S16 E41: Yonatan Sompolinsky on Bitcoin, Kaspa & Proof of Work

0:49Vlad CosteaWelcome to the Bitcoin Takeover Podcast. This is Season 16, Episode 41. We are live with Yonatan Sompolinsky, very proficient and at the same time prolific mathematician, who is best known, at least before launching Kaspa, for being quoted in the Ethereum whitepaper for his GHOSTDAG protocol, which is aspirational or was aspirational for Ethereum at one point. And he was also involved with Bitcoin early on. And I'm going to let him talk about that. And what else should I mention in this intro? Yonatan?

1:34Yonatan SompolinskyI was passively involved in Bitcoin early on, I don't think I contributed a lot other than handing out interesting thoughts and, you know, challenging the status quo. But other than that, it took me time to contribute to crypto, it wasn't from day one.

1:58Vlad CosteaWhy do some people think you're Satoshi?

2:00Yonatan SompolinskyI think because he or she left the scene and there aren't many other proof of work.

2:12So the people that, I think most of the people don't really think it, but just enjoy thinking it.

2:30Vlad CosteaThat's how I read it. Well, it's very funny because for example, the Cardano fans believe that Charles Hoskinson is Satoshi, the Litecoin fans think that Charlie Lee is Satoshi, the Zcash fans think that Zooko is Satoshi. Seems like we have a lot of Satoshis in this space. But what you have created with Kaspa is very different from the traditional blockchain. You have...

2:51Yonatan SompolinskyFor me, I will just add Alex Chepurnoy from Ergo. For me, he's my Satoshi. He wrote 3600 lines of code, the entire Bitcoin code. He wrote it from scratch. And he's a very, very Satoshi-like. He's the main, he's the first Satoshi person that I met in person.

3:17Vlad CosteaWho's that again?

3:20Yonatan SompolinskyAlex Chepurnoy, the founder of Ergo.

3:24Vlad CosteaInteresting. He's not on the usual suspects list.

3:29Yonatan SompolinskyWell, he shouldn't be. No one should be. But I'm just saying like, if you're adding fans, so if I'm also allowed to add a candidate because you are allowed to add, so I'm adding him.

3:42Vlad CosteaI think there are a lot of smart people in this space. And I was going to say that what you're doing with Kaspa is different from what we are accustomed to seeing in most blockchains, where there is only one block and the rest of them that get discovered by other miners get orphaned. And it's a system which basically enables all orphaned blocks. Basically, there are no orphaned blocks in this system. And it's much faster than the usual ones. I guess this also comes with various trade offs, which need to get addressed.

4:18But I respect something which is proof of work, has a fair launch and is different from Bitcoin. And that's why we're doing this today.

4:26Yonatan SompolinskyFor me, you mentioned the only difference between Kaspa and Bitcoin. Kaspa having multiple block predecessors, not having orphans. I truly don't understand why people treat Kaspa as complex or complicated to pass the knowledge on. For me, if you are a Bitcoiner or you understand the inner workings of Bitcoin, then by generalizing this and telling you that each block has multiple predecessors, then yes, of course, there are various challenges, beautiful challenges that need to be solved and addressed. But the fundamentals, I'd argue 80% are like Bitcoin.

5:22Vlad Costea80%?

5:26Yonatan Sompolinsky80%, I would say. I know that's a peculiar or not a consensus point of view with neither communities, but that's on my own.

5:45Vlad CosteaSo let me start with this. What is Kaspa and why was it not possible to build this on Bitcoin?

5:54Yonatan SompolinskyKaspa is a generalization of the Bitcoin protocol where each block references multiple, all the latest, the most recent blocks that its miner observed. And there are no orphan blocks, as you mentioned. And this structure technically is called a DAG, a directed acyclic graph. It forms a DAG, but conceptually allows you to, so you envision many miners creating blocks in parallel very fast. And then because you can point at them in parallel, you can point at several blocks, then you get this messy, supposedly messy graph of blocks just growing.

6:49And most of the dynamics of the network are quite similar to Bitcoin. For instance, you know, each miner might have different points of view. Miners need to kind of reason about what's happening, transactions, proof of work, difficulty adjustment. The fact that you are pointing at several blocks, so it creates a challenge. And the challenge is that there are, they may be conflicts within these blocks. And so this implies you just need to order the block DAG, the structure of blocks that are flowing. And you need to provide an ordering that is guaranteed to converge across all nodes.

