The Ethereum vs. Solana Debate | Justin Drake vs. Anatoly Yakovenko

Primary Topic

This episode delves into the debate between the Ethereum and Solana blockchain ecosystems, featuring Ethereum researcher Justin Drake and Solana co-founder Anatoly Yakovenko.

Episode Summary

In this debate-style episode of Bankless, hosted by Ryan Sean Adams and David Hoffman, Ethereum researcher Justin Drake and Solana co-founder Anatoly Yakovenko discuss the strengths and weaknesses of their respective blockchain ecosystems. The episode kicks off with a discussion on the positive aspects of Solana and Ethereum, highlighting Solana's high throughput, low fees, and user-friendly experience, while acknowledging Ethereum's robust security and extensive network. The conversation then shifts to the challenges each platform faces, including Ethereum's transaction latency and the complexity of its virtual machine, and Solana's issues with outages and centralization risks. The episode emphasizes the significance of network effects, the potential for future technological advancements, and the impact of economic security on blockchain performance. Both guests provide insightful perspectives on the future of blockchain technology, the role of hardware improvements, and the balance between decentralization and performance.

Main Takeaways

  1. High Throughput and Low Fees: Solana offers high throughput and low transaction fees, making it attractive for developers and users.
  2. Robust Security: Ethereum's extensive network and focus on security provide strong guarantees against attacks and double spends.
  3. Network Effects: Ethereum benefits from significant network effects and liquidity, making it a dominant player in the blockchain space.
  4. Latency and Performance: Solana aims to achieve low latency and high performance through hardware improvements and fast block times.
  5. Economic Security: The debate highlights differing views on the importance of economic security and its impact on blockchain stability and censorship resistance.

Episode Chapters

1: Introduction

The episode starts with an introduction to the debate format and the participants, Justin Drake and Anatoly Yakovenko.

  • Ryan Sean Adams: "Welcome to Bankless, where today we're exploring the frontier of Ethereum versus Solana in debate style format."
  • David Hoffman: "Is economic security a meme? Is issuance a cost? Some big dividing lines finally get sussed out in long form on this episode."

2: The Good

Discussion on the positive aspects of Solana and Ethereum.

  • Justin Drake: "Solana has lots of obvious good balls in its bucket. It has very high throughput, very low fees, very good UX, low latency."
  • Anatoly Yakovenko: "What I envy out of Ethereum is the mainnet has a very large distributed network across a very large number of physical end nodes."

3: The Bad

Examination of the challenges each platform faces.

  • Anatoly Yakovenko: "EVM is really designed by academics. It's just a VM that no one with VM experience would have ever built."
  • Justin Drake: "There's no client diversity. There's no formal spec. There's no PBS. There's just general immaturity."

4: The Ugly

Deep dive into the more fundamental and potentially unresolvable issues.

  • Justin Drake: "As soon as you get close to one, even four is relatively close, you start getting these weird effects where even optimizing a few milliseconds gives you a crazy edge."
  • Anatoly Yakovenko: "The core benefit of what Solana does is global synchronization. And when you move stuff into L2s and pre-confirmation, you're effectively running a centralized exchange."

5: Future Predictions

Discussion on the future of blockchain technology and predictions for Ethereum and Solana.

  • Justin Drake: "If Solana does go forward, you know, towards its end game, where it reduces slot times to the minimum, let's say 100 milliseconds, then, yes, I think we're going to see hyper centralization."
  • Anatoly Yakovenko: "I think if the parameters are set correctly, that we have like 130 millisecond block times, you can have a validator anywhere in the world and vote, and you will get about the same API as anyone else."

Actionable Advice

  1. Understand the Trade-offs: Recognize the trade-offs between high throughput and security when choosing a blockchain platform.
  2. Focus on User Experience: Prioritize user-friendly experiences and low fees to attract more users to your blockchain applications.
  3. Stay Updated on Technological Advancements: Keep abreast of the latest developments in blockchain technology to leverage new features and improvements.
  4. Evaluate Network Effects: Consider the network effects and liquidity of a blockchain platform when developing and deploying applications.
  5. Decentralization vs. Performance: Balance the need for decentralization with the performance requirements of your application.
  6. Prepare for Economic Security: Understand the role of economic security in maintaining the stability and censorship resistance of a blockchain network.
  7. Adapt to Hardware Improvements: Plan for future hardware improvements to ensure your blockchain solutions can scale effectively.
  8. Mitigate Centralization Risks: Implement strategies to mitigate centralization risks, especially in high-throughput environments.
  9. Optimize for Latency: Aim to minimize latency in transaction confirmations to improve user experience and reduce front-running risks.
  10. Engage with the Community: Participate in community discussions and debates to stay informed and contribute to the evolution of blockchain technology.

About This Episode

Ethereum vs. Solana. That’s the debate we’re having today.

On Ethereum’s side we have Ethereum Researcher Justin Drake, and on the Solana’s we have Solana’s Co-Founder Anatoly Yakovenko.

They give us the good, the bad and the ugly of each other’s blockchain, settling their endgame towards the end of the episode.

This was one, if not the spiciest debate we had on the show. We tried our best to give both guests time with minimal intervention. At times went well, at others it caused deep semantic rabbit holes. But overall both guests were able to communicate their messages.

People

Justin Drake, Anatoly Yakovenko, Ryan Sean Adams, David Hoffman

Companies

Ethereum Foundation, Solana Labs

Books

None

Guest Name(s):

Justin Drake, Anatoly Yakovenko

Content Warnings:

None

Transcript

Justin Drake
There is a reason why Ethereum data availability is worth more than Solana's data availability. Because of network effects. Yes. This notion of shared security, this notion of synchronous composability because of the economic security, not just that, but also access to TVL, access to applications. There's only one ens in the world, right?

There's only one Ethereum. And Internet. This is really weak. This is why I think it's ugly and it's not fixable is it's not a fundamentally hard thing.

Ryan
Welcome to Bankless, where today we're exploring the frontier of Ethereum versus Solana in debate style format. This is Ryan. Sean Adams. I'm here with David Hoffman, and we're here to help you become more bankless. So on the debate track today, Ethereum versus Solana, we've got a conversation between Ethereum researcher Justin Drake and co founder of Solana and Natoli Yakovnko.

Now, both of these were previous bankless guests. I'm sure you know them from their respective communities. And the prompt for this episode was really simple. Give us the good, the bad and the ugly of each other's respective blockchain ecosystems. Take the conversation wherever you want.

You could talk about architecture, you could talk about economic security. And indeed the guests did. David and I tried to give both guests time to debate each other with minimal intervention. And I think at times this went well. At other times it probably caused us to go down into semantic rabbit holes fairly deeply.

But overall, I think both guests were able to communicate their main messages throughout this episode. Sometimes they were attacking, sometimes they were defending. Occasionally they were throwing each other compliments. It's a very entertaining episode overall. If you pay attention to Anatoly and Drake on Twitter, you'll know of a couple of topics that are coming your way.

David Hoffman
Is economic security a meme? Is issuance a cost? Some big dividing lines finally get sussed out in long form on this episode. How did this episode come to be? It was actually kind of spawned out of just a simple ping from a member of the bankless discord asking if it would be possible to get Justin and anatoly on a podcast together to have a debate.

And I was like, hmm, I also want this. And since it didn't really hurt to ask, upon asking, both Justin and Anatoly said, yeah, sure. And so after a bit of scheduling, here we are. I think the one thing I'll say about this episode, and generally, like crypto at large, when people enter crypto, they generally find a tribe pretty quickly, like you come into crypto, you learn about it, and then you learn about a layer, one that you like, identify with more than others because it aligns with your vibe. The how and why people become members of the tribe is like.

Personally, I think one of the more interesting things to explore in crypto, and you can really see it come out in this debate. You can really see some of the dividing lines between just who Justin is and what ethereum is and who Anatoly is and what Solana is. So that's kind of like a frame of mind that I encourage listeners to go into this episode with as they hear Justin and Anatoly parse out some of these things, finally in a long form conversation. So let's go ahead and get to that long form conversation. But first, a moment to talk about some of these fantastic sponsors that make this show possible.

So whether you are convinced by Anatoly and you want to buy Solana, or you're convinced by Drake you want to buy Eth, you can all do it on the same place, which is Kraken, our preferred crypto exchange for 2024. If you do not, how would I count with Kraken? Consider clicking the links in the show notes to getting started with Kraken today. If you want a crypto trading experience backed by world class security and award winning support teams, then head over to Kraken. One of the longest standing and most secure crypto platforms in the world, Kraken is on a journey to build a more accessible, inclusive and fair financial system, making it simple and secure for everyone everywhere to trade crypto.

Kraken's intuitive trading tools are designed to grow with you, empowering you to make your first or your hundredth training. Trade in just a few clicks and there's an award winning client support team available twenty four seven to help you along the way, along with a whole range of educational guides, articles and videos with products and features like Kraken Pro and Kraken NFT Marketplace, and a seamless app to bring it all together, it's really the perfect place to get your complete crypto experience. So check out the simple, secure and powerful way for everyone to trade crypto. Whether you're a complete beginner or a seasoned pro, go to kraken.com banklist to see what crypto can be. Not investment advice.

Crypto trading involves risk of loss. Are you worried about the security of your cross chain transactions? Cross chains with confidence. Using Transporter is a revolutionary token bridging app designed to give you peace of mind. Powered by Chainlink CCIP Transporter is your trusted gateway for securely moving assets like ETH, native USC and link, and so many more across some of your favorite blockchains.

Over $2.8 billion has been hacked from token bridges to date. Transporter puts a stop to this by ensuring your transfers are protected by the most robust security features available. Provides level five security backed by multiple decentralized oracle networks and an independent risk management network. Transporter also provides real time tracking throughout your transaction with its newly engineered user experience, so youll never have to second guess the safety or location of your assets ever again. And the best part?

Transporter makes it simple. Whether youre a blockchain beginner or a seasoned trader, transporters intuitive interface lets you execute cross chain transactions with just a few clicks, no additional fees, just a low cost for using CCIP, which can be paid in link or your blockchain's native gas token. But don't just take my word for it, see for yourself why transporter offers a stress free bridging experience experience the future of token bridging at transporter IO and just send it taking self custody of your crypto is one of the most important things you can do on your bankless journey. It's also one of the hardest things to get right, with huge consequences if you don't. If you want help going bankless, talk to Casa.

Casa helps you take custody of your crypto assets so you don't have to wonder whether you're doing it right. Casa is a one stop shop for doing self custody the right way. With Casa vaults, you can hold ether, bitcoin, stablecoins, all with one simple app and multiple keys for the ultimate peace of mind, with a support team to help you every step of the way. But it doesn't stop at self custody, because even though crypto is forever, you are not. We all plan on making life changing wealth in crypto, but with Casa's inheritance products, life changing wealth can elevate to generational wealth for your kids and your loved ones who don't know anything about crypto.

With Casa, you won't lose your private key, and you won't accidentally take them to the grave either. Click the link in the description to get started securing your generational wealth. Bankless nation we are extremely excited to tee up this conversation with two esteemed guests. We have Anatoly Yakovenko. He's the creator of Solana.

Ryan
He's the co founder of Solana Labs. He is mister consensus at the speed of light. Anatoly, welcome to Bankless love that intro. Thank you for having me. And in the other corner, I don't know if I should do fighting meme for this, but Justin Drake, he is a researcher at the Ethereum foundation and he is Mister ultrasound money.

