Motion: Today's blockchains can't increase TPS without taking a hit on decentralization (Emre Tekişalp vs. Anatoly Yakovenko, cohost: Uri Klarman)

Guests:

Emre Tekişalp (@etekis)
Anatoly Yakovenko (@aeyakovenko)

Hosts:

Richard Yan (@gentso09)
Uri Klarman (@uriklarman, special co-host)

Today’s motion is “Today’s blockchains can’t increase TPS without taking a hit on decentralization.”

Today’s guests are from two different public blockchains Coda and Solana. The CEO of Bloxroute came on as special co-host. So Coda came to this topic from the angle that, adding full nodes will further decentralize the network but slow down performance. And Solana started by asserting that only the voting nodes mattered because they alone would provide the censorship resistance, which is the key to decentralization.

As usual, the debate then evolved, and covered a lot more ground. We got into things like weak subjectivity and Moore’s Law, and at the end meandered into the topic of nation-state attacks.

If you’re into crypto and like to hear two sides of the story, be sure to also check out our previous episodes. We’ve featured some of the best known thinkers in the crypto space.

If you would like to debate or want to nominate someone, please DM me at @blockdebate on Twitter.

Please note that nothing in our podcast should be construed as financial advice.

Source of select items discussed in the debate:

• Solana: https://solana.com/
• Coda: https://www.codaprotocol.com/
• Bloxroute: https://bloxroute.com/
• Weak subjectivity explained: https://academy.binance.com/glossary/weak-subjectivity
• Coda article on mental model of Scalability-per-unit-of-Decentralization: https://codaprotocol.com/blog/solving-the-scalability-trilemma

Transcript

Richard: 0:08

Welcome to another episode of the blockchain debate podcast, where consensus is optional, but proof of thought is required. I'm your host, Richard Yan. Today's motion is: Today's blockchains can't increase TPS without taking a hit on decentralization. Today's guests are from two different public blockchains, C oda and S olana. The CEO of BloXroute came on as a special co-host. So C oda came to this topic from the angle that adding full nodes will further d ecentralize the network, but slow down performance. And Solana started by asserting that only the voting nodes mattered because they alone would provide the censorship resistance, which is the key to d ecentralization. As usual, that debate then evolved and covered a lot more ground. We got into things like w eak subjectivity and Moore's law, and at the end meandered into the topic of nation-state attacks. I definitely enjoyed hosting it. Now while you were here, be sure to also check out our previous episodes too. We featured some of the best-known thinkers in the crypto space. If you would like to debate or want to nominate someone, please DM me@BlockDebate on Twitter. Please note that nothing in our p odcast should be construed as financial advice. Okay. Without further ado, let's dive in! Welcome to the debate. Consensus optional proof of thought required. I'm your host, Richard Yan, State motion. Today's Blockchains can't increase TPS without taking a hit on decentralization. My metaphorical left is Emre Tekişalp, arguing for the motion. He agrees that today's blockchains can't increase TPS without taking a hit on decentralization. To my metaphorical right, is Anatoly Yakovenko, arguing against the motion. He disagrees that today's blockchains can't increase TPS without taking a hit on decentralization. That is, blockchains can scale without sacrificing decentralization. And then to my metaphorical middle, is my special co-host Uri Klarman, who will be moderating the debate with me. Great to have you, Emre, and welcome back to the program, Anatoly and Uri!

Emre: 2:19

Thanks for having us.

Anatoly: 2:20

Great to be back.

Uri: 2:22

Likewise.

Richard: 2:23

Great. So here's a bio for the three gentlemen. Emre is the Head of business development at O(1) Labs, the company behind Coda protocol, the world's lightest blockchain. Coda swaps the traditional blockchain for a tiny cryptographic proof that tackles the scalability trilemma and makes it easier to develop critical apps that run natively in the browser. Previously, Emre was in business development at Coinbase and product management at Intel. Anatoly is founder and CEO of Solana, a layer-one public blockchain built for scalability without sacrificing decentralization or security. And in particular, without sharding, he was previously a software engineer at Dropbox, Mesosphere and Qualcomm. Uri is CEO and co-founder of BloXroute, a layer zero solution, aiming to solve scalability bottleneck for all blockchains in particular, the solution operates at the network layer and can be proven to treat all nodes fairly in propagating blocks. Uri is an interdisciplinary networks researcher. His specialty includes alternative content distribution networks, trustless peer coordination, and security. As usual the debate has three parts: an opening statement for both sides, starting with Emre. The second round is the body of the debate with me directing questions to the debaters. Both sides are highly encouraged to follow up with their opponent after hearing answers on the other side. And of course, they're also free to respond to each other's points raised during the opening statement. The last round is audience questions selected from Twitter and we'll end with concluding remarks from both debaters. Currently our Twitter poll shows 39% in favor of the motion and 61% against it. That is most believed TPS can go up without sacrificing decentralization. After the release of this recording, we'll also have a post-debate poll. Between the two polls, the debater with a bigger change in percentage votes in his or her favor wins the debate. Okay, so let's go ahead with the opening statement. And the first question is actually for both of you. Can you please clarify the concept of decentralization that we're discussing today? It seems reasonable to say that everything else being equal, you want a network with as many independently controlled consensus nodes that produce blocks, as well as full nodes that validate state. When the motion talks about taking a hit on decentralization, in your mind, which set of nodes is decentralization referring to? So Emre, please go ahead.

Emre: 4:49

So for this motion, it prioritizes, in terms of decentralization definition, full nodes, because that's where we believe the current bottleneck is. However, I'd first like to take a step back and, let me share why we believe the centralization is at least as important as TPS, and why it should not just be a lip-service to a meme that we have in the industry. So if you look at the current value created by public blockchains today, at least 80% of it resides on the two most decentralized blockchains. And obviously the question asks, the definition of decentralization, but even without a definition, we intuitively know that these are Bitcoin and Ethereum, right? And this value also includes all the tokens on Ethereum where, a clear majority of third party tokens reside. And why is this? It's an interesting question. But I'd like to explain it with a quick real life example. So as you mentioned before, my current role at Coda, I was at Coinbase. And, most recently I was heading up the business function of USDC, the stablecoin that Coinbase collaborating on with Circle. And, as we were starting it off, it first started on Ethereum, which was a no-brainer decision. But as we started exploring, growing it and adding it to other chains, the first thing we asked is, are any of them actually decentralized enough? Why is that question? Well, it turns out what makes a stablecoin a stablecoin, and not for example, Venmo, is the ability to tokenize it without having to worry about the security of the platform you tokenize it on. Right? You take it, for the most part, for granted. This is, as opposed to something else where it's pseudo-secure, pseudo-decentralized, let's say again, not picking on anyone, but let's say Stellar, you start thinking of, legal recourses, you may have to take with the centralized entity behind the chain in case something goes wrong, because the probability of something actually going wrong has just increased from infinitesimal in the case of Ethereum, for example, to a couple of percentage points. And those percentage points actually matter a lot from a business perspective when you have billions of dollars of value. So once we accept that it is important, again, the question becomes about, well, what is it? And it is a hard one, I'll be honest. But there's two sides to it, as you said and we've decided to focus at least for this motion on the full node count. Because, well, frankly it is the easier one to quantify and compare apples to apples between different blockchains. It's really hard to compare, t he cost of attacking a proof-of-stake chain versus a proof-of-work chain, which all have different consensus mechanisms but also, that's again, as I said, where the bottleneck appears to be, at least to us, and we did this by comparing various public chains and using real life data. And when you plot transactions per second, as a measure of scale against decentralization or a f ull node count as a measure of d ecentralization, it's funny in that you see almost a perfect, negative correlation, in that, you start hitting TPS limits at different levels of decentralization. So this is essentially the core of this notion. However, the interesting thing to a l so note is that this isn't a technical limitation in that, the more TPS you have, it's not you suddenly hit a network capacity and there's, less full nodes that can connect to the network. It's rather an economic relationship and perhaps the best example of this was with the Bitcoin scaling debate, in t hat, f o l ks didn't want to increase scale because the more TPS you have, the more data you have and hence the more expensive it becomes to run a full node and you end up having less number of full no d es.