7:40So again, the one liner about Kaspa is it's a generalization of Nakamoto consensus. It's a generalization of the structure of chain or tree, rather, in Bitcoin. And this generalization by pointing out that multiple predecessor blocks allows you to define consensus in a way that will, in a way that the security won't deteriorate as you increase the block rate. So let me perhaps present it in a more theoretical but telling way. In principle, if you form the blocks in a DAG, you can still order them according to the longest chain rule.

8:26You can still mimic Bitcoin's operation with a DAG. The protocol simply dictates all transactions in the longest chain are, you know, the order of the blocks is every block and every transaction in the longest chain of the DAG is ordered canonically in the same way, in the same obvious way. All other transactions are ordered after them or discarded either way. So this, so you see the DAG, the DAG itself, the fact that you wrap things in a new structure and you called it DAG, this in and of itself does not create, just creates problems, but that does not create any solution.

9:10It's just, it creates a framework for how to define consensus. And this is not a framework that I or my advisor invented. This is well known in distributed systems that ordering events in a distributed system through a directed acyclical graph is the canonical way to record the causality relationships within, between messages or events. This is nothing novel. This is nothing interesting in and of itself. It's very basic in any consensus, even permission, the traditional consensus to say every node in a network signs a message or signs all messages that it saw and propagates them in the next round.

9:53And then the forming structure is a DAG of messages and you can implement Bitcoin's protocol within this framework as well by dictating, as I said, that the longest chain of the DAG dictates the ordering and everything off-chain is either orphaned, discarded, or at least deprioritized in the ordering, which is what we published in 2015, the inclusive blockchain protocol framework. This is alluding to that DAGging framework, but without yet changing the chain selection rule.

10:34The innovation is not in the DAG, rather in the way you order messages or blocks in the DAG. If I convince you that the way I or Kaspa orders the messages is canonical in the sense that it satisfies the classic consensus requirements, safety, liveness, or simply convergence across all nodes eventually. So if I convince you in that, then you'll see that why it's a useful generalization when I come out of consensus, because now you can, instead of dictating the order according to the longest chain, you dictate the order according to the heaviest cluster, the heaviest k-cluster, or in DAGKnight, some more involved mechanism.

11:22So that's the first part of your, to the first question, but maybe we'll maybe want to elaborate on that, or you want to ask the more interesting question, why this can't be implemented on Bitcoin.

11:34Vlad CosteaNo, I'm well aware that this whole DAG concept was proposed first or explained academically in the context of Bitcoin. At the time, there were a lot of proposals, I think, that started pouring in in 2012, 2013, 2014 about how to improve Bitcoin. I think the bull market around that time has sparked the interest of a lot of academics. That's when we had Zerocoin, that's when we had Zerocash, which turned into separate altcoins in the end. And we also had proposals for Bitcoin 2.0, which I think the most successful of it became Ethereum. So I guess it was only a matter of time until all of these proposals to create blocks in parallel, and also have these DAGs to have their own chains to demonstrate that they're sound and that the math checks out and that the vulnerabilities are not the ones that critics expect. And I was going to ask you about the origins of this, like, when did you start working on DAGs and the GHOST protocol?

12:41And basically, what did you envision you're going to accomplish at the time, compared to what you did with Kaspa?

12:52Yonatan SompolinskyOriginally, it was motivated by academic necessity to choose a lab project, and allow me to transition from mathematics to computer science. This is the, this is the original motivation. The concept of, so there, so my pet is, my pet is making noise.

13:22Vlad CosteaIs that a cat?

13:26Yonatan SompolinskyIt's a turtle.

13:27Vlad CosteaOh, you have a turtle.

13:29Yonatan SompolinskyYeah, we call it actually Bitcoin. Say hello. Actually waved. Yeah, we released it, and then he came back. So I can, I guess, the tie with Bitcoin is unavoidable. So the idea of generalizing into, so the first generalization was not into a DAG, but rather to a different chain selection rule within a tree. So Bitcoin already has an implicit tree structure. So when Bitcoin has a fork, technically, you're supposed to see a tree. At least a miner that observes the two forks before choosing the longest chain. And when my advisor, Aviv actually wrote the first academic paper on Bitcoin and Red Balloons, when he was in Microsoft in the West Coast.

14:33And when we, when I started my lab project with him, his interest was mainly focused on why, what happens when you increase the block rate of Bitcoin, because it was very apparent that the 10 minute rule was chosen for good reasons. So we tried to understand what happens with when the block rate increases. And, and then we came up with the GHOST protocol, which is still not, technically you can call it a DAG, but conceptually it's really a tree. It's a, it's a different way to select a fork within a tree.