Justin, welcome to Bankless. Thanks for having me as well, guys. This is going to be a conversation between some of the design decisions and differences between Solana and Ethereum. I think this is a conversation that the entire crypto community will tune in for. And it's great to have you both.

I think the way we're going to organize this is in a few sections. So the good, the bad, the ugly. First we're going to talk about the good, then we're going to talk about maybe the bad, the things that are correctable over time, and then we're going to talk about the ugly. And of course, we can veer from one section to the other as you guys see fit. But let's start the conversation with something hopeful, something optimistic.

The good side of things. I want to start the question with you, Justin. So let's talk about the good. What is good about Solana from your perspective?

Anatoly Yakovenko
Oh, boy. Long pause. So if I were to think. Sorry, I was just unmuting. I had to unmute.

Ryan
Sure.

Justin Drake
So if I were to have a metaphor in mind, I'm thinking of the good as a bucket with balls in it, and actually think of the good bucket as being the one with the most balls. So the largest one with the most volume. And Solana has lots of obvious good balls in its bucket. It has very high throughput, very low fees, very good ux, low latency. It has good traction, millions of users, it has good adoption, good financial performance.

Actually, that has been amazing. Right. We've seen Solana race to basically position number three. So if you go to Coingecko and sort by fully diluted valuation, Solana is the number three layer one. And I'm tempted to change the quote by Michael Saylor and say that there is no fourth best.

If you look at fully diluted valuations, there's a ripple, there's BNB, there's ton, but they don't really compete, in my opinion, with Solana. So it really is those three players in the layer one space. I think what Solana is doing for Ethereum is that it's providing healthy competition. It's a sort of accelerationism, and accelerationism, partly because of good execution, but also because of contrarianism. Like the way that I think of Anatoly as this constantly contrarian person, like his brain is wired in a very unusual way, and I say this in the best of ways possible.

Anatoly Yakovenko
My wife, this is her favorite feature about me. And I think you guys bring forward interesting insights and you have good vision. For example, I think that you want to build this idea of a global state machine. What does that mean, global? It means that a single l one can support the whole world.

Justin Drake
Like, in my mind, we're going to see l one s that can support 10 million transactions per second, which is enough for the whole world. It's 100 transactions per second per human per day. And then the other big idea that Solana brings to the table is this idea of synchronicity, that not only is there high scalability, but also different parts of the states can call each other synchronously, at least if you're willing to pay for it. I also think that Solana has very good pace and reactivity. There's been all sorts of issues in the past that we're going to talk about in the bad section, but broadly speaking, it just keeps on chugging along and moving along.

And in terms of the insights, I think there are real insights that, anatoly, you bring to the table. For example, the fact that we live in this exponential world with Nielsen's law, right? The fundamental bottleneck for scaling blockchains is bandwidth. And we are in this very luxurious position where bandwidth keeps on scaling exponentially. So overall, this is a massive bucket of good for Solana.

David Hoffman
Anatoly, your turn to talk about Ethereum. What is in the good bucket for Ethereum? I mean, tons of stuff. Like, I think what I envy out of Ethereum is the main net has a very large distributed network across a very large number of physical end nodes and very large number of actual Internet links. So when you look at it as a whole and you think of what are these networks supposed to do?

Anatoly Yakovenko
The mission that Ethereum has, it's supposed to provide security guarantees that are stronger than an honest majority. If you end up with accepting honest majority guarantees, which are fine for a lot of use cases, it's a multisig. At the end of the day, I think you just can't really scale it to support the world in a trustless way. Just nobody, like you can't have banks and 401 ks and human finance trust some honest majority that could blow up at any moment. Nobody wants that.

So you need better than that. And I think Ethereum has been so hyper focused on this that when they started working on sharding, they pivoted away from that because there's no way to do it, which is awesome. It's actually like the coolest thing to see somebody be like, you know what? This is not going to work because of the security properties we care so much about. We're going to work on this other roadmap first because that one doesn't compromise something really, really important.

And the execution for people think Ethereum is moving slow. But if you look at the size of the network, its value, I would be terrified of any changes to Solana if it was at that value. It's just terrifying for something to go wrong. Outages at that point are more, much more catastrophic, I think. So Ethereum actually moves pretty fast for a project that size.

The four clients is awesome. I think this is, again, the biggest thing that you're supposed to provide. Is this better than honest majority security guarantees? And when you have four clients, you're not trusting a single piece of software. Even with the best of human intentions.

It's still written by humans. Right. The original BFT paper actually talks about you need to have separate teams building stuff in isolation so you can actually have this fault tolerance across the human buzz factor. So there's a lot of envy there. And I think that's awesome.

From my perspective, Ethereum is at a point where to cause a double spend attack, you need NATO to coordinate across the thousands of different ISP's and to have everyone execute like a thing all at once. And that's. I don't think you can do better than that in the real world. So it's at a point which I think every other l one should be NVSF and trying to get there as fast as they can, right? As fast as they reasonably can.

All of that, I think, puts it in a class, I think, in my mind, actually above bitcoin. So for what that's worth, we're starting. This off on a very nice foot then. Both of you said nice things about the other's primary ecosystem. Anatoly, just curious, your reflections on what Justin said about Solana, do you agree with his characterization?

Ryan
Like, what was going through your mind as Justin was saying these things? Sure. I mean, I think I find those to be positives of what Solana has accomplished. I think probably because of my experience working at Qualcomm for a decade, I saw how fast Nielsen's law moves and how fast Moore's law moves when it's targeting an industry that's in hypergrowth and your brain as an engineer shifts. You're not writing for code for hardware today or yesterday even.

Anatoly Yakovenko
You're writing for hardware that's going to be out two to four years. So you're expecting just everything to get faster. So something that I've been, like, kind of hitting the table with to all the engineers that are working on Solana is you got to build things to scale as if the hardware is twice as fast. Just assume it's going to be twice as fast, how you're writing the code, how you're designing the SIMD's, and kind of leverage that. And I think that focus has paid off.

Honestly, that was a big bet that a small team, a very small team that had way less funding than all of our competitors took from 2018 to 2020 to launch. And we launched faster because we were like, don't make this complicated. Just throw a thread pool there. It takes work, right? It takes engineering, but it's much easier to think of problems like, okay, like, even look at proof of history.

It's not. I remember talking to Justin, like, a week after we became an official company about vdfs, and our solution was this very, very, extremely dumb approach. Just verify the VDF with as many cores as you have available to you, and don't worry about this fancy math that I don't understand if it's secure or not. And that worked. So stuff like this, I think, is underappreciated, and I'm glad that Justin has recognized that in some ways.

Ryan
So, Justin Anatolia has said some nice things about Ethereum. Said it would take NATO to disrupt the network and cause it double spend, and, like, maybe they couldn't even do it. Said, it's more. I forget the phraseology, anatoly, but, like, on a higher pedestal than bitcoin, let's just say. So what was going through your head as Anatoly was extolling the virtues and the values of ethereum there?

Justin Drake
Yeah. So one thing that Anatoly said, which I agree with, is that it's kind of bad design nowadays to try and rely on honest majority. And it is true that our previous sharding design did rely on honest majority, but this is something that we have moved away from just through technological innovation. So now there's this idea of data availability sampling, where you don't have an honest majority, you have an honest minority, and we have this new flavor of sharding, which is dank sharding. And not only have we removed the honest majority assumption, but we've also added synchrony.

So it used to be that we had a 1024 or 64 asynchronous shots that would each have its own independent proposal, and instead, now we have kind of this one big proposer that basically publishes a big block that everyone is capable of verifying, is indeed available using very, very little bandwidth. And this technique is called data ability sampling. And I think Anatoly had a podcast called no sharding and eventually decided to change the name. And I think this is good because the fundamentals are really in favor of sharding, or at least dank sharding, in terms of other things that he said is one, is that it's very, very difficult to double spend ethereum. Now, while I agree with that statement, I think this is kind of the wrong lens.

And the reason is that there is a much easier attack than double spend, which is basically around censorship and liveness. So anatoly focuses a lot, in my opinion, on safety. And to a very large extent, I agree that safety is not an issue. One, because the validity of blocks can be checked by full nodes. So if you try and just create an invalid chain, it will just get rejected.

And for specifically reorgs and double spends, there's this notion of slashing. It's just so expensive to do the attack that it's basically impossible. But there is these more subtle attacks around liveness and sensor resistance, which I think are the ones that are most important and I think are ones that Anatoly is under indexing upon. Maybe we can talk about this later. And another thing that Anatoly said is Simd, and maybe the listener might not know what that means.

So I just wanted to highlight, it is the equivalent of an EIP for Solana, and it means Solana improvement document. Beautiful. As we kind of wrap up this good section, I want to give also a chance for each of you guys to maybe elevate something that is unsung about your own ecosystem. So Anatoly talked about what's good about Ethereum, Justin talked about what's good about Solana. All of a sudden, this is really feeling like therapy.

David Hoffman
But we'll ignore that there's something wrong with therapy. David, we did not mean for this to feel like a therapy session, but nonetheless, here we are. But Justin, maybe you can advocate for Ethereum. What's something unsung about Ethereum that you think is definitely in the good category that is worth elevating? And of course, we'll go to Anatolia afterwards.

Justin Drake
Right? So one of my mental models is that the layer one space is winner take most. And so we need to ask ourselves, who is going to be that winner? What are the shelling points. What are the structural advantages?

And in my opinion, Ethereum has won in the sense that it is the place with the most network effects, with the most liquidity. It is the one that is the most mature, the most secure. And so that is maybe the best compliment that you can give to Ethereum is that it's preordained in some sense, as an Antoni mentioned, to flip bitcoin and become number one. And this meme where there's this guy who says it's not just about winning, but everyone else needs to lose, well, it's a little bit like that just because of market dynamics. I really do see a world where at the layer one level, the network effects are so strong that it makes it extremely difficult to be an alternative layer one.

And I think this is one of the struggles that Solana will have going forward. Anatoly, feel free to reflect upon that and then also just elevate something that you think that is unsung about Solana that deserves more praise. Yeah, I guess from my perspective, I kind of treat these systems as back end systems and developers are expected to learn and switch every two to five years like it's just part of life. So it's really, really hard to have a moat. And you see that with languages and databases and stuff like that.

Anatoly Yakovenko
Even though Oracle is still around, but makes most of its money through licensing. And people use redis and SQL Lite, and as hardware improves, SQLite becomes easier and easier to deploy for the kind of use cases that you would need, like Oracle, my sequel before. So you kind of have like a lot of tension and competition in the backend, even though being first mover is an awesome advantage. I think what's unsung about Solana, probably that I think why we're succeeding in a lot of ways is because the developers are not infrastructure focused. For the most part, they're focused on apps.

If you look at trying to get users, you need to give them something to use. That's fundamentally you need to provide value to a user. If you look at Solana, investment dollars are smaller than ethereums, but vast majority of them are going to product set or user targeting. You can actually, I don't know if this is true or not, I haven't done the numbers, but my guess is that actual product investment on Solana and Ethereum might even be compatible. So what's unsung is that how well the Solana devs are executing on actually user capture building products and stuff like that?

More so than people talk about Ethereum having network effects. I think most of those around in the infra side, which is just a piece that's not necessary if you have a high performance, low cost chain. So there, I mean, while I would agree with the statement that Ethereum culture to a very large extent focuses on infrastructure, historically, I think there is a shift now. So we have l two s, like base, arbitrum, optimism, where you just come in as a dev and you just don't have to think about infrastructure. It just scales magically.