Richard: 8:52

So it's interesting that you refer to the chart, the one with the negative correlation between the TPS and the number of nodes, which I don't think was specified fully in the article, but I think that was what sparked this debate between the two of you on Twitter. So Anatoly, would you like to have your opening statement and then address what was being said by Emre?

Anatoly: 9:15

I think the way I approach this is a thought problem: what is the difference between a blockchain and Postgres running on AWS, right? Because you can have arbitrary[inaudible] of safety on AWS. Effectively between Google, Azure and AWS, you can have as much safety as you can possibly have. In fact, the likelihood of losing data is much more tied to your credit card expiring and that account being closed, than the actual hardware failing. So to me, the interesting difference between permissionless networks and the centralized systems is censorship resistance. So what that means in proof-of-work in Bitcoin, is that there is an economic cost to censor a particular transaction. And because this economic cost of Bitcoin is quite high, you're spending hash power, and the opportunity to be a block producer is open and available to anybody with any hash power. It's a system that in a very genius, beautiful construct is eventually censorship resistant. And that is something that I think proof-of-stake networks, including Solana are going to struggle to replicate. And I think that they fail to replicate in the long-term they will fail to be truly decentralized. So if you take censorship resistance is the key factor. Then what you really optimizing for is the minimum set of nodes in a proof-of-stake network that can halt the network or the liveliness threshold. For some, for most protocols, this is somewhere around 33%. I think Avalanche can, can be configured somewhere between 40 and 20. And I think, I don't know what they're going to run their network with, but effectively let's just call it 33%. So if you look at most proof-of-stake networks today and you take their stake weighted nodes and you see what is the smallest set of machines that adds up to 33%, it's not a very large number. IOTA famously only has one coordinator, but Tezos and Cosmos- I think Cosmos takes about five nodes to get to 33%. With Tezos it's seven, and as everybody usually loves to derive EOS, it takes eight n odes to get to their l iveliness th reshold. So in a lot of ways EOS is more censorship resistant than any system of Tozes and Cosmos, even though a lot of folks tend to think of those two has being more decentralized.

Richard: 12:07

Okay, great. So let's move on to the body of the debate. We have been speaking generally about what decentralization means and the state of decentralization for various other projects, but let's get back to the crux of the disagreement. So one of you thinks that in today's architecture, somehow TPS cannot go up without sacrificing decentralization. And if you think about that in the terms of censorship resistance, and it sounds basically you were saying, if you want to scale, you basically have this problem with onboarding new nodes into the network. Right? And I don't think we've actually specified whether we're talking about full n odes or b lock producing n odes, but regardless there's a general problem with onboarding these new nodes when you increase the TPS. And then the other person does not agree with that. So maybe we can start having conversation around that and maybe starting with Emre.

Emre: 13:03

So I totally agree with Anatoly and that censorship of resistance is the single most important thing. And we can focus on or dig deeper on the consensus node side of things. Whereas he said, it's the level of decentralization and proof-of-stake is a bit harder to get compared to proof-of-work. Although I slightly disagree there, just to give an example, the consensus mechanism that we use at Coda, for example, Ouroboros a variant of Ouroboros, which is Cardano what the Cardano team came up with. You can configure it to have, 46 or up to a 49%, I believe, threshold, but without going into that, the one interesting thing about consensus nodes is that those nodes are incentivized to some extent, to keep honest or stay honest right, that's, that's the entire reason by, let alone proof-of-work chains, proof of state chains have at least the large or big enough ones has st ayed s afe up until now. The interesting thing with full nodes, which are on the outside outside of things are, they're not incentivized at all. So that's why it becomes harder economically for these folks to keep running, and maybe to just double click on why full nodes are important. They keep the game theory in check, right? The game theory that makes public blockchains possible requires that there are as many independent folks as possible, that validate the transactions that have been broadcasted and that the correct ones have been actually put into the block by block producers. And the faster you can detect basically any attacks, the more resilient your chain is. In that world, and the full nodes obviously have other benefits as well, but in that world, full nodes are crucial and the more expensive it gets for them, the less resilient censorship resistance decentralization, pick your meme, you have. And that's ultimately why, I think again, I want to refer back to data also because so many of the statements in our industry ends up being hypotheticals. We're just not seeing T PS being able to be increased, whilst maintaining decentralization, at least with today's architectures.

Anatoly: 15:28

So if these nodes are not participating in consensus, they're just processing data, they're just looking at the information on chain. What can they do when there's a double spend?

Emre: 15:43

I mean, they're the ones that raise the alert and raise a flag, right? And say, look, there's something wrong going on here in the network.

Anatoly: 15:52

How would they do that?

Emre: 15:52

By observing transactions in the network. Because they're a part of the peer to peer network of the..

Anatoly: 16:00

But the consensus nodes that are voting or not?

Emre: 16:03

They are as well, but if they're the ones attacking it right then, and if the ones that are honest are not a majority, then it's much harder for these folks to detect that a nd that they can be sure that the way...

Anatoly: 16:18

To attack the network, right, somebody needs to acquire a certain number of keys and then partition Binance and Coinbase, and then double spend in each one. And somehow the full nodes that are part of this network that are voting are not getting partitioned, right? I see a cognitive dissonance that if there's some amount of honest nodes remaining in the network, which is the assumption that every network operates on, then why wouldn't those nodes raise the alarm? And they're not voting nodes, right? They're just full nodes. Then, what does it matter? To me, the idea that full nodes that are not participating in consensus count for anything is moot. But, and this is where our disagreement came from. But I think this debate is can you increase decentralization without taking a hit to TPS? Well, the way you can increase decentralization, right, is by increasing TPS because fundamentally if your transaction layer is fast enough, you can fundamentally use it for consensus like we do. And if your TPS goes up, therefore the number of nodes that participate in consensus goes up. And if the number of nodes that participate in consensus goes up and you can spread your stake, that smallest set that is required to censor their network goes up as well. So, imagine 20,000 machines equally split, that takes about 6,700 machines to hit to that censorship threshold, right? That's 6,700 keys, that an attacker has to take control of, and to spin up an epoch, to start censoring the network. And that is a hard problem, that's hard to do, right? So to me, it's actually, I think, impossible to build decentralization without increasing TPS. And I don't think full nodes come into the picture at all, because they don't actually improve any interesting feature of decentralization.

