. And actually, because you know, there's good reason for that. So

In a blockchain like Arbitrum, in a layer two, there is something called the sequencer. And there's also something though called the validators. And I think these are two roles that people often confuse. And so let's talk about a transaction, the life cycle of a transaction. So I send a transaction to Arbitrum.

Um

I could send it directly to Ethereum, but generally I'll send it to the sequencer, and the sequencer will do what its name suggests. It will sequence that transa transaction and set, I will include it in this sequence, in this order, you know, your transaction will be included there. And the sequencer sort of gives this fast promise. And remember, the sequencer actually is then going to go ahead, you know, it's going to post these transactions, which call the batch poster is going to post it on Ethereum. But the nice thing about the sequencer is it gives you a promise that's faster. It gives you a promise that's uh even faster than Ethereum, which you can, you know, rely on to the extent that you're comfortable trusting the sequencer with that promise.

Now, what happens from then on? Like let's say the sequencer

Included a bad transaction. Or the sequencer said, like, hey, um, take money out of David's account and put it in Steven's account. Like, what would happen if the sequencer did this thing? And the answer is, um, on a chain like Arbitrum that has operating fraud proofs, it would be rejected as invalid. Because what happens is there's another uh role called the validators, and the validators are the ones that actually execute the transactions and say, is this valid? Is it not valid? So the sequencer sequences them and it says, here are the transactions, and the valid validators go ahead and run those transactions and said, okay, you know, David paid Steven, or oh, okay, actually, this transaction that the sequencer included isn't valid. It doesn't have a signature. David didn't pay Steven, so we're going to reject this transaction. Now, typically for efficiency, the sequencer isn't going to do that. It's going to, you know, exclude the bad transactions. But the point is, when it comes to a trust and security, we're not relying on the sequencer for security. We're actually relying on the validators. Okay, so what is the issue then? Why do you need a decentralized sequencer? Or in other words,

Before we go into decentralizing the sequencer, can I ask, well, who are the validators? Because if the validators are also doing the checking for us, uh, who are these people? Because then now uh now all of a sudden I'm trusting that they're doing their job correctly, right?

Excellent question. So today on Arbitrum One, there's about a dozen validators that are on this allow list. And that includes like relatively well known entities such as off chain labs, uh Google, it includes uh the Ethereum Foundation. So it includes like, you know, and all you need to do but do, by the way, is trust one of them. That's the nice thing about the validators. Only one of them has to be on

Honest.

Yes, it's like the single honest validator assumption that you've, you know, that that phrase, that's what that means, as long as one of them is honest. And by the way, this is made possible by Ethereum again. How do we get around the typical consensus result that says you need to have a supermajority? The answer is we have Ethereum. And therefore, layer twos can do their consensus with only one honest validator in their layer. Because what does that honest validator do? It appeals to Ethereum and relies on its consensus. So again, this really staged layer approach where uh layer twos can have that. But that's not the goal and that's not the ideal. And actually, it's a good time we're having this conversation because Arbitrum Bold is a brand new um protocol for uh the challenge and fraud proofs that's going live hopefully soon. Um, I would expect it to be voted on in the next few weeks, um, you know, to be put up for a vote in the DAO in the next few weeks. And when that happens, validation will open up for everyone.

Okay, so we have we have 12 sophisticated validators who have uh the roughly the who have like the whitelist to be able to uh do this like kind of sophisticated job of making sure all the transactions are valid. That's been the state of things, and now that's going to governance vote to get opened up to the whole entire world. So the whole world can become one of the people that says, hey, that val that transaction wasn't valid.

Yes. And the nice thing there, and by the way, it's not actually that sophisticated a thing. It's running a node basically. And right now, if you run a node and your node says

that's invalid, you know, this the whitelist will say, okay, now you can actually go on chain and and challenge that, put down a deposit and challenge that. So that's going to be opened up to everyone to do it exactly.

Okay. And then now who's the sequencers? Because the sequencer is a single box, right? So there's one operator, the sequencer.

Yes. So so this is yeah, a great question. So the sequencer today is

There's two questions here, which is who's a sequencer and who gets to choose or change who the sequencer is, right? So those are two related questions. So the Arbitrum DAO, in the case of Arbitrum, gets to choose who the sequencer is, but today it um it gives that responsibility over to the Arbitrum Foundation, you know, to choose uh um who who who who actually runs the box of the sequencer. But I think that's actually a really interesting point because I think this is something that people miss. Right. One of the criticisms of the current centralized sequencer is the sequencer gets uh all this value, right? And that's number one. And therefore they're gonna have a disincentive to decentralize.

Right.

Now, I want to like clear two two misnomers here. Number one is there's this other meme called sequencer revenue. And I don't know if you think that exists or it shouldn't exist. Or at least an arbitram it shouldn't exist, but probably uh I think this is probably true for most chains on Ethereum, uh for most L2s.