Justin Drake
It just works. You don't have to worry about scaling about fast confirmations and things like that. And, you know, the metrics speak for. Themselves, actually, with no trade offs. Right?

So there are trade offs being taken, and in some sense, what these teams are doing is that they're making even more trade offs than Solana and being willing to kind of play the even longer game. In some sense, Solana is like fake it until you make it. So it comes in with like lots and lots of technique to look at that, and then is using that as a strategy for user acquisition. And then as there's a new fire that comes in, you kind of have to fix your technical debt. And in some sense, what the l two is all about is playing this game even harder.

So you have chains like base and optimism that don't even have a fraud proof, right? That's the game that they're willing to play. They're willing to fake it until they make it. They have a totally centralized sequencer. But the good news, the fundamentals, is that if you're willing to project ahead, there are ways to decentralize all of these things.

A lot of my personal research in the last six to twelve months has been around, for example, decentralized pre conformations. How can we have the 100 millisecond user experience that arbitrary bayes provide, but in a decentralized fashion? It turns out it's not that difficult. Not only that, in some sense it destroys the user experience of Solana. Because Solana has 400 millisecond block times, about 5% of the blocks are skipped.

And on the other hand, with a decentralized pre conformation mechanism that I suggest you have 100 millisecond preconfirmation. So we can have the best of both worlds. We can have the speed of light pre conformations and the decentralization. I think we have moved into the bad section, squarely feels out of the good section here. Maybe to just set up the bad section again, the good, the bad, the ugly.

David Hoffman
We've moved into the bad. Let's do it. Bad characterize as temporal the current problem set, whereas the ugly are things that are just like completely untenable, unrecoverable, final in nature. But the bad is just like the current problem set of respective ecosystems. And so we've now drifted into the bad.

Justin opened up the good section. So, Anatoly, maybe we can throw this one to you. Just the bad about Ethereum, we've kind of like touched on it a little bit. Justin does touch on it with talking about transaction finality just now. How would you characterize the current problem set, the bad side of Ethereum right now?

Anatoly Yakovenko
I mean, like EVM is really like designed by academics. It's just a vm that like, no one with VM experience would have ever built. So it has a lot of technical debt, like design debt. It's fixable, but it's like blood, sweat and tears to do that. And I think you're going to see very slow progress there.

And because it's so central, it kind of permeates to everything. The expensive l one and the split between l two s is kind of breaking a lot of the financial dynamics and splitting liquidity, which diminishes the value of Ethereum as an l one two years ago. So if you're now as an alternative l one, like Solana, you're not really competing with the Ethereum ecosystem. A developer has to make a decision. Do I deploy on eth l one, eth l two, or where do I put my focus?

I can't focus on all of these different environments at the same time. It is difficult. Right. And that means that Solana now being one aggregate layer might be an easier and simpler choice for a lot of devs. So in a lot of ways that's maybe fixable, but I don't know, that's a hard one.

And because of the split in terms of, you're going to have different confirmation semantics, different levels of fraud proofs, different levels of security. People will take those factors into account and they won't all decide to all go in one place. So it's going to be very hard to have one single dominant winner. But it's possible there'll be a power law distribution there. Are you kind of characterizing that as just developmental friction?

David Hoffman
Yeah, it's just kind of diving through quicksand. Exactly. But the quicksand is never ending. It's just going to get worse. This is, I think, kind of this.

Anatoly Yakovenko
It's both a strength, but it's a weakness. Right. You have a chronic copy of options and too many options create developer friction and create user friction at the top. I think maybe there's decisions that the l one can do on its own, but this is governance, and that's above my pay grade. This is interesting in Italy, because it feels like you're talking about the bad, but you're also bringing in some of the ugly too.

Ryan
In indicating that some of these things might be unresolvable, would you say that's true, or is that a step too far? I think they're hard to resolve. Okay. They're hard to resolve. These are hard problems because they require a lot of coordination and potentially sacrificing.

Anatoly Yakovenko
But the proof of work to proof of stakes, which sacrificed. Right. The Pow miners. So it's possible, right, to, like, it's possible to fix these, but it's hard. Justin, back to you.

Justin Drake
Right, so one of the things that Ana totally asked is, as a dev, should I launch on l one or l two? Well, I'll answer that for you. There is no future for the l one directly. The l one is a settlement layer. Now, a fair question is which l two?

And the reason why it's a fair question today is because the l two s are like silos. There's fragmentation on ethereum, but my vision is that we can fix fragmentation, and there's a whole bankless episode on that whereby the silos basically meld together. It's almost like what we're left with is the boundaries between a smart contract. You could argue, for example, that and uniswap being separate smart contracts are in some sense siloed. But really, you benefit from synchronous composability.

So synchronous composability is the strongest form of melding and removal of the frontiers. Now, it is true that it's going to take significant amount of engineering to get to this magical land of synchronous composability. And there's two pillars that we discussed in the episode. One is shared sequencing. That is actually fairly trivial technically.

It's mostly a coordination thing where we all need as an ecosystem to choose a sequencer that we will use as D shared sequencer, that can give us synchronous composability. And I have a specific thesis here, which is that the based sequencer, meaning the l one sequencer, the Ethereum sequencer itself, is the primary candidate to be that shared sequencer. And then there's this other highly technical thing, which is real time proving, and by that I mean real time snark. Proving we need to be able to make a transaction and a few milliseconds later have a snock proof proving that this transaction is valid. And so now the various l two s don't need to trust each other.

And so now you can send assets in and out of l two s essentially immediately. And they act like contracts at layer one where there's no longer this boundary. So yes, today, as a dev, you do have to make a bet for the short and medium term, but eventually ethereum will fix its fragmentation and will be one massive virtual machine. Now, one of the things that you said is that DevM is, for example, hard to parallelize. And there's just some cap in terms of TPS.

Anatoly Yakovenko
That's just one of its problems, like 32 byte register storage values. There's a whole bunch of things that are really fugly about it from a VM design perspective. Right, that's fair. But you know, like these 32 bit registers, what do they ultimately impact this performance? So, like, any single instance of the EVM will have some cap on how many transactions per second they can process.

Justin Drake
Let's say that this cap is x. Well, what we can do is just have multiple instances of the EVM, right? We can have 1000 instances. And now with synchronous composability, if you're willing to pay for this, then you basically have the ability to spread your state across multiple instances of these evms and still have them act like one super EVM. Now, another thing worth mentioning is that we don't have to use the EVM.

The beauty of rollups is choose your virtual machine. And if it turns out that SVM is the best and greatest virtual machine for the rest of humanity, well, guess what? What we're going to see is we're going to see SVM roll ups. And there's actually teams building exactly that. One of the best teams on Solana has migrated away from Solana is now building an l two.

They're building a roll up on Ethereum, putting the SVM there, and essentially decided to really buy into this thesis that the l one is winner take most, and Solana is probably not going to make it as an l one. And so really the next best thing that they can do is launch as a roll up on Ethereum and be a dominant execution chain on top of the incentive value. I think you're a little bit misstating their intentions. They're going to run a product on Solana and one on Ethereum. I don't think they're going to migrate away.

Anatoly Yakovenko
Sorry, I didn't mean to interrupt you, but you're kind of oversold. No, this is the end of my rant. You don't have to use the EVM. If it turns out SVM is the answer, Ethereum will use the SVM. So the core benefit of what Solana does is global synchronization.

And when you move stuff into l two s and pre confirmation, you're effectively running a centralized exchange like it's trustless and you can verify it, but you have a single spot, whether you're trading in New York or Singapore, where stuff happens, and that's fine, but that's not a single giant global state machine. The whole point of Solana is to synchronize data around the world. The fact that it's high throughput, the fact that it's in parallel, and all this other stuff, these are secondary effects. And why data synchronization is really, really important is because if an event happens in Singapore, that newswire has to go travel to the market where it's executing in New York. And by the time it gets there, what I want is the markets running in Solana to already reflect that price.

So a trader that gets at Newswire in New York cannot arbitrage it between their centralized l two, their Nasdaq exchange. Doesn't really matter to me what that is. With sub millisecond preconfirmations in that little region. It's going to be as good at price discovery as this giant state machine that is propagating information around the world as fast as possible. The part where it's high throughput is a side effect of us needing to get to that low latency and global distribution of data, because that requires parallelism and a whole bunch of cores and fast links and all this other stuff.

That's the core important part that Solana brings to the world. So when all these svms and everything else launch in these separate little environments, that's fine. It's just Solana's competing with centralized exchanges, it's competing with binance, it's competing with Nasdaq, and all those things are faster than those l two s. But what they can't do is the fact that that if some event happens anywhere in the world, you can permissionlessly change the price of a market that is impacted by that news event in Singapore and be competitive with everyone. That's the really, really cool part.

And when I think about the world in the future, 100 years from now, in Sci-Fi finance land, I see that as the winning piece. I don't see all this myriad of different environments with local markets and local information synchronizing in these local spots. I see one giant state machine with neutrinos going through the center of the earth, because that's faster. So you're highlighting three things. One is that we're talking about throughput.

Justin Drake
Two is that there's latency, and then three is this idea of synchronicity, where everything comes together and you're saying that throughput is not very important. It's a commodity and everyone has it. Great. I would say that's maybe one of Ethereum's biggest weaknesses. And so if it's going to become a commodity soon, that's all the better for Ethereum.

Now, the other thing that you brought up is synchronicity. And here I do think we can fix Ethereum's asynchrony using real time proving and chat sequencing. So now let's focus on the middle one, which is what you say is the most important, which is the latency. Asynchronous. You can't like fix the shared sequencing by moving it all into one giant block producer, because then that thing is the Nasdaq.

Anatoly Yakovenko
All the information happens there, right? And if I'm jump or whatever, I just run my shit right next to it. That we do agree upon is this idea of having very sophisticated block builders and having unsophisticated validators. This is part of your endgame vision. And really what we're saying here is that the sophisticated entity, which is only the leader for a very small amount of time, for 12 seconds, in the case of Ethereum, is getting rotated on a continuous basis.

Justin Drake
Now, are you saying that this number, 12 seconds, is just too long because there's just too much time to have a monopoly on a global state? Our goal is, you know, the latency around the world is around 120 milliseconds with speed of light through fiber. If we get to neutrinos, it'll be like 40 or whatever, but we're not there yet. I want multiple block producers on Solana rotating every 120 milliseconds. So there's one in Singapore, one in New York, one in every region in the world, as many as we need, and every 120 milliseconds they switch.

Anatoly Yakovenko
That's the end state. What you're saying basically is that whoever is this sophisticated entity that has monopoly over the next 12 seconds, they act like some sort of centralized exchange, and this is not great. Well, there's a few things here. One is that I expect transactions to eventually become encrypted. So today we have plain text mempools, and soon we're going to have encrypted mempools.

All these trade offs will delay when information becomes globally available. Okay, so we are talking about latency. I thought we're talking about the multiple variables to optimize. And this is the challenge. Like if you encrypt data and it's revealed later, you're delaying.

Justin Drake
When that is not necessary. Data is available to the rest of the global market. If you have a fully homomorphic encryption, you can execute the transaction immediately, even though it is encrypted. But the result of that that is important to the rest of the world. So me as a market maker, and if I'm putting price at risk and this shit happens later and I can't respond and cancel, you can respond, right?