Uri: 18:28

So I want to jump in here for a second, with a question for both to Anatoly and Emre. So first of all, I'm glad that you all agree that decentralization is an indicator for censorship resistant. So on Anatoly, I think you said y ou w ere saying that only voting nodes really count, and t hey a re the one who matters because if somehow an attacker gets some l arge, let's say 40% of the network or something like that, what would exactly the n on-participant like the full n odes who don't participate in the voting, what would they do? And I think it's not actually accurate two things here. One is that you don't actually need to hack the keys in order to get the majority or something like that. Think that the US government comes up with an executive order saying y ou're not allowed to do X, Y, and Z, or allow to accept transactions that are touching this wallet or something like that. And if a majority of the voting nodes, whether that's POW that's POS are following that rule, there is, Bitcoin works so well because miners are not coordinating, but it is possible to reach the situation where more than the 33% are actually doing something malicious. Like it's not that it couldn't happen, and..

Anatoly: 19:58

I agree.

Uri: 19:59

And I do think, or from my perspective, I'm not entirely sure you're right, that non-voting nodes, let's call them just full nodes, It's easier, so full nodes, who aren't miners and aren't validators, they carry their economic weight, right? It's hard to measure it. It's hard to quantify it. But if all the major exchanges and all the big businesses who actually utilize blockchains and crypto, etc, are not going to follow some chain that doesn't follow the rules, even if majority of the hash power, or even if the majority of stake follow that path, then the minority chain, to an extent, can actually turn out to be the big one or the real one, or I don't know how to call it. So, I don't completely agree with you there. On the other hand, I'd be happy to get your comment on that. But at the same time, I also want to shoot something towards Emre. That is you're very focused on data, which is great. So I'm fair... And you came from Intel, which is also great. So I'm fairly certain, you can evaluate how much bandwidth, how much storage, how much CPU you need to run a full node. And you'll see that it's very, very little, right. Your home computer can now process 3000 transactions per second, it's super easy. And if you have Intel I99, you can do 20,000 transactions per second and 60 megabit per second link is e nough for 15,000 transactions per second, and storage isn't a real issue because A it's cheap and B if you're very limited, you don't actually have to store all the blocks since Genesis, right? You can prune it. You can store the state and only capture only store w ith the last X blocks. So I'm not sure, I'd be happy for your take on that. Who says that the requirement that like, Oh, they'll need to pay more t o r un f ull nodes, I disagree with that. Like, I don't think that is based on data. And I'd be happy to hear both y our points.

Anatoly: 22:11

It's Uri to your, to your latter point. You're preaching to the choir here because that's our whole thesis. That by the time anybody ships anything complicated, the hardware costs are gonna drop by half again. But like this stuff takes so much work to build, but Intel and Nvidia and Samsung are just gonna keep shipping semiconductors. But your first point is actually, I think there's a bit of, the tail leading the dog effect in all of these networks and that when you have something like proof-of-work, which has, I think eventual censorship resistance, because it takes constant economic c osts to block transactions. Right. The effect is that all the folks that are participating in this network t o don't have control of this hash power, simply shrug their shoulders and say, well, we can't actually censor anything. So we have to accept these transactions, right. They kinda shrug at the regulators, shrug at the governments and say, well, whoops, this is, this is truly a censorship resistant network, right. And the theoretical model for censorship resistant networks has existed and the general solutions, since day one, right. It's basically the thing we're trying to build. And if we solve that in p roof-of-stake, and I don't think any network has, the only way I imagine we solve it is we actually in crease t hat set of keys that must be broken to such a large number that these participants that are holding the balances right th ere, clients, the wallets, when somebody as ks h im, I need you to block this. They, again, shrug their shoulders because the network with, 10,000 machines that must be stopped for me to do that is impossible for me to do that. Right? I'm just Binance: how can it go hunt down 10,000 different validators across the entire globe? And that, that to me is the key part, right, is that becomes the cycle of the user s tru st the network to be censorship resistant, and because of that, those balances is a value.

Emre: 24:32

So maybe just to also pine on that point, I fully agree by the way that, further decentralization on both ends is as much needed. I think the main crux of our notion has been that there's been so much more work, thanks to the amazing teams like S olana a nd many others that have been done on the consensus n ode side, that part has stopped t he becoming the bottleneck and the other side, the full node side now has become the bottleneck. And again, the disagreement is basically whether you need i t or not. So let me touch u pon that point that, Uri, you made on the second front. So first of all yes, like if you are a node that's already connected to the network, it's relatively cheap on a home desktop machine that has i99 and has a GPU and as an SSD, yada, yada, the other, right. But we live in a mobile age, a nd you want as many people as possible to be able to easily connect them, become a full news networks, w h ether or not you want to invest all the way. There is another question, but i t is a fact that not everyone has that same level of power. The problem, again, I think further increases when you consider the fact that not everyone will always stay connected to these networks. So yo u've a lways ha d t his issue of folks dropping out and then, ha ve t o c atch up in between. And just to give a real life example there, J a m eson Lopp, I think he runs these tests every year of how long it takes to sync a full bitcoin node from scratch. And I think the latest one was six hours again, using a$2,000 desktop machine. It's not terrible, but not everyone has six hours on there and a$2,000 desktop on their hand. And then on the other side, whether or not you need actually to sync the entire chain, y es, I think there's many different points of view here, but, the notion of l ight clients and SPV nodes, etc, has b een a topic probably since Bitcoin existed at this point. And it's still not concluded, r ight? People still condone. And again, this is more or less school of thought at this point. So w e can definitely argue about this, but that full nodes are required for a trusted access to a chain. And trust is a, in our society these days is a n eroding concept. So we are of the belief that if you can, it's best to be a full node on th e n etwork. And one last point, o ne of my pet peeves is this notion of Moore's law so, being someone who h ad t he past life was living and breathing, this notion being at Intel Moore's law future is uncertain in my opinion, right? Again, this is another school of thought thing, but we're about to hit, t h e limits of physics. An d e ven before that, t h i ngs are not doubling up every year or every other year or ev ery two years. So that's also another thing that I believe is not, is not for granted. And thus, we should just do our best to be able to using, some more breakthrough technologies to be able to ensure as many full node as possible.