15:13A few, a few weeks, or maybe not much more than that before we published the paper, the guy from Lightning, Christian Decker, and his advisor, Roger Wattenhofer, they published the paper that showed that the latency of blocks in Bitcoin or simulation of latency as the, as the, as the block rate or data grows. And this was, it complemented our thinking on this and GHOST was mainly an erroneous attempt to improve Bitcoin's protocol into GHOST. And we erroneously claimed that GHOST can scale to any block rate and recover the same security guarantees.

16:10This was flawed. This was a flawed claim. To my, or to our defense, we weren't, we were more from a game theory background. We weren't yet too well versed with distributed systems. And by the time I went to financial crypto conference to show the paper, we already knew there's a problem, but we didn't have a final analysis of is it broken or just not improving Bitcoin as we thought? Which is, it turns out to be the latter, but we knew exactly what the attack is.

16:48And, but when I met there Gregory Maxwell from Bitcoin Core and he, I told him the protocol and he immediately showed me the attack. He didn't even need to think about it, which I found, I was amazed. I knew the attack, but I thought like, it's, it's, I thought it will take longer, a lot of time to, a long time to explain it to people. So that's, so that, so that was the story of GHOST. It's not, it's still a nice protocol, but it's not achieving what we thought it achieves.

17:28And very early, not long after that, Aviv already, Aviv, my advisor, Aviv Zohar, he pointed us towards the notion of DAGs. But again, DAG, there's two things and it's important for people to remember this, DAG is just a framework. So when I say to point us in the direction of DAG, I don't mean to those early 2011, 2012 ideas of circling in Bitcoin around keeping the chain, you know, longest chain, than adding things from the side. Rather, I'm talking about a pure, a true, a clean generalization of Nakamoto in the sense that there's no tricks.

18:13There's no more, there's no like a drive chain or things from the side. Like there's no second rate citizens. It's just like, take all blocks and try to order them and linearize the DAG, so to speak, in a canonical, in a canonical manner that will converge across all blocks. And that's the, that's when the journey, the real, the real DAG journey, a journey started.

18:41Vlad CosteaYou mentioned drivechains and coincidentally, it's been 20 minutes and I have to present a sponsor and one of the sponsors is LayerTwo Labs, which is building the drivechains. Now they have the software ready so you can test it on the Bitcoin signet. You go to layertwolabs.com slash download and you're going to find side chains for Zcash, Ethereum, prediction markets and something like Namecoin, if I'm not mistaken. You can give it a try and you can see if this is the kind of future that you want for Bitcoin.

19:12And if you do like it, you can advocate for BIP-300 to get activated on Bitcoin, even though I'm pretty sure you can do drivechains with OP_CAT, but they're not as great. They don't offer the same security guarantees as something like BIP-300. And I was going to ask you, what do you think about drivechains? You mentioned that.

19:35Yonatan SompolinskyI don't have a strong opinion on that and not, and not an intelligent one at least.

19:43Vlad CosteaLike in principle, it sounds really good. And I think it's the kind of Bitcoin maximalism which I like, which is about providing the functionality on this monetary layer, as opposed to creating other networks for different purposes.

20:12Yonatan SompolinskyThat's a weird definition of Bitcoin maximalism, or at least a new one that I'm less aware of.

20:23Vlad CosteaYou have the Bitcoin maximalism with a capital B, and with a lowercase b, which means for the network or for the monetary layer, the 21 million coins. And there are people who say we should build everything on this network, and there are people who say we should build everything to use this currency. Of course, there are intersections, but I think this is the easiest way to distinguish. And I think drivechains are building for an economy which happens on top of Bitcoin, and also uses BTC, the token, for everything.

21:00Yonatan SompolinskyThe second part completely makes sense of the lowercase Bitcoin maximalism. I think maybe it's called the Nick Szabo of Bitcoin maximalism, criticizing the social scalability of launching new coins.

21:22From my understanding or my comprehension of the Bitcoin culture, there isn't much support for creating lots of functionality on top of Bitcoin. That's at least the Bitcoin maximalism I'm more familiar with. That was the spirit and atmosphere in Bitcoin maximalism back in the day, in the days between 2013 and 2018. It wasn't that everything should be built on Bitcoin, but more of a distaste towards finance and other functionalities.