Anatoly Yakovenko
It means that I have to have wider spreads. Well, I can't because it's fully homomorphically encrypted. Well, I can come in with a transaction which says if the price is x under the encryption scheme, do such and such action. So I can still arbitrage a market even though they're encrypted. When me knowing when this stuff is actually executed or not, is delayed and that increases spreads.

There's a system where everything is executed as fast as possible, and that information is available as fast as possible to the entire world. That system will have the lowest spreads and the best price discovery. That's the ideal. Okay, so you want the decryption to happen as fast as possible. Now here's the thing.

Justin Drake
We can have the decryption happen in a matter of milliseconds. And the reason is that we can have decentralized pre conformations. We can be in a position where the entity that is receiving these encrypted transactions can commit to including those or including them in a specific order. And once the commitment has been made, if they were to change their mind, they would lose a very large bond that they put forward. So those are communication cycles.

Anatoly Yakovenko
There are hops there. It's one round of latency, which is the same as with Solana. If I want to interact with Solana, I need to do, at minimum, one round of latency. Sure, but the goal there is that you submit and you know that it's going to be accepted simply because the block producer has accepted it and has very low probability of failing at that point. So you as a trader, you can start building on the next block as if that data is already there.

And the market's already reflecting it. Okay, so you're saying there's two rounds of latency instead of one round of latency. Basically there's one round to send your encrypted transaction, get a pre confirmation, and then the second round in order to actually know what the result is. But actually that can be reduced to maybe one and a half rounds of latency. Because as soon as the easiest way.

For Ethereum to have this is to just take Solana blocks and dump them into ethereum da and say that Solana is an Ethereum l two and be done with it. One of the things that we're not ready to do is have these really short block times. And we'll talk about this in the ugly section. And the reason is that that introduces ugly typing games and that starts weakening the security of the chain. So one of the concepts, but again, we'll talk about it later, is slot to ping ratio.

Justin Drake
So the slot duration, for example, in the case of Ethereum, is 12 seconds. And ping time roughly, let's say, is 100 milliseconds, just to make the math easy. So we have a slot to ping ratio of 120, you have a slot to ping ratio of four. And I believe as soon as you get close to one, even four is relatively close, you start getting these weird effects where even optimizing a few milliseconds gives you a crazy edge. Then you end up in a position where all the participants, the validators, the block builders, everyone, is basically building radio towers.

And that's fine programming FPGA's. It is not fine for the validators to have access to radial towers. That is not fine. The verifiers don't. The validators are the people verifying the data.

Anatoly Yakovenko
They're going to be receiving it sprayed around the world. I don't think that matters there. We've already seen timing games in Solana where the validators are willing to, to wait a little bit so they have a bit more information and they can get a similar edge just by optimizing their latency and using radio towers. That's fine. The world is not, because what will.

Justin Drake
End up happening is that only the validators with the best infrastructure will survive and everyone else will have a suboptimal yield. Let's say the yield is whatever x percent. Well, we're going to have those that get x percent and those that get half of x. What's the difference between the super mev optimized yields, ethereum versus plain vanilla yield? Is it 4.4 versus 4.44 or something like that.

One of the things I've tried to do when thinking about Solana is actually trying to steal, man. It, it turns out that blockbuilding specifically, and I think this is something you might agree with, is something that can be decoupled from validation. Yeah, totally. So we already have proposal builder separation, but we can go even further. We have this notion called execution tickets.

We have this notion called execution auctions. I don't think all the latency games with blockbuilding and Mev will be relevant in the long term. So that's squarely kind of a temporary bad thing. But what is much more fundamental is timing game for attesters. Those that are making the votes, they are meant to be as unsophisticated as possible.

And I think this is something you agree with. Now, if I am an attester, a voter that has, let's say, 20 milliseconds or 50 millisecond edge over all my other fellow ad testers, I'm going to be making more money. How? How? Because I'm going to be making more informed vote.

Like, whenever I vote, I'm going to be voting correctly on the head, as opposed to voting on some sort of block that gets reorged. And then me getting. The way that Solana distributes data is we get, effectively, a sample that the rest of the network has to block. So there's very little to optimize there. It's a very small portion of the overall returns for the validator to skip to.

Anatoly Yakovenko
Never pick a bad fork. It's a 100th of a percentage. Because the way turbine works is that you receive a whole bunch of samples from the rest of the network, and you know that everyone in the network got the block. So if you vote on it, you know that everyone else is going to vote on it. Like, you don't need to wait two to 50 milliseconds to see.

Do I see anyone else vote on it? And vote on it, too, especially in the environment where the validators don't execute any code, they just vote on. Did I receive the data? Are we picking this fork or not? When you look at Solana, it's a long run of blocks that are all built on top of heaviest fork.

So what percentage of blocks actually are forked? 5%. So, like, 5% of your votes are skewed. It's an insignificant difference in return. But that's today.

Justin Drake
But that's not the fundamentals. And the reason is that put yourself in the shoes of a block builder, sophisticated, who has the right. To build the next block. No, like, what they're going to do. Returns from votes are not tied to the value of the block.

Well, they are. They are. It's subtle. Put yourself in the shoes of a block builder for the next lot. What they want to do is they want to gossip and build their block at the very last moment.

And the reason is that, from their perspective, every millisecond counts. And what they're going to do is that they're going to gossip the block right at the point where whatever number of votes they need, let's say it's two thirds. Let's say they need two thirds for their block to get accepted. They're going to gossip just at the point where two thirds plus one of the validators are going to see it, and then one third of the validators, they're just not going to see it in time. And so they're going to vote on some parent or some other block and they're going to get penalized.

And these effects are just massively amplified when the slot to ping ratio is too small. That's not what we see in the data you actually see. That's because today the blockchain is immature revenue. We don't even have PBS on Solana. Just come on.

You're not a mature chain right now. Revenue on Solana is equal to Ethereum. L one. Yeah, but you don't have PBS. This is something that's coming soon.

Presumably. I'm going to have a call with the Jitto guys on PBS. Specifically something that is basic infrastructure in 2024 that you don't have. The actors on Solana are not acting rationally, like they're just running the default software because they're not sophisticated. But over time, and this is something.

Anatoly Yakovenko
That we've learned, the amount of benefit that you get is insignificant because you only have 400 milliseconds to build your block, you will get overrun by the next guy. And if you will miss your block, you miss it. Miss all the rewards. You said it yourself, you only have 400 milliseconds. So let's say that you have a tiny edge.

Justin Drake
Let's say you have 40 millisecond edge. That's 10% more immediately. Why would you wait to transmit your block at the last moment and let the next guy an opportunity to overrun you? Why wouldn't you transmit your block as fast as possible? Because you're going to submit it right at the point where two thirds of the validators are going to see it on time and one third of the validators are going to be, why is.

Anatoly Yakovenko
That more profitable than to do this? Because you win a few more milliseconds versus the naive strategy. Milliseconds in terms of what? Math? Yeah, yeah.

Justin Drake
And, like, if you have 400 millisecond slot times, four milliseconds is 1%. And as a block builder, I'm competing relative to the other builders, and so my profit margin might only be 1%. And so by having a four millisecond edge, I could double my profit margin. I don't think so. I just don't see that as, like, a realistic result.

Anatoly Yakovenko
Like, look at Ethereum. Okay, so look at Ethereum right now. What is the difference between a totally no mav client and a fully maximum mav client in terms of APY? It's huge. It's like something like two x different.

No, it's not. It's like 4.4 versus 4.5 Apy. No. Justin, can you kind of help us handhold some of the listeners who might have lost the thread a little bit? So you're doing some MEV calculations with regards to block times, and I think you're alluding to an acceleratingly centralized network as a result of increased value capture.

David Hoffman
Can you kind of present some napkin math and where that actually points to your conclusion? Right. So, as a rough heuristic, mev grows linearly with time. So if you have 400 milliseconds, that's worth twice as much as having 200 milliseconds. And one way to think about it is that there's twice as many transactions that come in during this period of time, and so twice as many fees.

Anatoly Yakovenko
That's not true, though, because a lot of transactions land before your slot. So you've accumulated more than a block's worth of transactions prior to your slot, and then you may get 400 milliseconds more. But your period of when you're actually receiving data is anytime. Just like MNMM pool people just submit shit ahead of time. So you have a bunch of data that you can work with.

The extra 400 milliseconds account for a very small window that. So when you're cutting that 4%, you're talking about getting the very last bit of people that are trying to snipe something, maybe, or just stay late in the. When they clicked something. Look, I'm one of the operators of the ultrasound relay, which is the top relay on Ethereum, and I have very good visibility on timing games. All the builders will co locate all the relays will do crazy optimizations to shave a few milliseconds.

Justin Drake
This thing is real. Even with twelve second block times, it's going to get insane when the block times are 400 milliseconds because the marginal improvement is for every, let's say, millisecond that you get. From a relative standpoint, it's much bigger return on investment than with Ethereum, where you have 12 seconds. I think when you have high throughput, the number of transactions you're dealing with is much higher in any individual one, and the ROI is much lower. So you're talking about median fee on Solana is 0.2 cents.

Anatoly Yakovenko
So that extra 400 milliseconds, extra 200 milliseconds, even over the amount of time that you're waiting and receiving transactions, is accounting for a very small amount of actual fees. No, I mean, if your stop time is 400 milliseconds and you have a 200 millisecond edge, which you wouldn't, that's too high. But let's say you have a 40 millisecond edge, then you get 10% more time to build your blocks, and you get 10% more transactions and therefore more fees. But you don't because you get a bunch of transactions before your block starts. But these are the ones that the previous builder didn't want to pick.

Justin Drake
They're like garbage transactions. No, these are just people retransmitting and trying to land. And they're not garbage transactions, they're high value transactions. Okay, but that retransmission is a Solana problem. This is not something that we have in the context of Ethereum.

In the context of Ethereum, we have low value transactions that kind of just sit in the mempool for slots and slots and slots. They're all low that don't get picked up. Like, the median transaction fee in Solana is 0.2 cents. So 50% doesn't matter what the medium transaction fee is. Sure.

Anatoly Yakovenko
So the average transaction fee is two cent. Let's say that last 200 milliseconds is going to get you 500 times what two cent worth of extra data maybe at the best, but don't you just. Make it up on volume? I think that's Justin's point, is like, sure, we're measuring things in milliseconds, but Justin's talking about it being relative. Those transactions are not higher ROI.

Go compute this. Go look at the flooded transactions and see. Do you actually get higher value transactions in the last 40 milliseconds versus the first? And even then, does it even matter to delay sending the previous transactions versus just I have 400 milliseconds left in my block. I will take the highest paying transactions at the last moment and send those out.

Like, there's no real. Like, you have to show me that there's a massive financial advantage to not transmit whatever you have early in the block, because you only have 400 milliseconds and you got to generate those sticks and send them out. And this is happening in Solana. In a current environment, we only have one block producer, but the design is for concurrent block producers. So now you have multiple block producers around the world that are not aware of what each the other one has.

So they have to race to fill the block together. And if they delay in that one particular one gets dropped, that guy has pushing his boundary in terms of losing all his block rewards. I feel like some of these details maybe require some follow up after the episode if Justin wants to prove something to Anatoly, and Anatoly wants to refute it. The interesting point here is I think the goal of Solana is to solve these problems with hardware. We reduce the block times to what the physical limit allows, let's say 120 milliseconds.