Uri: 27:45

So if I just take a moment to conclude the two points that you brought up, starting from the later one, and you're pretty much saying that, yes, you can do that. And it is, it isn't terribly hard to achieve that. But if this, if we want this to be as decentralized as possible, well, now people have smartphones and you want them to be able to do that on the frontier always keeps getting further and further. So if we want real decentralization or decentralization as possible, then yes, maybe you can, maybe it's not an issue on desktops or not a major issue, but now we actually want that more mobile. So it's even more decentralized. And Anatoly your point was that yes, we are. It's not only that the voting nodes are important. It's also the non-voting nodes that you accepted. They are also substantial and important, right?

Anatoly: 28:47

My point was that without censorship resistance in the core level, the non-voting folks don't matter. And this is a tail wagging the dog. But the reason that they don't even have a voice a Bitcoin is because Bitcoin is eventually censorship resistant. But for that to be true in a proof-of-stake network, you actually have to increase the censorship resistance set of the voting nodes and the proof-of-stake network. And the only way to do that is t o increase TPS, f undamentally, because those things depend on how many cryptographics messages per second that verify from everyone e lse i n t he network.

Uri: 29:25

These are both good points.

Anatoly: 29:29

I think we basically agree. That's the fun, that's the joke here. We need more decentralization and you guys are attacking it from well, if everybody is full node, that's a lot easier to achieve, Right? We are trying to increase the number of messages per second that the consensus node can verify, right, and the end of the day, it's the same thing. I think one notion that you mentioned is this idea, do light clients matter or not? And this is I think the fundamental debate between weak and strong subjectivity and full node verification is a requirement for strong subjectivity, right? You have to verify the entire ledger from the start to the finish when you get to that result, that this network is sound, I don't know if this is going to be possible in Solana, ever. Right? So we've given up on strong subjectivity from day one and took the full approach of weak subjectivity, where anytime somebody joins the network, they use TOFU- Trust On First Use. And as long as they maintain that connection, right, they can continue running it. And then light clients, and SPV, and all this other stuff is the fundamental way, how everyone else reads the network. So I think there's interesting differences there, really fundamentally philosophical differences.

Emre: 30:54

Yeah. I think Anatoly hit on a lot of good points there one of our core beliefs and that's of the entire Coda project, as well as before my time has been that, cryptography is increasing at a pace specifically zero knowledge proofs and that, and achieving[inaudible] has gotta be possible without taking too much of a hit on performance. We are seeing some empirical data there, a lot, literally a hockey stick growth in the performance of these proofs. But it's also interesting that obviously, since you're adding compute o v e rhead, b ecause the y're cr yptography, the mo r e ad v anced op erations, one without such a baggage will always be faster, but it becomes, then, the question of, is the lat ter on e decentralized enough, to be able to support the most-- I don't want to say-- valuable but for a lack of a better word-- the use cases with the most dollar value attached to it, at least in terms of transaction value? And there, I still believe, taking some hit on TPS for further decentralization will be important. And again, I love, as Uri said sticking to data if possible, that's what we'r e see ing today, right? From again, EOS and TRON, and some of these other networks are obviously of a d ifferent nature. But the y just haven't been able to create as much value as the two dece ntralized cha ins.

Richard: 32:26

So I wanted to touch upon one point that Emre mentioned about Moore's law. So I know this was also touched upon in our previous debate Anatoly, and I think this is also related to the concept of hardware decentralization that you mentioned on Twitter. Can you elaborate on that point in the context of today's debate?

Anatoly: 32:45

People typically talk about Moore's law, right? Some people get confused and thinking about frequency scaling, where the speed of the underlying transistor technology gets faster and faster. The reality that it's just how many, how much silicon do we have available to us for compute is the crux of it. And Moore's law in that sense is not going to stop because the wafer sizes are going to increase. The yields are going to get better. The number of chips that they can stack right now, vertically in a single[ inaudible] is going to increase. And those optimizations are going to continue, growing at a pace that they've been going up, I think the end of my children's lifetimes, right? I think my grandkids might not see the end of that. Because those all fundamentally increase the number of context-free writes, computation that doesn't overlap in state, right? How many single instruction, multiple data stuff, can we do? How many vector dot products we can do? Literally if humanity stops figuring out how to scale vector dot products, we should all be working in bunker coin, because this is where we effectively predict the future, right, and optimize our decisions and economy and everything else that we do. So from my perspective, I think Moore's law and that sense is going to continue going. And that is very much applicable to consensus and smart contracts platforms, because we've demonstrated, we can easily define this problem as a set of non-overlapping operations, right. Over a common censorship resistant messaging layer. So you have, our capacity to handle the nodes is based on non-overlapping computation mostly. So it's horizontally scalable. So a s core counts increases, number of lanes and the CPU cards increase, we can basically add more validators so we can increase t his censorship resistance, to the limit. And I think that's g oing t o continue indefinitely. And this a pplies to some extent, t o both wire communication, fiber and memory bandwidth, and even wireless. So 5G is this horizontal scaled, 4G sort of, because it uses, d ozen different connections at the same time and we'll d o path routing. And these approaches are going to continue, right. People are g onna go from, I don't know, what's DDR5 right in terms of memory bandwidth, right. It's double a DDR4.

Emre: 35:31

Yeah. So I won't go into the Moore's law debate. It's probably as, if not more discussed, when ETH 2 will ship or something like that. But, and I do agree, if Moore's law ends up being a thing, then we have much larger problems as a society. That being said, I think it's not the right framing for where we are as an industry, because it's not we're all happy and sufficient where we are, as long as we can double, let's say, T PS or any sort of performance metric for a blockchain, as long as we can double it every year. We already have probably 1000x or 10,000x or whatever leap. I mean, Solana has awesome TPS performance numbers, but again, in our perspective of being, at a per decentralization un it o f decentralization level, we have thousands of X of improvements we believe we need to make. So it's not we're happy with how much hardware capacity is available. And that's why we believe Moore's law is not the right level of framing here. When we get there, sure, bu t we have just so much more to catch up than being satisfied with doubling every year.

Uri: 36:45

So, first of all, I have to say, it's so much fun to sit on a committee discussion with someone like an ex-Intel guy, an ex-Qualcomm guy. You know what you're talking about. It's super refreshing when you're talking about Moore's Law and is this the right framework to think about, Is it not. I want to ask the same question, but from a different perspective, let's assume that increasing TPS hits decentralization, is this all relevant? Yes. You want me able to do the trillions of trillions of transactions that you could do on AWS in a decentralized setup, but if you could do just billions, maybe it's enough, maybe just a million, which again, is it percentile of a percentile of a p ercentile, maybe that's enough. So is this discussion at all relevant? Can we achieve enough with a good level of decentralization? Is it really worth discussing? Is this a theoretical question or does it actually a ffect the real blockchains that we're helping design here?