It's chain revenue, right? Like, you know, it's like, you know, to give an analogy, you have someone at the you know, collecting, you know, you had a you're at a theater, someone's collecting tickets at the door, and you say, okay, this is the ticket, ticket man revenue. No, this is like the revenue of the theater. This is just the person that's collecting that revenue, but like this person doesn't get to keep the money that they collect. They're they're the front person for a much larger operation. And it's the same thing actually for the sequencer. We fundamentally internally don't use like the term sequencer revenue. It probably exists in certain places in documentation because you know it it uh it's kind of pervasive in the community, but at least my preference is the term chain revenue. This is revenue that belongs to the chain. Now, the chain has to also appoint a sequencer. And the cool thing is that today the answer to that second, that first question is who gets to choose for the sequencer? It's already the Arbitrum Dow. The Arbitrum Dow gets to choose. So even though the Arbitrum Foundation um is actually running, you know, it's in charge of running this box of the sequencer, it's not getting the profit. All that profit is going to the on-chain treasury of the Arbitrum DAO, which means there's zero economic incentive not to decentralize because you can basically, you know, uh the DAO can say, hey, instead of you, I'm actually want this committee to run this consensus protocol to decentralize the sequencer. We know how to do that. And it's something which we're building together with espresso. And by the way, the time boost protocol, which is our uh favorite protocol for um for how the sequencer operates and how it assigns priority, was chosen to be easy to decentralize. And that's literally the one that we're working on to decentralize. So um literally why it was designed. Um but the point is there's no economic disincentive because it already belongs to the DAO today. Off chain labs doesn't get sequencer revenue. Uh the Arbitrum Foundation doesn't get sequencer revenue. It all goes to the Arbitrum DAO, and they could just as easily say to it, and just as you know, you might say, hey, instead of setting up one toll booth, we're gonna have, you know, 12 toll booths, you know, a distributed toll booth. The toll booth collectors are still not the ones that are making the money. It still goes to the port authority, if you will.

Let me see if I can uh uh steel man uh a response to that. So uh with a single box, a single centralized sequencer, even if it is governed by the DAO, and even if all of the revenue of the sequencer goes to the um goes to the chain, goes to the DAO, there's still the potential of this one single sequencer uh censoring a particular user or something. They could still do censorship. So in the grand roadmap of like the generalized Ethereum layer two space, there's always been this idea of having a more multiple sequencers in order to get around this like uh censorship concern. Because if we have multiple sequencers, one of them starts to censor someone. Well, there's other sequencers that are not making that same choice. And so therefore that user just has to wait until uh a new sequencer uh puts a new uh transaction into the chain. Uh and so this has been like the kind of the uh the grand idea of like a decentralized sequencer. Uh and then I think it this opens up the economics conversation of like, well, if there is multiple sequencers, does that invalidate this idea of like chain revenue because one of the handful of multiple sequencers can keep that revenue for themselves because uh because now they are in control of sequencing the transactions. Maybe they take some MEV and actually give less money back to the chain. I I think that's maybe a uh a steel man of a rebuttal. Like, uh tell me if I'm wrong or how I'm wrong, or or what's your response to that?

Yeah, so to the first thing you said, like absolutely fully agree. That is the actual incentive to decentralize the sequencer. There is a centralization aspect or a censorship aspect. I want to get back to that in a second because there is already a lot in place that actually makes the censorship not as bad as you might think it is. But yes, that is the reason why you'd want to decentralize the sequencer. Um to the second point, I think, you know, um you have to sort of step back for a second and say, what does it mean to decentralize a sequencer? You know, before we think about how they can economically uh potentially manipulate it. And there are different models here. So there are some models that say, like, and this is not the model that that Arbitrum uh is doing today, or it's, you know, it's on any roadmap that I'm familiar with, but there is a model that, you know, what's called Miva, you've probably seen MEV auction, which is basically says, hey, auction off the right to be the sequencer to the one who's able to pay the the most. They'll go ahead and collect that MEV. And then sort of, you know, this person, it's kind of like you have a single sequencer at any given point in time, but then over time, uh, that that rotates, right? And that helps you from a censorship perspective because you can say, hey, even if this current one is censoring me, you know, in 10 minutes or an hour or whatever it is, like I'll be able to, or in the case of a base roll-up by the next block, I'll be able to now go to the new sequencer. Like that, there is obviously a censorship resistant benefit there um to that. Now, is it possible that now this individual sequencer will, you know, even if it's off-chain, set up some sort of thing that says, okay, you gotta pay me some fees to be included and start trying to collect, you know, out of band revenue? Of course, uh, it is possible, but they're limited by the fact that you know they don't have um, you know, sort of um, you know, a monopoly on this until you can just wait them out and wait for the next one. But the type of sequencer decentralization actually being considered in the arbitram ecosystem is a bit different than that model.