We have multiple block producers. That requires more bandwidth, so we need more cores and bigger pipes. And we create this very highly competitive environment from the hardware itself. So when you have these games, it's very, very hard to get an edge over the default. And that's possible.

You may get an edge at any given time, but it's not going to be 20 x or ten x. It's going to be basis points. So I want to zoom out at this point. So last time we kind of handed the mic to Anatoly, who mentioned the bad. If I were to sort of summarize the bad, it's almost like the bad and ugly.

Ryan
I think what you were saying, anatoly, we got in the minutiae of, like, you know, millisecond block times and Mev and all of this. But I think what you're saying is the reason Ethereum is under optimized is because it's not doing transaction confirmation as close to the speed of light as possible, right? And it's not propagating this global shared state at that level. And the longer the window you have, the longer that window is for front running and queue jumping, and the transactions go up and the spreads go up and the fees go up, and the worst experience it is for users. Okay, so, Justin, what's your take on that without going to the minutiae?

I don't know if you want to steal man and stole his case for this. But why does Ethereum not do this? Why is this not bad for Ethereum not to optimize for global shared state at near the speed of light, as he's saying, in the way he's saying it, maybe give your reason for that, and then maybe you could get into some of the bad that you see on the Solana side of things. So, one of the pieces of good news here is that at the end of the day, it's largely a parameter. It's some sort of number that we can pick.

Justin Drake
How many milliseconds per slot? Is it 400? Is it 12,000? And if film got it wrong, then maybe this is something that we can change. The way that I see it is that it boils down to this slot to ping ratio.

So we're bottlenecked by speed of light. Here. It takes roughly 130 milliseconds for light to go all around the earth. And ping times roughly can be approximated as 100 milliseconds. And what happens is that there's boundary effects between the slots that are on the order of this ping time.

This is where the so called timing games can be played. So if you have 120 ping times that are all one next to the other, and then you concatenate the next one where they touch each other, where this potential for timing game is actually very, very small, but once your start times become so, so small, then the edge cases start to accumulate. All of Solana engineering isn't solving that. And this is how turbine and provision and all this stuff works well together, is that boundary is deterministic from any validator's perspective. Like so, when you receive all the data from potentially n number of block producers, how you receive it gives you a very, very good statistical guarantee that everyone else saw the exact same partition.

As timeliness is fundamentally subjective. There's no way to have a global agreement on when we receive, every different validator sees the packets at different points in time. All of our engineering is to make the best case, the average case. This is what all of Solana's engineering is about. What will happen is that the block builders will gossip the block just at the right point, so that two thirds of the validators see it on time.

They vote, the block gets accepted, and then one third of the validators are left in the dust. They can't do that because of how the blocks have to propagate through turbine. There's a deterministic path that shuffles the data around the world with erasure codes. So when I receive it and recover that block, I know that with a very high guarantee that everyone else has done the same thing. So when I vote based on that information, it's very, very sound.

Yes, but what you're talking about here with azure codes is data availability. I'm talking about timeliness, which is a different thing, which is that same thing when on my clock, which millisecond did it arrive? And fundamentally different routes on the Internet. Internet take different amounts of time. There's jitter, there's all sorts of issues.

Anatoly Yakovenko
And we sample all the routes. So when 50% of my routes, for every shred that has to take a different path around the network arrive, I know that statistically they have arrived for the vast majority of the network as well. So when I vote, I know it arrived for everyone else. The end game is that the builders have a direct pipe to two thirds of the closest validators. They don't.

Justin Drake
Like. They don't. They don't, Solana, they can. It's just an implementation detail. There's nothing in consensus that prevents them to just have a direct pipe.

You know, even the colocation, Justin, why does that matter? Why does any of this even matter? Why have we been talking about this for so long? There is a global world, right? Your direct pipe is as fast as everyone else's because you still have to get your blocks to the validator on the other side of the world.

Anatoly Yakovenko
It has to go through the same fiber. The reason this matters is because it is a centralization vector. Okay? What happens is that elevates the barrier to entry. In order to even be able to, to play the game, you need to have outstanding networking infrastructure.

You don't. And by outstanding, I mean above and beyond what is available to the commoner. Every commoner can go look through their phone book and find a data center. In their city that has data centers won't make it. And part of the reason is that you're going to need access to, like, private links.

No, you won't. That only, like the HFT firms have. That is where we're heading. No, you don't. If the Solana network is actually distributed around the world, sure, in a world where everyone stuffs older validators in, like, two data centers, that's bad.

But if the validators are actually physically distributed around the world, those HFT links are not going to help you. There's going to be incentives for the validators to migrate. Why? Why? Because they're going to be closer to the builders.

Justin Drake
Therefore they're going to be able to make better votes and therefore have a higher return. Again, like I've said, you can't make better votes to get an edge when every block is built on top of the heaviest forest and you have statistical sample that tells you, yeah, everyone else saw this block. This is the whole point of, like, Solana's design, is to eliminate the forking choice from that as much as possible. So, Anatoly, is your prediction that, like, making these design decisions for Solana will not result in centralization of the network and the way Justin's describing, and Justin, is your prediction that these design decisions will centralize the network in the way that you're describing? It feels like we won't be able to resolve this in the scope of this conversation.

Ryan
We just have to kind of wait, do we not? And, like, do we have any concrete predictions that either of you would make some point in the future that we could look back on maybe years from now and say, oh, you know, this is how it happened, you know, like, one of us was more right than the other. Justin, do you have a concrete prediction here as to, like, what the outcome of this might become so we can table this topic and move on? Yes. So if Solana does go forward, you know, towards its end game, where it reduces slot times to the minimum, let's say, 100 milliseconds, and this is my understanding of what Anatolia wants to achieve, then, yes, I think we're going to see hyper centralization within any given slot.

Justin Drake
So in slot n, what we're going to see is, like, all the attesters just being in a data center or in a few data centers that are all proximate to each other. And Anatoly, if you are right, none of that happens. Right? Kind of like the properties Solana has today remain in place. It doesn't centralize in the way Justin's talking.

Anatoly Yakovenko
I think if the parameters are set correctly, that we have, like, 130 millisecond block times, you can have a validator anywhere in the world and vote, and you will get about the same API as anyone else. All right. I think the difference is going to be basis points. Well, we'll have to see in the future, we'll have to replay this episode, this clip in particular, and see which prediction pans out. Celo is the mobile first and EVM compatible blockchain that's built for the real world and designed for fast, low cost payments worldwide, driving real world use cases like Opera minipay, one of the fastest growing web three wallets with over 2 million users users across Africa, Celo is seeing a meteoric rise with over 375 million transactions and a million monthly active users.

David Hoffman
Not to mention tether and circle just deployed native USDT and USDC on Celo supported by leading exchanges like Binance. And now Celo is looking to come home to Ethereum as a layer two with a game changing proposal. Core contributors at C Labs aims to leverage optimisms, op stack pioneering and transition as the biggest l one to become a layer two, with Testnet arriving at earliest summer 2024. With the Celo layer two, gas fees will stay low and users can even pay for gas using ERC 20 tokens including native USCC and USDT sending crypto to phone numbers across wallets using social connect. But Celo is a community governed protocol make your voice heard in the Celo forum to shape the future of Ethereum.

Follow on Twitter and explore the ecosystem built for the real world on celo.org quests new projects are coming online to the mantle layer two every single week. Why is this happening? Maybe its because mantle has been on the frontier of layer two design architecture since it first started building mantle da powered by technology from Eigenva. Maybe its because users are coming onto the mantle layer two to capture some of the highest yields available in DeFi and to automatically receive the points and tokens being accrued by the $3 billion mantle treasury in the mantle rewards station. Maybe its because the mantle team is one of the most helpful teams to build with, giving you grants, liquidity support and venture partners to help bootstrap your mantle application.

Maybe it's all of these reasons all put together. So if you're a dev and you want to build on one of the best foundations in crypto, or you're a user looking to claim some ownership on Mantle's defi apps, click the link in the show notes to getting started with Mantle launching a token worried about the IR's? Talk to Toku for five minutes to learn how to save hundreds of thousands of dollars for you, your team and your investors. Now, if you do an initial consultation, Toku will cover the cost of your token valuation, which is what you need for your launch. So don't wait.

Reach out to the team right now@teamoku.com. dot that's team.

Ryan
We're still on kind of the bad section, so no, I think that was. Definitely the ugly section. Yeah, Justin, yeah, it's all combining, isn't it? Justin, do you want to give your take on either the bad, the ugly, or both? For the Solana side, that's outside of the scope of the conversation that we just had.

So, something new? Yeah. I mean, I didn't touch upon the bad facts, but in some sense, they're not really important. And as David said, they will just dwindle with the passage of time. So there's all sorts of bad facts, like the outages being very proximate to the largest financial fraudster since Madoff.

Justin Drake
The misleading marketing. Even today, I went on Solana compass. I thought this marketing was kind of gone. But still, there's this marketing that Solana can do 65,000 transactions per second, which is, first of all, that was a claim that was made four years ago, and it's still not achieved. You can go run the code yourself and test it at a very, very simple transfer.

Okay, show me on production. There's also bad present facts. There's no client diversity. There's no formal spec. There's no PBS.

There's just general immaturity. But again, I think these are not fundamental things because they're being worked on. There's another client being worked on. I'm sure there's some academic somewhere that's working on a formal specific. And I do know that JItto is working on PBS, so all of these things will get result.

But then there's a final category of items in this bad bucket, which is the bad takes. And in my opinion, there's a couple of bad takes, Anatoly, that you have been spreading, which hopefully this podcast can help you spread less. One is this idea that issuance is not a cost to the network. Now, there's a very simple reason why it is a cost, and it has to do with taxes. My assumption, Anatolia, is that you pay your taxes.

Is that correct? Of course. Okay. So if you were to stake and you were to make, let's say, a million dollars of income staking, my understanding is that you're based in California, and I looked it up. The income taxes in California is roughly 50%.

So you will have to pay half a million dollars of income taxes on your staking. That is direct sell pressure. That is a cost to the network. Now, Solana is, roughly speaking, a $100. Billion network sell pressure is not a cost.

Ryan
Wow. We have to go into this. Okay. Wow. Like, the value of an asset is not determined by how much is sold.

Anatoly Yakovenko
If you're talking about how much Apple stock somebody sells is not a cost to Apple. Okay? So someone has to pay for these costs. So, for example, in bitcoin, like, someone has to pay for the electricity, someone has to pay for the e waste and the hashrate, that is a cost? Yes.

That is an actual cost to run the network. That's a physical cost? Yes. It's a physical cost, sure. But like taxes, even though it's not physical.

So every share of Apple stock sold is a cost to Apple? No. Let's say that there were no taxes. Let's say that we live in a world without taxes. So if I sell Apple stock and I pay taxes on it, that's a cost to Apple?

Justin Drake
No, this is different. Are you differentiating, Anatoly, between a cost to apple and sell pressure? Are these kind of different concepts in your mind? Sell pressure is a local event. That's temporal.

Anatoly Yakovenko
But the value of the company is based on Apple. It's based on their future cash flow. And in the US, the monetary market will expand to capture that value. Somebody will go buy all that stock, because the future cash flow of Apple is worth so much. So, selling Apple stock is not a cost to Apple.

That's, like, obvious to everyone. Apple does not issue stock to pay for its security. So Solana is a roughly $100 billion network. Roughly speaking, 5% of issuance per year, which means that it's not solar Solana security, which means that 2.5 security of. The network doesn't come from the issuance.