Emre: 37:46

I have a short one there and then I'll let Anatoly... But I think it's not we're asking, Oh, wouldn't it be so great if we had just twice performance improvement, right. We're asking today a nd h ave been for a number of years, for so much more so that's why I believe, we're not at a hardware limitation, we're not bounded by hardware, we're bounded by the fundamental architecture of how these things are designed i n that, one other pet peeve I have about our i ndustry i s that we call these things like, Ethereum calls itself, the world computer, it's a terrible computer. So maybe that's the wrong way to think of them. Maybe they're not computers, maybe, folks have called t hem trust machines. Maybe they are a different type of machine and we n eed to think of them differently. But yeah, that's my short spiel, on this topic.

Anatoly: 38:37

I guess, fundamentally I am trying to brute force this problem because you guys can imagine a difference in level of security, if the set of machines that we need to break into is, eight or below, like EOS and every other proof-of-stake network or 2000, right. If we actually have a globally distributed set of nodes that have very tiny stakes, right then the 30th and minimum 33%, it takes 2000 machines all over the world for you to fly around it and go break into for that network to be just, to have a liveness hit, that network has achieved a level of security that I think it would be foolish to use anything else if it was fast enough. And I think the only way to achieve that level of security is again, by driving it into Moore's law, where network cards and bandwidth and memory and CPU's continue scaling and leveraging that to do more cryptographic operations per second, just for managing the consensus side. And this isn't even, I'm not even talking about TPS for users. What I'm talking about is figuring out how to create this censorship resistant set of computers at such an extreme degree that, it is impossible, no, it's basically, the odds of breaking into that thing are like the odds of the universe go absent. How do we do that? Right. To me, it seems like the only viable way to do it and truly maintains security is just by using hardware because it's fast enough for a lot of things already, right? For our network, we generate two and a half messages per second per validator. So if you have 50,000 TPS, which you mentioned would roughly be like 150 megabits Mbps, which is widely available everywhere. That's 20,000 computers. We can get them all equally s tate, which is a, the hardest problem. I think the social impact of that, how do we get all these people that stated their validators and ended up with this high concentration of wealth a nd just the top floor o r five machines, how do we actually get them to spread it out across this entire set? But that is, to me, the harder problem to solve. We can solve that-- then we have the world's computer. And it runs like, what, 200 megabit connections? Which is basically nothing. But that's how I view this problem. And I think that if you take the step away from proof-of-work away from pure, objectively measured, like"we w ill start from the Genesis and we compute the entire state, and this is the social truth," if you take that leap, you basically have to go all the way to where w e're at. You have to build something that is so censorship resistant, that is unfathomable for it to be broken. And to be fair, I think zero knowledge proofs might be a path there as well. So it may be possible that the Coda approach is just as viable, because I can imagine a world where you have 100 million iPhones that have a copy of the ledger, right. On the zero knowledge proof. And even if they're not all communicating to agree on this, the fact that those copies exists and the statistical likelihood of all of them being correct is so high, that in itself is a measurable censorship resistant network. Even if the consensus mechanism isn't leveraging it, that is an interesting experiment to run. Anything can happen.

Emre: 42:42

Yeah. I'd say maybe, one of my favorite parts of this debate is that both projects come at the problem from the exact opposite ends. And I think Anatoly probably goes to that point you made about weak vs strong objectivity and then how it implies censorship resistance. Yeah. So it is indeed interesting. And I think what will be interesting to see is, which of the two approaches accrue more value, not in the network valuation sense, but perhaps more in the aggregate value of different applications that are created on this platforms, or the t o ken a ssets, etc. That's one thing I'm looking forward to see.

Richard: 43:26

Okay. By the way, Anatoly, one thing you mentioned on Twitter is that Solana takes a linear hit on TPS for note counts. I feel this translates to user TPS capacity goes up if node counts go down, does this not go to show that TPS comes at the expense of decentralization?

Anatoly: 43:44

Yeah. But the capacity goes up exponentially every two years. Right. So if it's a linear hit versus an exponential, it's kinda insignificant.

Richard: 43:53

Right. This goes back to your argument about how on the back of hardware performance increase in general what we just talked about that effect will be trumped by the hardware performance increase.

Anatoly: 44:05

Yeah. Even at the current state of affairs, right? The 5G rollup is going to put a one gigabit device in everyone's hands, right. Globally. So the backhaul to mat to handle that is going to be enormous. And the fiber to the data centers that support all of this stuff. And the interconnect across the entire globe is going to just dramatically increase in bandwidth that is the more constrained resource. In the messaging side of consensus. I think compute is more or less already insignificant. We went from using 4 GPU cards 1080 TI to 1 in the span of two years because there was both the architecture improvement and process improvement.

Richard: 44:52

So maybe just the reel back a little bit, if I may summarize the issues that you guys have touched upon in relation to the debate topic today, I feel that the main disagreement is still in whether decentralization is a matter of censorship resistance for block producing nodes or the democratization via onboarding as many full nodes as possible because in Solana's approach, it just seems that there's going to be limits in terms of hardware for the full nodes, to be able to get on board because of the hardware constraints. On the other hand, the Coda approach, which basically says that you can run a full node in a phone browser that would obviously boost the amount of full nodes very, very easily. So am I right in summarizing the crux of the disagreement right now being whether decentralization is a concept of just the censorship resistance concept in relation to block producing nodes, versus whether there's a need to expand and include as many full nodes as possible?

Emre: 45:59

I think so. And one other thing I would add is full nodes also helped with, just being able to give state to any user that wants to use the blockchain. Right. So just think of an example of today's world and Ethereum, where, take your favorite dapp. And you're an average user, who just has metamask and doesn't otherwise know how to use a box, and you just want to get that sweet interest rate from a DeFi protocol. You essentially connected the chain and get your state from, as a whole sorts of a suite of middleware providers, whom all are like these full nodes and even worse, archive nodes on the chain. One benefit that we believe the succinct blockchain approach with which Coda has brings is that, every single one of those users can trust us the, get their state from the chain, just because they have to download the small proof. And that's how they're able to be able to a full node. So not only does it again in our approach of having stronger objectivity, provide much better censorship resistance, but it also makes the whole, maybe not the UX because that's usually shrouded from the end user, but especially for developers, the whole developer experience just becomes easier because they're able to provide a better UX ultimately to their end users.

Richard: 47:23

So Emre aside from Solana, what are some other chains you're looking at right now that you don't have the capability of adding significant amount of full nodes without compromising the TPS.

Emre: 47:35

Barring the ones I'm not aware? I think this is an issue or a challenge, not an issue, but a challenge. If you want to have strong objectivity that every other blockchain has or every other traditional blockchain architecture has, because the fundamental issue is about becoming a full node, right? And the only way to become a full node to download the entire chain and verify it yourself, whereas with a succinct chain that uses zero knowledge proof to do the same thing, the beauty of it is that there's only that that verification is only done once and you have approved for it more like a certificate that says, Hey, this verification was done correctly on here you go. You can check it for yourself. Right. It's, we, we do see other teams trying to tackle this problem of becoming a succinct blockchain, but to our knowledge, Coda is the furthest ahead so in that case, that's why we're seeing, this as a different approach.