It's

Going to be a distributed committee that sequences, right? And therefore, um, you know, the the property that you'll have is it's not that like there's one sequencer now and one sequencer in an hour, and one sequencer an hour later, and therefore that's how you decentralized. It's actually at any given point, it looks like there's one sequencer, but it's actually a consensus protocol that's uh backed, you know, by uh who knows, say a couple dozen nodes or or the like. And maybe these nodes are voted on or chosen by the DAO, um, who can actually participate in that committee. And and that, you know, I think it's it's a lot harder to uh to actually um, you know, for them to try to uh collect revenue. No, sure, sure, if you manage to like break the key uh majority assumption of this protocol, then they can they can do those things. But it's gonna be much harder for one or two people to sort of you know out of band do things because they're just you know a node. And as long as the consensus property and the honesty property is is uh is respected for this committee, then it's gonna do the right thing and it should be you know cartel proof and and you know you know the ability to re to resist such attacks. And that's the model that we think about. So there's really not then one person that can insert themselves and try to do that. Uh and also, by the way, again, remember the DAO will have control over this. So if there is evidence that someone is doing something and trying to uh you know do things that are shady out of band, then they will hopefully be replaced by by the by the DAO and they won't be part of this process anymore.

Right, yeah, the uh the the Dow boot as the like the last line of defense.

Yes.

Uh can you tell me what you guys are up to with uh Espresso? Because I think that the first time I actually ever heard about Espresso uh was actually talking to you at ETH Denver 2022 or something very, very long ago. Yeah. Because this conversation is related to like Arbitrum's actual plans to like, you know, quote unquote decentralize the sequencer. Tell us about Arbitrum's plans to do this and and how does Espresso work into this conversation?

So Espresso is building a marketplace for sequencers and a really interesting product as their core product. But also we are working with them as a research collaboration, research and development collaboration. You know, primarily they're doing the development, but we're contributing heavily on the research side. We actually build a, you know, kind of what I said, a decentralized version of Time Boost. So just for two seconds, what is Time Boost? So another thing that people don't often discuss with nuance when they talk about decentralizing the sequencer, it's like, okay, we've decentralized the sequencer. Now what would you like it to do? Would you like this decentralized sequencer to do first come, first serve ordering? Would you like it to do MEV auctions? Like there are many, many, many ways to decentralize the sequencer, and they look very, very different and they have different properties, right? You might say on the one hand, like, I want to do a first come, first serve auction, and I want this decentralized sequencer to guarantee that this fairness property is happening. On the other hand, you might say, I actually want to do an MEV auction. I don't care about any sort of like first come, first serve, fairness. I just want to make sure that this decentralized property is making sure that this auction happens properly and the chain is getting maximal revenue. Very, very different things with very, very different objectives, but they're both still somehow decentralized in the sequencer. So time boost is actually uh kind of in the middle of those things. It turns out that for users, like average users of the chain, they're just playing their game or doing their DeFi transaction and just want to get their transactions submitted relatively quickly, they want something more like first come, first serve. Doesn't necessarily need to be exactly first come, first serve, but they want fast inclusion.

MEV searchers really want the ability to pay for ordering and say, hey, I want my transaction earlier than or or you know the first in the in the next block or something like that. And what Taiboo says is it kind of gives you the best of both worlds is it guarantees fast inclusion for everyone, but it inserts up to um

About 100 to 200 milliseconds of delay, where someone can buy a priority lane ticket and say, I want my transactions included, you know, with priority ahead of you know in the in the next block. So what you get is a really nice property because you get the average users are happy because their transactions are not going to be delayed more than a couple of hundred milliseconds. And the MEV searchers are happy because rather than having to pay Amazon or or whatever to try to gain a little bit of advantage, they can actually just buy that directly from the chain and the chain internalizes the revenue. And the last property of time boost is just like arbitrary today, it doesn't allow front running. So the sequencer sees the transactions first and it doesn't allow any sort of front run of front running. So you don't know that, you know, my you don't know my transactions before it's already been included. There's back running. I might try to get an arbitrage opportunity and get in really quickly, but I can't try to front run you um with any knowledge of what your transaction is. And that's where time boost is. It comes, so a time boost is a policy. Now, again, you just like first come, first serve or MEV auction, you could do centralized time boosts or decentralized time boosts. Um, what Espresso is working on with us is building a decentralized version of time boost. And it turns out that the time boost design was literally designed to be uh decentralized and it's very decentralization friendly, um, despite what some some uh may say.

Uh including what some may say. Some somebody may have said on some podcast somewhere that there was it was a pretty hot take that optimistic roll ups have captured Ethereum governance or are in the process of capturing Ethereum governance. Um uh what when you heard the statements, what what was your first reaction to that? Like, is there any merit to this at all?

No, there there's zero merit to that at all. And actually really it really bothered me. Um it was kind of upsetting because

A there's zero evidence. There's actually evidence to the contrary. So one of the bigger, you know, more controversial conversations of the all-call developer developer call in the last, I would say, year or so was this issue of 4844 versus withdrawal. So it seems like ancient history now that we've enabled withdrawals and you can, you know, uh take it, you know, uh take out your uh you know the ETU staked. But um there was a big push among many core development teams, and I'm not saying they had the wrong reasons, to be clear, to to actually tie these two together. And it was the Prism team, which was initially the only PRISM is part of off-chain labs, um, was the only uh core development team that was adamant and said, no, we

think that this is going to delay withdrawals. We think that shipping withdrawals should be our highest priority. Um we do not think that withdrawals should be uh should be tied to four eight four four and they actually uh moved four eight four four um you know to the to the next uh hard fork.

And

what do you say?

They moved it back, yeah.