That's absurd. $2.5 billion per year of taxes is being borne by the Solana token holders. If we were to live in a world where there were no taxes, then this cell pressure would not exist. That is definitely a burden to the network. This is a complex thing to analyze, because people that are offshore can earn their rewards offshore and only pay taxes on value they appropriate to the US.

Justin Drake
Okay, but let's take your example, Anatoly. Do you contribute to sell pressure of Solana? Probably. Okay. But, like, is that a burden to.

Anatoly Yakovenko
Sure, and I pay my taxes. Okay, good. Yeah, of course. Sure. Like some fraction that does not change the future cash flow of any asset.

Like when somebody sells their stock. So it has no relevance on that. It is the exact same thing as proof of work. In the context of proof of work, the minus. Absolutely the opposite of proof of work.

Proof of work actually have to go buy electricity and burn an external exogenous asset to run this. You have to go buy an exogenous asset called us dollars in order to be able to pay your taxes. And in doing so, you're selling your solana. That's not a cost to the network. Anatolia, you said something else here, which is like, issuance does not provide economic security.

Ryan
Did I hear you correctly there? I'm wondering if you could say that again. It does not provide security. Security of the network is the physical boxes and links. Like, if I'm running a system that cares about detecting double spends and invalid state transitions, I run a whole bunch of nodes, a whole bunch of clients, and make it really, really hard for an attacker to partition my clients such that they can double spend.

Anatoly Yakovenko
So if they try to double spend, they have to reveal two headers to two different sides of the network, and they don't know which nodes are mine. So it becomes the more nodes I run and the longer I wait, the harder it is for the attacker to do the double spend. So it's irrelevant. Like, this is why we don't see double spend on the Solana testnet. We have as much economic security in the Solana testnet as the Ethereum mainnet.

It's as good at defending against double spends. The Chihuahua tendermint chain has seen as many double spend attacks as Ethereum mainet because it's as good as defending it against double spends as Ethereum Mainet. So this ties into the second bad take that you've just spread and you constantly spread, which is is basically that you're saying that economic security is a meme 100%. And you are right in the sense that economic security is not necessary. To verify the correctness of blocks, I run a full node.

Justin Drake
If a block's invalid, I just filter it out. And then there's two other things that economic security could potentially be helpful for. One is, as you said, double spends. Now, it turns out that doing a double spend, you need to have a very, very good reason for doing so. And there's a little bit of a tangent, but we can also just completely solve double spends with cryptography.

There's this notion called one shot signatures, and it turns out double spends is not fundamental to consensus. But now let's focus on where economic security is fundamental, which is to do with liveness and censorship resistance. Let's say that I can 51% attack a chain. Then what I can do is I can selectively filter out transactions. For example, I can filter out oracle transactions.

I could filter out liquidations. I could filter out fraud proofs. I could filter out buy orders or sell orders. I can manipulate markets. Now, let's say that I can make a billion dollars manipulating markets through the selective censorship.

And the economic security is, let's say, $1 million. Well, now it's a no brainer for me to invest the $1 million to go 51% attack and make a billion dollars of profit. And so economic security is not a meme. It is fundamental to censorship resistance. When you talk about security, it implies safety.

Anatoly Yakovenko
No, this is probably. No, there's two aspects of security, safety and liveness. This is basics. Sure. But when we talk about security, you would want to place safety over liveness.

Justin Drake
No, they both are important. Allow a network to run with an invalid state transition, and under any circumstances. No, no, they're both important. We've talked about this. The invalid state transitions are easy to fix just because you can run a full node and you can just discard them immediately.

This is something that has been known for 15 years, since bitcoin. Sure. If you run a full bitcoin node and some sort of miner tries to create an invalid bitcoin chain that prints, you know, 10 trillion BTC, it will just get rejected, of course. And that could cause a liveness failure. If the vast majority of the network is trying to produce invalid blocks, you have no confirmation.

Anatoly Yakovenko
You can't tell. Right. It's effectively aliveness failure. So you pick safety over liveness. Any case, like economic security to a lot of folks implies safety.

Justin Drake
No. So when we talk about censorship, resistance, and liveness, it does have impact there, but it does not provide enough. So it's not a meme. Oh, I think my point is that other networks, like tendermint networks, like Chihuahua chain, like the Solana testnet, provide equivalent amount of security there, even for liveness and censorship resistance. Yes.

No, because I go into our, I buy whatever required number of tokens, let's say I need to spend $10 million to get enough tokens, Chihuahua tokens, to go 51%, attack it, and you take. All the math for a day. Yes. And people recognize that. You take all the math and they fork and slash you.

Okay, so you do agree that now it basically depends on how much profit you make. Now, we have ambition in Ethereum, which is to be the Internet of value. We want to have the reason why. Trillions, tens of trillions of dollars, the reason why that are settled and traded on Ethereum. If you can disrupt the Internet of value, even for 1 hour, even for 1 minute, that is worth an insane amount of money.

Anatoly Yakovenko
It's not. And we cannot have the rest of the world, tens of trillions of dollars secured by $1 billion of economic security. The rest of the world runs on t plus two to t plus ten settlement. It handles trillions of dollars, it has outages all the time and liveness failures. It works.

So you're trying to replace it with a better system. I get it. But you can't tell me that an hour outage or ten hour outage is fundamentally going to stop the world and everyone's going to be panic on fire. That's just bullshit. Yeah, economic security protects against a very specific attack, but that attack does not exist on any other chain because it requires more than epsilon stake.

You have to go buy a huge chunk, it's detected. Once it's detected, the attacker is slashed. So that effect exists on every chain. The fact that it's detectable and the fact that it's easy to trivial to slash, those are the things that secure the Solana testnet and the tendermint Chihuahua chain. And this is why you don't see these attacks anywhere.

So you can't tell me like why like this meme is bad. It's like you're using it as bad marketing. Ethereum has world war three level economic security. It's bullshit. Because one, I can't tell looking at the stake, and there's no way you can prove to me that all of that stake didn't come from the ICO.

There's zero way you can guarantee that the cost basis for all the current stakers is not what they pay, like the ICO price. There's no way you can prove that it's impossible. I trust that it's not because I'm a rational person that believes in mostly functioning markets and most people are honest and stuff like that. But if we're doing hard analysis and like, what can we actually prove? You cannot prove that.

So the actual economic security based on some number, based on the current spot market price is meaningless. Real security is detectable, it's apparent in any network. And that's why we don't see them in even testnets. No. So I think we've agreed that economic security is not a meme because it is relevant to this one specific.

It is a meme because it is. Relevant to this attack called censorship and liveness failure. It's a meme because every proof of stake network does as good of a job defending against that, as Ethereum may know. Okay, so what you're saying basically is that if this attack were to happen, then the social layer can come in and do social slashing, which is correct, but there's a few points here. One, social interventions are extremely expensive.

Justin Drake
We've gone for one in Ethereum's history is called Ethereum classic is called the DAO is not something that you want to go into. So it's what, $300 billion worth of economic security to defend against an attack that never happens in any proof of stick network. But it's not a meme, it's never. Happened because two things need to happen. One is you need to have a sophisticated attacker who knows about this attack and is willing to plan it.

And two, the incentives need to be there. Right now, almost all of the value secured on Ethereum is ether itself, and it's staking. So in some sense there's nothing to attack but project yourself ten years in the future where there's short ether, tens of trillions, hundreds of trillions, potentially take. Over an Ethereum testnet, cause it to double spend with my economic attack while shorting Ethereum and say, look, there's a bug in Ethereum. Everyone panics, Ethereum drops, and there's a massive economic incentive to attack an Ethereum testnet that has zero cost.

Anatoly Yakovenko
Right? I assume in terms of economic security, yes. This particular attack wouldn't necessarily work because the social layer knows that it can fork it and it doesn't affect the rest of the market. But once you're willing to censor oracles, you're willing to censor fraud proofs and censor liquidations and market buyers in market sells, then you're in a position where you can harvest toxic meV. And if the amount of toxic MEV that you can harvest is greater than the cost of attack, then yes, it's going to happen.

Justin Drake
Just because it hasn't happened doesn't mean it's not going to happen. Now, you said that the social layer can intervene. That is correct, but that is a very expensive process. How is it any more expensive than rebooting from an outage? Well, in the case of Solana, these outages don't have real significance because there's no value on Solana.

Anatoly Yakovenko
It has the same revenue as ethereum. It's not about revenue, it's about TVL and things like that. Solana has 0.7% of all USDT. It is like, minuscule. Why are outages important for TVL?

That's sitting at rest and doing nothing. Why is revenue is what's actually important. No, no. It needs to be active. So there needs to be over collateralized loans.

Justin Drake
That needs. Which is determined by revenue. Right. A proxy for all that activity is actual on chain revenue. No, like a lot of the revenue for the validators, you know, come in this form of sandwiching.

Right. Like this is one of the primary forms of revenue on Solana right now. Sure. So Mev plus priority fees on Solana are about equal to Ethereum. L one.

No, but we're talking about like, really, really toxic levels of MEV. Like, sandwiching is nothing compared to the type of MEV that you can harvest if you were to do a fair two. So if somebody does that and that's bad, the network calls, slashes them, and moves on, and it's a 24 hours thing. Okay, let's say I can make $10 billion from this attack and it only cost me $1 billion. That is a bad outcome.

Anatoly Yakovenko
It's unlikely that that could be true because you would have to go buy up all of this. It seems unlikely that the revenues, within 24 hours, the maximum mev revenues could be that high, $10 billion. That are basically the threshold here is if it takes 24 hours to go fuck over this attacker, right, they're doing the bad thing. Everyone halts, they get slashed in the fork, and everyone continues. That's a 24 hours process.

Maybe they have to make more money than the 50% of the stake within that 24 hours. Okay, so there's a couple things here. One is that they can pay whatever 24 hours has the most impact. Like they can pick whenever the Fed decides to change their rates or whatever. Sure.

So there's the potential for revenue. Right? Secondly, it doesn't take 24 hours, like we're talking potentially an event that can take days to resolve. And thirdly, in your end game document, one of the things that you highlight, Anatoly, is this idea of sampling committees. You talk about committees of size 200 where let's say you have 20,000 validators.

Justin Drake
You pick a committee of size 200, and now only 1% of the validators are in control of the chain for whole epoch, which is, let's say two days, what are you going to do? You're going to slash one third of the 1%, which is like 0.3%. And now the attacker can do the attack again and again and again and again. And so what you're suggesting in your end game is ugly because the social slashing doesn't even work there. I would imagine the stake representation there is going to be higher than 1% simply because when you're sampling power law distribution of nodes, your 200 nodes already make up more than like 50% of ethereum.

What an attacker will do is that they'll spread their stake equally across validators. But then they're not going to be majority in the subcommittee. If you have like 10% of the stake, you're going to have 10% of the subcommittee. In any case, there is a degradation of the amount that is slashed, is you're not going to slash the whole attacker. The attacker can do the attack several times.

Anatoly Yakovenko
Sure. And again, that's just like in Chihuahua chain. They buy 10 million worth of Chihuahua coins. They do the attack, they get detected, they get slashed. This is why it doesn't happen, because.

Justin Drake
The revenue that you're trivializing slashing, you cannot trivialize social intervention. So we are full circle to Chihuahua chain. Again, I don't think either of you is convincing the other one of the points. So Anatolia positions. There's never been this level of attack in any proof of stake network.