Anatoly: 48:34

So Coda is still proof-of-stake based, right? So you guys did dip your toes into weak subjectivity because you don't know if I have a long range attack, that's generated multiple versions of this full node.

Emre: 48:46

Yes and no, I'm in, that's the variant of Ouroboros we're using, has formal proof, against this l ong r age attack a problem. I can't fully explain it myself, but yeah.

Anatoly: 49:00

I have a super majority of the keys at any points of the chain, right. A super majority of the discarded keys. And I can always generate a different diverging for it. But if I have a, there's no way to work around that problem.

Emre: 49:20

In a proof-of-stake chain do you mean?

Anatoly: 49:20

Yeah. As long as if I am able to get a super majority of the stake weight at some point in time with a network of old discarded keys, then I can generate a divergent fork, from which is indistinguishable. So, what is measuring the weight of any of these things, right? And consensus is, for proof-of-stake networks, is weighted cryptographic signatures, and proof-of-work it's real electricity. If there's no physical energy being spent to secure the network, then it is somewhat weakly subjective.

Emre: 49:51

Yes. I believe in a fork case you are correct though. I'll be honest, I'm not the consensus experts in our team. So, may have to go back, but yeah, maybe you're right. It's not entirely like strong objectivity in that, for example, assume a proof-of-work chain version of Coda. Yes. That would be maybe bulletproof. But the guarantees that we get out of Ouroboros have given us comfort that, a full node that's verified via ZK-SNARK, gives us very similar properties.

Anatoly: 50:26

This is like, where I think the rubber hits the road, right. Is if you have a kajillion light clients out there that may have a full copy of the ledger and some other corrupted, right. And now there's the version full copies of this thing. What i s the social truth?

Emre: 50:45

Right. I see what you mean.

Anatoly: 50:47

You're gonna, you're gonna go to the consensus nodes and ask him, so at the end of the day, right, you have similar problems, like any proof-of-stake network. And I'm not sure that a full node adds that much additional weight versus a partial node that just generates some fraction of the state. Right. Because Ethereum or anything, or us included, if I have a particular application, right, one of our clients-- they filter out the transactions they don't care about, they only look for the ones they do and they maintain their own copies of effectively a state machine, and index them and do whatever they want with them. So if somehow we were under, we get attacked and every key that's out there, that's in consensus gets destroyed, when they do state reconciliation with a chain they'll stop. Right. All the exchanges will halt.

Emre: 51:48

Yeah. So the way we actually prevented a majority of that problem is, Coda has this notion, which we call a transition frontier in which, the canonical proof, which we identify as th e c apability to become a full node, is provided after, X number of blocks after which there's like 99, like couple of nines of guarantee of settlement. So we yeah, we prevent a bit of that issue, but yeah, I'm not go nna s ay it's a hundred percent be cause t o your point, y e a h, proof-of-work chains are the only ones that can provide that. However, w e believe this transition frontier concept provides much of the benefits without having to burden the user with having to fully download the entire chain.

Anatoly: 52:45

It does a full note copy, provide over somebody that's just storing the hashes, the Merkel hashes of the entire chain.

Emre: 52:55

Well, the latter is what is identified as a SPV client, essentially, right? In that you just provide, you just prove those Merkle roots, but are unable to verify the, in each individual transaction yourself, which to be frank and they used CXO model matters more, but it's also important in and account based blockchain.

Uri: 53:15

Anatoly, you may be for, for our listeners, I think because this is going back and forth. Could you make broad yet concise argument rather than in small questions?

Anatoly: 53:27

I think full nodes are by definition, full nodes that are not participating consensus, don't provide additional security because when that hits the fan, you end up with trusting and reconciling the state amongst the consensus nodes and the financial parties involved. If you had USDT on your platform, the real source of truth is going to be the ledger that USDT uses to reconcile against her network. They're not going to care what the client nodes say, if they contradict what you, as USDT has, they're simply not going to care.

Uri: 54:09

So let me just play the devil's advocate here, or maybe clarify some of those things you're asking, well, what's the difference if you have just a full node or just an SPV or a light client, which just keep the hashes of everything. What if you're thinking about the full nodes? Well, the full nodes are following the rules and if they receive some level, when you block signed by all the stake or the necessary, like the two thirds, and that has a double spend or something you illegal or a spending without the right signature or something like that, if they would alt they would see that and would not accept it at all because they can see what the rules are and what the rules are. If you're just following and grabbing, I think the state, without being able to see what's really happening, you might not be able to see that?

Anatoly: 55:05

Okay, here's a realistic attack vector, right? You have somebody partitions, Coda, and you have Coinbase and Binance on one side and a bunch of random users, a hundred million light clients in the other. And they partitioned the network and get a hold of a super majority of the keys. And they produce divergent blocks and Coinbase and Binance gets one set of state and everybody else gets the other, what is going to be the real Coda a week later after everybody figures out what happened?

Emre: 55:33

I think the benefit of yours is not in obvious partitions or obvious attacks. It's more in, controversial ones. Where in a similar to what you was saying earlier, I believe it's, it's giving the users the ability to choose which, which partition they are on. Right. So yes, in an obvious one where, there's hundreds, the majority, clear super majority residing on exchanges. If that's the case, then yes. That one wins. But what happens in a more controversial one, the notion of everyone being a full node, it gives people the choice. If it's the other way around, then you're basically have to go with whatever, if you're an SPV node or, just a normal, yeah. Not even a node, you don't get to pick.

Anatoly: 56:17

That's a very fair point, but I think that matters more when you don't have Coinbase or Binance on one side on the same side, if all the exchanges are equally split, then which one wins, it's probably the one with the most active users, which is a very weak subjectivity way to decide.

Emre: 56:37

Fair. Yeah. There is a meta subjectivity topic there that's more on the social layer, but, yeah, I think the different approach philosophically that we have is maybe doesn't apply i t to the world today where a majority of the tokens o n a, on a blockchain sit on exchanges, but, in a future utopia, which we're all going for, where everyone is able to independently j oin these networks a nd do whatever they want. I think it becomes a bit more important th an t hat case.

Richard: 57:04

Hey guys, I have a naive question. What is the significance of Coinbase and Binance in Anatoly stylized example?

Anatoly: 57:11

That's where the majority of, of wealth is stored, right? My point is the meme of that Tether is going to decide the real Ethereum, right? That it doesn't matter what the Ethereum community does. The real Ethereum is the one that has USDT on it.

Uri: 57:26

Is that really the case? Do you think everybody would agree with that statement?

Anatoly: 57:30

This is again a thing of the tail wagging the dog, right? I think the community needs to build something that is viable for us to migrate to. And until that happens, you're going to have this very hard migration of, from Eth1to Eth2, and that's going to be very slow and painful.