Anatoly Yakovenko
It happens in proof of work networks because you can do an anonymous 51% attack. And this is the big flaw in proof of work, is one, is you can roll back two weeks worth of ethereum classic with an anonymous chain of links and nobody knows who you are. And you effectively robbed everyone that believed the previous thing. But in proof of stake, you have branded validators that compete for stake, and they're staked for an epoch. So there's a cool down and warm up period.

So they're known. And it's very, very hard to get to even point to what you can do. And once you pull off the attack, everyone knows who did it immediately. Like, the information is out there, so you're completely left yourself open to being slashed immediately. Let's move to another bad item that you might have.

Ryan
Or maybe we're in ugly territory. So we've talked about economic security. What else is on your list? This meme is benign in Ethereum, because Ethereum does a really, really good job with engineering. And they actually do have a lot of validators and a lot of links and independent clients.

Anatoly Yakovenko
All that stuff they do. It's a dangerous meme for everyone else because people will, like Luna, will roll up to some absurd number of stake and say we have massive economic security where it's like, not real in any way. And that gives people false confidence. I would actually like. What I would wish Ethereum would do is just like, forget that meme, just talk about the really excellent engineering.

These attacks are impossible because the engineering is really, really good. So Luna is incomparable. Like, they were building a stable coin and the design was just flawed. They were not. But their TVL, the amount of stake that they had staked to secure their proof of stake network was massive.

And they could literally just go, say we have massive economic security that is incomparable. And it takes a lot of work to go and parse those details and to tell a person that's not plugged in. No, you can't compare that economic security to Ethereum's, but you can literally easily say, look, that's 100 nodes all run by the same people. This is bajillion nodes geographically distributed with a bajillion links. This is secure.

This is not. Okay, so once again, on that item of economic security, is there anything predictive you could say? I think listeners might be asking, why does all of this matter? Right. So, like, Justin, why is that bad in your book that, you know, you think that economic security does matter and that it's bad for anatoly to propagate it and say that it doesn't matter?

Ryan
And I think listeners might be wondering why. Like, why did we just have this conversation, this debate about economic security? And if you're right, is there something predictive that you can say about some kind of future attack that Solana or chains like it might face if they're not considering economic security? Because we could test that in the future. Somebody with actually weak security will have strong economic security, and they will get wrecked.

Anatoly Yakovenko
This is the bad outcome? That's your prediction, Anatoly? Okay, this is the bad outcome that I think, as an industry, we should strive to avoid. Right. Because then you're giving people a very, very easy meme to, like, brand themselves with, versus really, really hard engineering, which is hard to do and build and deploy.

Ryan
I see. And you're saying that at some level that's already happened with something like Terra Luna using economic security as a meme? Justin, how would you put a bookend on why this matters? I think it's more of an intellectual point for Solana, specifically because, ironically, Solana has very high amounts of economic security, because the vast majority of its soul is sticks. I don't trust it.

Anatoly Yakovenko
Don't trust the economic security on Solana. Well, you do trust that the social layer can slash them if they were to misbehave. But you can't say that the chain is more secure. Like, what is it 20 times more secure now than a year ago? You cannot possibly say that.

Justin Drake
You can. No, you can't. You can, because there is money to be made from 51% attacks, and now the money to be made needs to be so much higher to compensate for the loss of slashing. Now, in terms of a concrete prediction, my prediction would be that some smaller chain that has much lower amounts of economic security than Solana will get 51% attacked and will be harvested for toxic mev through, for example, the censoring of oracles. Okay, all right, let's move on.

Ryan
Are we into the ugly territory yet from your side, Justin, or are we still on the bad? What's ugly about Solana? To finish up this conversation, then maybe we talk about kind of the end game of both ecosystems, and we call this a wrap, but what stones have we left unturned? Yeah. So happy to finish off on the ugly part.

Justin Drake
And it's kind of interesting because it has to do with things that seem innocuous. And specifically, one of the things that I think is ugly is the fact that Solana is an l one. The mental model that I have is that it's decided to create an island in the Pacific as opposed to being connected to Wall street. Right. There's existing network effects.

These are extremely strong network effects. And by being an l one, you don't get to tap into these existing network effects. And now, if my thesis is correct that the l one is going to take most, then Solana is not going to make it. Now, let's look at concrete numbers, and I think these are extremely telling. You have Solana, which is a four year old project.

How much USDT has it accumulated? It's accumulated under 800 million, which is about 0.7% of all the USDT. Now let's look for comparison on the l two s, a single l two. Arbitrum, which is only three years old, has three times the amount of USDT. Why is that?

Because they are connected to the epicenter TVL is of. TVL is irrelevant. It's not irrelevant. It's useless. It's not useless.

Anatoly Yakovenko
It is. How much revenue is arbitrum making off of that bajillion in USDT? Like, who cares about it? I have no idea. But it is a sign that people are using USDT for loans and savings and payments and all sorts of other reasons that USDT is.

No, it's not. That's actually, like, very, very misleading, bullshit. Amount of TVL in a protocol is meaningless. Big data protocol had 6 billion TVL. What happened?

That was an Ethereum protocol that had like, a big run last cycle, doesn't exist anymore. What happened? I thought TVL was really important. If the average user puts forward, you know, $100 of USDT wherever they're living, and let's say there's $100 million of TVL in a specific chain, that means that on average, roughly speaking, a million users. That would be if you have actually a million users that are actively using that money and doing commerce with it.

True, but that's not what TVL actually is. TVL is people fishing for rewards. Look at TRon, for example. Tron has a huge amount of USDt TVl. Why is that?

Justin Drake
Because TrON is actually being used for payments. It's actually being used by real people. You can't say that TVL is irrelevant. It is one of the strongest signals that you can have that the chain is actually being used. Again, there's many, many counterexamples of that.

Anatoly Yakovenko
I'd have to look at actual Tron data. How much of the TrOn TVL is somebody's money sitting in a protocol, earning, like, useless tokens? How much of it is actually spent on peer to peer payments? You got to separate those out. Like, TVL is not a good signal at all.

Ryan
Sure, Justin, beyond TVL, can you finish your thought on the ugly? And then we want to get anatoly to respond in your comments and then state his ugly before we get to then. So finish your thoughts on the ugly. Right. The fact that Tron has a lot of TVL is, again, a point against TVL being a good signal.

Justin Drake
No, TRON is, like, providing tons of utility for payments. This is a well known fact that a ton of payments in Asia. Let's move beyond the TVL. What else you got, Justin? Right, so basically what we have is that Solana is moving towards being isolationist and being disconnected.

Now, the l one is only one aspect. There's more to it. One is that Solana has no support for, like, clients. It has no real time mercurialization of the states. There's literally zero, like, clients.

And what that means is that the only way to interact with the chain, unless you're willing to be on a data center, if you have a phone, or even if you're another blockchain. Let's say you want to build a trustless bridge between Solana and Ethereum. You can't. And so this is why the only kind of bridges that you have are trusted bridges, like wormhole, because Solana has no, like, client support. And my prediction is that it will never have, like, client support for a couple of reasons.

One is that it's taken this shortcut of not having mercury, and it turns out that mercurialization is the bottleneck. And so if it were to reintroduce it, it would affect its writability and its TPS, and that's not something that it wants to do. You don't need mercalization for, like, clients. Like why? All I need is data availability proofs.

Anatoly Yakovenko
That's totally doable. And you need a path to the vote from my transaction that asserts the value of the state. Like I just need to submit a transaction that says check the state is equal to this hash. Okay, so basically what Anatoly is suggesting is that every time you want to read the chain, you make a transaction. That is crazy.

Justin Drake
That is absolutely insane. No, it's not. Yes, it is. Why not? For a high throughput chain with median fees below 0.2 cents.

Because you make orders of magnitude more reads, more reads on the blockchain than you do writes. So that would like the transaction that. You care about verifying can do that as its last instruction. You don't have to actually do this for every, every time you need to do a read. It's that if you're a payment provider that needs to receive payment flow in your UI, you force the user to do this.

Anatoly Yakovenko
They sign the transaction, you then do the whole SPV proof and DA, and you're done. It's an extra instruction for people that want to want that cost. That's it. That's bundled in a transaction. Good luck with that.

Justin Drake
My personal prediction is that no one will be paying for that. Everyone will go through trusted intermediaries. They're going to be willing to just trust whoever helios or helium or whatever it's called is telling them all the RPC providers and they will not be making a transaction first. If it's a commercial differentiator for those guys to do it, they can bake it into their APIs. Justin, do you have any more ugly before we turn it back to Anatoly?

No, that's most of it. The isolationism combined with the timing games in Italy. Just because we want to make sure you have an opportunity. Opportunity to. We asked you about the bad section.

Ryan
I think that bled into the ugly for Ethereum. But is there anything from your perspective that we haven't talked about that is ugly about Ethereum's design choices, value architecture? Why is Ethereum fundamentally broken? Yeah, I mean, one thing we haven't talked about just to spark things maybe is like running a node at home. And it seems to me that maybe Solana design decisions do not think that that's very valuable at all.

We could talk about that, but the floor is open. No, Nielsen's law takes care of that. I already have one gigabit at home and my test boxes work fine. I think the ugly parts on Ethereum that are not fixable is this focus on ultrasound money and decentralized money. And this is what pigeons it into a specific direction, which is means you're trying to build the lowest, smallest possible l one.

Anatoly Yakovenko
And it's very, very hard to do that and do all the execution on all these l two s and have any value capture on the l one. Execution is what makes money, is when you have priority fees. People competing for content, they want this particular dog coin versus another one. They're willing to pay extra fees to be there first. That's all in the execution layer.

That's where all the money is made. And that pays for a normal user that doesn't care about whether they land in this 400 millisecond block or the next one. And they just want to pay for a mobile phone to that user. They don't give a shit. Like when they click that purchase button, they don't care if it takes extra 400 milliseconds.

So their fees are fractions of a penny, right? Does this make sense? This is all because execution is able to capture high average fees and provide a low median fee. That's really, really hard to do in the settlement part. And when you move all the value capture away from the l one, you're left with, like, what are you left with?

It's just trying to compete with bitcoin and being the most better money. But that is not like real. It's all. It's a meme.

Decentralized money, I agree, is important. When my parents fled the Soviet Union, had these technologies existed, they would have preferred USDT on tron versus bitcoin or ethereum. And there's no way you could convince them that, hey, you're going to take this massive move and flee this country, that if you take ethereum, it's going to be within 90% of the value that you put in within a year, because the variability and all this other shit makes it a much riskier bet. So you need stable coins for normal people. Once you accept that as the thing this decentralized money meme puts you into, design decisions that are all kind of ruin the cool part of the network, which is execution, can pay for super, super fast, cheap stablecoin payments.

Ryan
Justin, I want to get you to respond to that, and then we'll end with closing thoughts from each of you. How do you respond to what Anatoly said about the ugly of Ethereum? So, Anatoly is right that there is no money to be made, no value to be captured in the settlement of execution, which is just verifying a snark. But there is this other part of settlement, which is data. And it turns out that data scales linearly.

Justin Drake
Snarks are magic. This is exponential scaling, where you can take arbitrarily large computation and shrink it down to a single snark. You cannot do that with data. Every user needs to consume a certain number of bytes. Now, one of the things that Anatoly said is that we're trying to build the smallest chain possible so that we can extract the highest fees.

That is completely wrong. We have a whole roadmap of scaling, and it's not about per transaction fees being high. We actually want per transaction fees to be low. It's about the product of the number of transactions times the median or the averaged transaction. So we want to have 10 million transactions per second, each of which are paying a tiny fraction of a penny.

But in aggregate, that could be hundreds of millions of dollars per day. But to the L one, they're just paying for data. That is correct, right? To the L one, they're just paying for data. Yes, and data is a commodity.

Anatoly Yakovenko
Like, there's no way the L one can charge more than a small multiple off the cost of bandwidth. Okay, so let's say that we have a validator and they're on like the best connection that is available to them, you know, from their home or whatever it is. Sixty four cents a terabyte. That's the cost of data. Not everywhere, but that's the cost that I can get.

And it's going to be everywhere within a couple of years. You cannot put the whole Internet on one Internet connection. That's just not going to work. Sure, but today I can get sixty four cents per terabyte. Within ten years, the world is going to be below that and my price is going to be even lower.

Justin Drake
Yeah, but there's this notion called induced demand. This is the guy who said, whatever, like 100 memory is going to be sufficient for the whole world. Like you need to project into the future. Today we might only be doing ten transactions a second, but in the future we might have our personal AI's do millions of transactions per second on our behalf. And so we're never going to be able to fit everything on a single Internet connection.

The more resources we have, the more we want to consume. But like the data portion, it's opaque, it has no content based pricing, no time based pricing. It is just pure commodity throughput, pure congestion. So as the l one cannot charge more than a constant factor off the physical cost of it, because if it charges a hundred x off the cost of the physical access, somebody's going to build network effects. This is like, again, this is weak.

There is a reason why data availability is worth more than Solana's data availability. Because of network effects. Because of. Of network effects, yes. This notion of shared security, this notion of synchronous composability because of the economic security, not just that, but also access to TVL, access to applications.

There's only one ens in the world, right? There's only one Ethereum and Internet name system. This is really weak. This is why I think it's ugly and it's not fixable is it's not a fundamentally hard thing. I thought we agreed that synchronicity is important.

Anatoly Yakovenko
Yeah. If you're on a separate chain and then suddenly you become asynchronous relative to all the applications that are living in the other chain. Sure. And so you pay a premium for synchronicity and composability. If its main purpose is to charge for l two data, this is how it's going to end up.

It's going to be very, very hard to capture value, like real value. Capture. All the stuff you're talking about is very hand wavy because you can't do cash flow analysis on it. Like Apple. I build this phone for x, I sell it for Y.

Right? Maybe it has value, maybe not, I don't know. But it's wobbly to me. If you're charging just for data, there's a fixed amount cap that you can charge for it, because if you charge 100 x on the cost of data, it's just open to competitors. There's just no way you can run a business that runs 100 x cost on a commodity service, like just providing data.

Celestia, whoever else will eat your lunch, and they'll be like, this is the better place for l two s. It's just going to be downward spiral. So this is like the hard part. I don't know if it's fixable. It's easy for me to argue how an execution layer will survive and work.

Ryan
Justin, brief comment on that, and then let's get to closing. So what's your brief comment on what Anatoly just said? Fundamentally, we're going to be limited by bandwidth. Even though it grows exponentially, we always want to consume more. That is the nature of humanity.

Justin Drake
And we can't put all the Internet of value, every single transaction, through one home Internet connection or even one data center Internet connection. So fundamentally, there's going to be congestion. And why is there going to be congestion? Because everyone is going to be willing to use this prime resource, which gives us network effects, gives us access to liquidity, gives us access to security. So you're expecting that the Ethereum l one will have blob congestion 100%?

Yes. Like right now, there's no blob congestion. So it's going to be under spec'd for its demand? No, we're going to have as much as we can. But even that's not going to be enough.

Anatoly Yakovenko
I think you're overstating demand. My guess is that the top 100,000 transactions per second will cover ninety nine point nine nine nine nine nine percent of value. Underestimating the demand is like saying there's only demand for five computers in the world. No, no, that's not true. There's going to be way more transactions than you expect we're going to have.

Justin Drake
Google does. Yeah, again, you're talking about. But science fiction future stuff. Instead of looking at a current data, Google does about 40 to 80,000 searches per second. That is a single global application that has scaled to every human and permeated their lives.

Anatoly Yakovenko
That is 80,000 events per second. There's always new applications every time. We've ten x'd bandwidth, we've unlocked new applications. You know, we've gone from text to. Images to videos, not like transaction events.

Ryan
So there are many points of difference throughout this conversation, and I have thoroughly enjoyed everything that's been said. I pray my competitors build for a world with infinite demand. Overestimating demand is as bad as underestimating it. Anatolia, once you get to your closing arguments, maybe talk about what you think the end game is going to be and how Solana fits there, and where you think the points of debate are, or dissimilarities here. And just like, close us out on the discussion so far.

So you have the floor. I mean, I think Solana has accomplished something really awesome. We have high average fees and low median fees. That means users can get very low access to the chain fractions of a penny. I think l two s are still not even within ten x.

Anatoly Yakovenko
And the network makes sustainable revenues that can pay for all the hardware ten times over. So it's a miracle, as far as I'm concerned. Holy shit. That there's enough money to be made here, that these boxes are going to run forever and access for users and use cases that I care about, which is just selling products through stable coins, like selling stupid phones, that works. That's like, I've created something that is useful for people.

And Solana's goal is to maximize this Moore's law and Nielsen's law, as far as I can go and optimize the network, because every time we do that, it's a benefit to the users. Spreads get lower, latencies get better. All these things converge on just better outcomes for all the users, and hopefully it still remains financially stable. So that's division, and it's very, very simple. There's no infinities there.

It's just like the current release works well at 16 cores. Let's get it to work better at 32 cores. And Anatolia, how does Ethereum fit in that kind of end state vision of what Solana is? Do these coexist? Does one become larger than the other?

Ryan
How do you think of that endgame? I have no clue. If this decentralized money thing, it might work. Bitcoin is worth a bajillion dollars, and it doesn't make any sense to me. Network effects, TVL, all these things could matter.

Anatoly Yakovenko
But it's not hard. It's not something that I can bet my work on, that I would want to bet my career on. I want to build stuff that I know is going to work. From my perspective, the more l two s and networks and different environments there are for digital assets, the better, because the more that stuff exists, the easier it is for a price discovery engine that runs as fast as the speed of light allows to actually find product market fit, that there's just more stuff to do price discovery on. Right?

Like, so please go build as many l two s as you can. Build as many ultrasound monies as you can. Like, go, go forth. We're going to build a chain that runs as fast as physics allow, given the hardware gods and where they go. Justin, what are your closing arguments on this whole discussion so far?

Ryan
And what is Ethereum's end game? Where does Solana fit in the picture here? I mean, hearing Anatoly give his closing argument, I think of Anatoly as a pure technologist. I am grateful for all the technology that it has provided. But from my perspective, blockchains are one of the most multidisciplinary kind of creatures that exist out there.

Justin Drake
And it's more than just about the technology. It's about the culture, it's about the memetic forces, it's about economic forces, about network effects. It's about forces that are way larger than just pure technology. Anatoly is building open source software, coming up with new virtual machines, new gossip, new networking technologies. All of these things have little moat.

They can be copied very easily, and they are in the process of being copied on Ethereum Altoos today. What is difficult to copy are network effects, memetic forces, brand, reputation. This is what will stand the test of time. And the network effects are so large that it makes it difficult to have an isolationist policy. We want to build everything from scratch, not invented here.

I don't want it. I want everything to be on my little Solana island, disconnected from reality. And that will mean that fundamentally, you have headwinds, constant headwinds, and this is being reflected. One of the very interesting things is when you have an exotic island, you have exotic life forms that emerge. You have weird things like geo bodens and dog with hats and all of these things.

And you can have circle jerking within this little island where everyone's happy worshipping these little images. But even Solana, which is the primary place for meme coins, is still minuscule. Look at, look. Solana has tens of thousands, if not hundreds of thousands of meme coins. If you take the sum of the market values of all the meme coins is about $10 billion.

On Ethereum, you have one meme coin called shiba inu, which is $15 billion, a single meme coin. And this is where we see the distinction between quality and quantity. Anatoly and Solana is going in for quantity, quantity, quantity, and at the expense of all sorts of things that are more important in the grand scheme of things. So just because I asked Anatoly the same question, I want to ask you the same question. So the end game for Ethereum, what does Solana look like?

Ryan
What does Ethereum look like? Is one larger than the other? How do you think this plays out, Justin? Again, in the endgame, I think it's winner take most. I think the network effects are extremely strong.

Justin Drake
And so there's just like, there's one Internet, there's going to be one Internet of values. There isn't. I mean, okay, maybe, arguably, there's like, two Internets. There's like the chinese Internet behind the firewall and other Internets. But that is kind of a bit of an anomaly, really.

You can think of there being one big Internet with massive network effects. And what I think will happen is that we're going to see a flourishing of activity where the network effects reside and grow. And in order to have these healthy network effects, you need to have synchronous composition, you need to have shared security, and also you need to have shared economic bandwidth. The monetary premium is not just about economic security, it's also about economic bandwidth. The end game for example is to have decentralized stablecoins.

How do you build decentralized stablecoins is by taking pristine collateral and issuing stablecoins out of that. In order to have trillions of dollars of over collateralized stablecoins, we need trillions of dollars of pristine collaterals. And so you can't discount the importance of memetic value. And there's going to be strong network effects where these decentralized stablecoins are issued. And what I think will happen is that ethereum is the most likely candidate of winning it all.

Now, you might ask, if the layer one landscape is winner take all, how can you justify the Solana $100 billion valuation? The way that I rationalize it is that there's a 1% chance, let's say, that Solana can become the Internet of value. And that is enough because there's a 1% chance that a 100 x is from $100 billion to $10 trillion. And so that is a worthy bet to be making. But in 99% of the time, if you were to roll the dice of humanity and you run these parallel simulations, you're going to have a dominant network, the Internet of value, that is going to eat everything up.

And the lead that Ethereum has is so high that a little bit of technology that is easily copyable and that is being copied right now with the l two s is not going to be a sustainable moat. Well, there we go. I think we have to end it there. We've gone on for almost 2 hours. Gentlemen, thank you both for joining us and having this conversation, this discussion.

Ryan
It's been at times exciting, at times heated, but always done with good faith, I think, from both sides. So I certainly appreciate it. Anything else you want to say before we conclude that? Anatolia, you look like you want to say something. No, this is great.

Anatoly Yakovenko
I don't think anyone's mind is changed, which is funny. It's usually how it works with each other. So we'll keep arguing about this. Yeah, that's usually how these go. Justin, was your mind changed at all?

Justin Drake
Not really. I am looking forward to the feedback from the community. I mean, I want to see, for example, if people really buy the fact that paying your taxes and cell pressure is not a cost to the network. That just seems episode. But we'll see.

Ryan
Well, I think there will be some follow ups in community. As always, you can let us know what you think of this episode. Let us know if you want a round two at some point in the future. And maybe we convince these guests to come on and resolve the unresolved questions, but we'll end it there. I gotta let you know.

Of course. None of this has been financial advice. We have no idea what the price of soul will do. No idea what the price of ETH will do in the near term future. You could definitely lose what you put in.

But we are headed west. This is the frontier. It's not for everyone. But we're glad you're with us on the bankless journey. Thanks a lot.