Emre: 57:53

But that is an interesting point, question actually, when you think of it again, a hypothetical example, Tethers bank accounts are frozen and the only way to either the company can unlock them is they say, we agreed to freeze half of the accounts because they're doing things that government doesn't want then. And if there's a fork of Ethereum, right. W here actually this is a bad example t hat I think of it because the money is actually in the bank account. So, but think of this in the Maker DAO case, right. DAI gets frozen because, the government takes over. Then it wouldn't be the case that, the fork that the majority of the users prefer ends up being the winning fork.

Anatoly: 58:37

Yeah. I mean, for DAI's case, if it works depends on the value of Ethereum. So the Ethereum fork that wins I think is going to be the one that exchanges except for ETH, right, And are able to transact that fork with USDT. So I think to your point, I think it is, even though it's very easy to poke holes and say that exchanges are these big financial institutions aside. I don't think they actually do. I think the protocols and the constructs and the social force of knowing that everybody has the same full copy and understands this particular state machine to have concluded that is the source of truth. I think that's what pulls it all together. And obviously if there is a way to implement this zero knowledge groups, I think that's a very powerful tool. I have no doubt will succeed.

Richard: 59:34

Awesome. So let's move on to the audience questions section. We have one question and it's sort of unrelated to the debate today, but it's still relevant for the blockchain space. So this person is asking, what are your thoughts on the impact on the notoriety of crypto from the Twitter attack that just happened?

Emre: 59:52

I think there's two views. One of them is that despite this clearly being a hack of, out on Twitter and then just to hackers using the soundest money, they think there is the media portrays it as a crypto scam. Right. And it's just, I think shows how early we still are. Friends were asking me. And one exemplifies is if a thousand years ago, a year ago, like the Genovese bank was, broken into with people, call it a gold attack? Probably not. It would have been a bank run. So that's one, one perspective. The other one, I think is just interesting, but, any publicity is good publicity. Even today after, the hack is not resolved obviously, but, Biden, Joe Biden was tweeting. I'm not going to ask for a Bitcoin, but please donate to my campaign. So people, hear more and more about Bitcoin. So in a way maybe it is a good thing. But I think overall, this is still showing that how much the entire space has still to seen to be seen more legitimate and as a, thing that is here to stay.

Richard: 1:00:59

Okay, great. How about you, Anatoly?

Anatoly: 1:01:01

I would largely agree with Emre. I think this attack just seems really weird that somebody that had access to all of these accounts would do this, the lowest bottom theater form of a crypto scam when they could have done a million things. Yeah. It's a dumb thing. I don't know what it says about Twitter, and Twitter probably being our truth-censorship-rersistant network, the only one that we have, and that may be the way we resolve forks is whoever wins a Twitter battle, right?

Richard: 1:01:41

Well, for me, I think the issue actually brought to light this other threat to crypto and Bitcoin in particular, and that is nation state attacks. And this was featured on one of the previous debates. So Joe Biden was tweeting about it. Politicians were paying attention to this and there's this whole narrative about how, if Bitcoin did not exist, untraceable criminal activities like this would not be as easy. So the counter argument there is, well, Bitcoin, isn't the problem, but criminals are the problem. Right. But I think the issue here is that Bitcoin seems to make it easier to hide the criminal activities. At least that makes it really easy for the government to basically garner political capital and public support, if they want to attack Bitcoin for their benefit. So I'm really curious from you guys' perspective, the projects that you guys are working on, what additional censorship resistance are you offering that is different from the kind being provided by Bitcoin and how does your project guard against this nation state attack?

Emre: 1:02:44

Yeah, so the one angle we have is something I alluded to earlier in that, c urrently the way many people use Bitcoin or any other blockchain, right, is to use the centralized gateways, the easiest ones being exchanges, such as Coinbase. But, even if it's not that, yo u use a wallet provider or you go to a website to sign the tr ansaction from very few people, actually right, connected to ne twork and, e i t h er via becoming a full node or SPV, an d, send the tra nsaction or host their wallet there. So if, as you said tomorrow, the US government decided that, all r igh t, it' s illegal to provide Bitcoin services to anyone in the US and then how do these people access to network? Not a lot of people will be a ble to become a full node and then, just conne cted with it directly. So, and in this case, even SPV nodes I have an issue, right, because they are by definition providing a Bitcoin service. So they are actually doing something illegal, which there'll always be, folks who figure out how, but it, it just becomes much harder. So in that way, Coda, what we're working on, enables folks to not have to rely on the centralized gatekeepers. So there's a clear benefit there, which we have fortunately right now, not seen a dire need to because there's, there's no set, no such legal issue. And if it does end up being a case, we have a lot of more problems than, Coda being a solution to this but yeah, we believe that this does indeed provide an extra added benefit in terms of censorship resistance.

Richard: 1:04:23

Okay, great. How about you Anatoly?

Anatoly: 1:04:25

I don't really worry about state-level attacks. I honestly think that that's just not a problem.

Richard: 1:04:33

Why can you elaborate on why you think is not a problem? I mean, from there's incentives for them to attack and there's means to attack.

Anatoly: 1:04:40

Why?

Richard: 1:04:41

So we were going into a different debate now, but the incentive would be, there's AML and terrorism, funding of child pornographers and so forth. And then obviously there's also the threat to the status of Fiat currency, undermining the government's capability of controlling the monetary system. So there's, that's the incentive. The means would be, if we were talking about Bitcoin, then that would be, say, seizing of mining facilities, prohibition of on-ramp off-ramp and also just political narratives, basically Donald Trump Tweeting about how Bitcoin is facilitating all these terrible activities. And therefore basically using his rhetoric to drive down price, which would affect security.

Anatoly: 1:05:28

All those attacks are much more complicated than the state has the tools that it has his disposal to find the bad guys, if they really wanted to, right. If the United States government thought that somebody there was enough illegal activity occurring in Binance, how long do you think Binance would stay operational?

Richard: 1:05:50

So you're basically making the argument about the competence or resource of a government?

Anatoly: 1:05:56

In a national security sense, Bitcoin is not an issue for the United States, no matter how much dollar denominated value accrues. Because the dollar is what it's valued against, and it's effectively no different than Amazon stock or gold. None of that matters, right? I think the US government, if they wanted to, could find whoever they want, if they're using Bitcoin, it's even easier because they can effectively have cryptographic proof that that's the person that's doing the bad stuff, right. It's an immutable public ledger. I think it's the dumbest way to do anything with illegal financial activity, at a large scale.

Uri: 1:06:40

Can I question that? Let's imagine, but to any country suffering a major crisis to its local currency, if people moving to alternative solutions, black markets, all these kinds of things, or smaller economies of less developed countries, which are slowly being eaten by alternative currencies, which are better, it's fairly easy to think of a scenario where state actors, not necessarily the biggest States, but even smaller States have a lot of resources at their disposal say: Oh, we want to stop this from happening. Whether that to directly attack whether we want to prevent our people from using it, these kinds of things. And if you think of cryptocurrencies is similar to BitTorrent, well, BitTorrent was only, threatening the large media companies, etc. So it was a failed attempt to take it down or to stop, illegal streaming, etc. But the stakes here are a much higher. I don't remember who it was, who tweeted about saying, countries that went to war for much smaller things. So I'm a bit surprised that your approach that: Oh, I'm not really interested or think state actors.

Anatoly: 1:07:59

So is Venezuela going to attack our validators inside Europe and America and China?

Uri: 1:08:05

A different example, think of Egypt before the Arab spring. But I don't know if you're familiar with the details, but just generally speaking, what's happened is that there was a lot of arrests over the internet and forums, etc. And then Mubarak decided to try to shut down the internet. And what happens then in Egypt, which is super interesting regardless, is that, well, people couldn't complain about it, on the internet. So they went to the streets, so that didn't work well for Mubarak, but generally speaking, think of the great Chinese firewall, think about countries shutting down their own internet networks, preventing people from talking with outsiders, which means that if you're participating in the consensus, you're being thrown out of the consensus, not necessarily bring the system down, but it does partition it. And if you're thinking about China specifically, or really any major economical center, you'll have big chunks of the network, which could be shut down. So I'm not saying that they will come for Coinbase and try to take it down, I don't know, Ethermine or something like that. But attack vectors are there. And it might affect not only the consensus, but users.

Anatoly: 1:09:21

I think the small countries will see any of these networks as an extension of the big powers. They're effectively, any use of Bitcoin outside of the United States, as equivalent to using the dollar. Because at the end of the day, what do most people do with it, is they dump their local currency, get Bitcoin because somewhere it can be changed into dollars, or yen, for some extent or euros. Right? So I don't, I don't see those kinds of attack vectors of a concern of the big powers. And all the validators and all these networks are basically running inside US, Europe, China, right? So, effectively, I'm just not concerned. All these powers have much more effective tools to get to whatever they want, than trying to go censor a validator, right? Are they going to go after Bison Trails?

Emre: 1:10:24

But Anatoly, I think I disagree on that as well, with same angle as what Uri said. And that it's about access, not about consensus attacks. And this happens, right, like countries in monetary regimes where the country zone fiat currencies having whenever is having challenges, o ther foreign currencies, the popular ones like USD becomes illegal. And yeah, you're right. Maybe it's not that, they care about Bitcoin because ultimately people want US dollars. But if the easiest way to get us stars is Tether on Ethereum, then, who says they won't shut down Ethereum? Not shut down, Ethereum, sorry, but shut down access to Ethereum. Right? As I gave an example, like just passing a law that says no company on our borders, inside of our borders can provide service for this. In that case, this becomes the needs for a network, more li ke the Bittorrent becomes, I think more pressing because again, the reason Bittorrent was resilient was that, a nyone could connect to it. All you nee d is a software client tha t yo u can get from anywhere.

Anatoly: 1:11:27

Light clients will serve the purpose just as well.

Emre: 1:11:30

But, but again, light clients, again, you need to get data from others, which end up being service providers, in which case they are also doing something illegal, which I agree is a bit harder to crash. It's a bit harder to identify, but it's still something...

Anatoly: 1:11:45

Outside the borders, right? Is Egypt going to shut down Infura? Or every other version of Infura? Every other provider? If it's on the internet, there's a way to connect to it.

Emre: 1:11:57

Well, it depends on the size of the country, then they can put pressure on whichever country is hosting the service. Yeah. So it just becomes a tail risk.

Anatoly: 1:12:05

If you're not in trouble with US, Europe or China, It doesn't matter. Nothing else matters. If you're in trouble with one of those, nothing will help you. So this is why I'm not worried about those effects, right? I just don't care.

Emre: 1:12:21

Yeah. I think, again, this becomes more of a matter of global politics. In which case my view is the world we are used to is, is rather fast seizing to be the reality. And, trust in the order of things in institutions is degrading. So it's better if we're able to provide direct access to these technologies to each individual. But I agree that that's, that may not be there today, but I think again, a worldview discussion is that my view is that we're getting there in an uncomfortably fast pace.

Richard: 1:12:56

Okay. That's actually a great segue into our concluding remarks. So, it's been a fascinating debate. Thank you guys, both for coming on and thanks Uri for being that gracious cohost. So maybe let's get you guys the concluding remarks and just say a few things to synthesize your thoughts. Maybe some things we picked up from your opponent during the debate, starting with Anatoly.

Anatoly: 1:13:18

I think that Emre makes a really strong case for the duplication of data o f full nodes i n the actual replication a bout network effects that maintain a level of security and censorship resistance outside of consensus. And that is something that I think shouldn't be overlooked. So I will definitely... I didn't think that before this debate, but talking to him, I have much stronger c onvictions that it's possible to achieve those goals.

Richard: 1:13:56

Okay, great. How about you Emre?

Emre: 1:13:59

Yeah, I appreciate that Anatoly from the same perspective, I think what's been new to me is that Solana's perspective of, hey, look, if you are okay with weak subjectivity and a more pragmatic view of the world as it is today, then it is possible to achieve way more performance blockchains then, without having to necessarily require perfect decentralization at the full node side of things.

Richard: 1:14:31

Okay, perfect. Uri any concluding remarks from you?

Uri: 1:14:35

All I want to do now is crabbing Anatoly an Emre, and grab a beer with them and talk about like the networking layer. This has been great fun for me.

Richard: 1:14:43

Okay, perfect. Well, thank you for your insights on Emre and Anatoly. Maybe we'll have you back on the nation state attack debate at some point. How can our listeners get in touch with you?

Emre: 1:14:53

Yeah. On our end, my Twitter is@etekis or my last first last name. And f or more on Coda, they can visit c o daprotocol.c o m.

Richard: 1:15:05

Great, Anatoly?

Anatoly: 1:15:05

I am@aeyakovenko on Twitter. But, Anatoly@solana.com if you want send me an Email, or just go to salona.com to find anything and everything you need to know about our network.

Richard: 1:15:18

Perfect. And Uri?

Uri: 1:15:20

Just Google, Uri and BloXroute then you'll find me either on Twitter or on our website. It's easy to find.

Richard: 1:15:26

Perfect. Well, thank you guys. So listeners, we would love to hear from you and to have you joined the debate via Twitter. Definitely look out for the vote in post debate poll. Also feel free to leave your comments on Twitter. We look forward to seeing you in future episodes of the Bockchain Debate Podcast. Consensus optional, proof of thought required. Thank you, gents. Thanks again to Emre and Anatoly for coming on the show and thanks again to Uri for co-hosting. What was your takeaway from the debate? Don't forget to vote in our post debate Twitter poll. This will be live for a few days after the release of this episode and feel free to say hi, or post feedback for our show on Twitter. If you like the show, don't hesitate to give us five stars on iTunes or wherever you listen to this. And be sure to check out our other episodes with a variety of debate topics: Bitcoin's store of value status, the legitimacy of smart contracts, DeFi, POW versus POS, and so on. Thanks for joining us on the debate today. I'm your host Richard Yan. And my Twitter is@Gentso09. Our show's Twitter is@BlockDebate. See you at our next debate!