22:11So for me, the correct term or the more telling term is crypto minimalism rather than Bitcoin maximalism. Meaning it's a statement that crypto, Satoshi's invention or Satoshi's values, should begin and end with creating a stateless monetary base.

22:39And all the rest shouldn't happen, or either shouldn't happen or shouldn't decentralize. This is at least the movement as I gathered.

22:54Vlad CosteaBut if we appeal to Satoshi, he's the same guy who supported the creation of Namecoin and also proposed that Namecoin is going to get merged mined with Bitcoin. Namecoin also has 21 million coins, is merged mined, which means that transactions are paying fees to the Bitcoin miners. And I don't think anything interesting happened with Namecoin lately. It was pretty promising early on. It's basically what ENS on Ethereum turned into. And I guess Satoshi was not opposed to using the technology to build other chains. That's what I'm trying to say.

23:37Yonatan SompolinskyI'm not saying otherwise. I'm just saying that Bitcoin maximalism messages and takes in the first decade of crypto were of the type opposing other functionalities. I'm not necessarily saying Satoshi was a maximalist in this sense. But the known core devs, the hardcore culture around Bitcoin was definitely that. The main critics of Vitalik weren't, you should do this with BTC, with lowercase Bitcoin. And that wasn't the main statement. The main statement was DeFi is a scam. And I think more of a distaste towards finance in general.

24:27Vlad CosteaI sort of disagree here, because if you go back to the article that Vitalik wrote in 2014, which explicitly comes up with the term Bitcoin maximalism. He references a Reddit thread where someone said, well, you go ahead and build your EVM and whatever is nice with Ethereum. And we're going to take that and put it on Counterparty. And at the time, and still today, Counterparty is this protocol on top of Bitcoin. It's not another layer. It interprets OP_RETURN messages in a certain way to create tokens.

25:02And it could have been so much bigger if it wasn't for the limit that core developers set on the amount of data that they could put into OP_RETURN messages. It was sort of a war before the scaling wars that happened within the community. And we could have had something very similar with Ethereum on Bitcoin with Counterparty if that one was allowed to flourish.

25:27Yonatan SompolinskySo you think that there was a technicality that prevented Bitcoin core devs from migrating or copying the EVM concept into Bitcoin?

25:44Vlad CosteaYes. And I interviewed Andrew Poelstra last week, and he said he's opposed to the idea of having other assets on Bitcoin that compete with BTC. Because it sort of perverts the incentives of miners, where you can have an NFT or something which is maybe worth more than a block reward. And you're giving the miners the incentive to try

26:11Yonatan SompolinskyTo reorg the block. Well, that's also the case on Ethereum. You prefer not incentivizing miners and other assets. But this is a weird take, and I'm curious whether Andrew has voiced this in the past. Maybe I was less informed than I thought. But from my recollection, this wasn't the main talking point in Bitcoin circles. Like calling Ethereum a scam because of the ICO or because it was a general distaste of finance. So I guess I'll push back the question to you. So was this a main, as you understand, it was the main talking point in Bitcoin circles? Like this EVM DeFi is great, we should just do it with lowercase Bitcoin?

27:12Vlad CosteaI think there was at least a subset of Bitcoiners that were hoping that this is going to happen on BTC as opposed to allowing Ethereum to grow as a standalone chain. And especially the big blockers, they were hoping that a lot of activity is going to happen on the Bitcoin chain. If there was going to be an increase in the block size, there was a lot of hope that there is going to be smart contracts. And a lot of the stuff that actually happened afterwards on Bitcoin Cash, if you pay attention to that.

27:45They did CashTokens, they did CashFusion, which is sort of like a CoinJoin. Now they're even trying to do contracts and stuff to basically... So they enabled OP_CAT, they want to do ZK roll-ups also. It's kind of interesting because they're showing what could have been done on Bitcoin if there was a little extra capacity.

28:13Yonatan SompolinskyI Beg to differ, this is not my impression. Of course, if ZK roll-ups were production ready at the time, then people's opinions and the influencers opinions could have changed according to their projection. But it is what it is. What came from them was a general rejection of Ethereum. And I don't think this was nuanced. You can mention CoinJoin on Bitcoin, on the Bitcoin Cash. These are still very related to raw money, raw payments.

29:06Ethereum drove the innovation, the need for the roll-ups. It wasn't, at least not after Ethereum launched and the divide, it wasn't promoted heavily by Bitcoin circles. But I'm happy to be proven wrong if you have a more informed take.

29:30Vlad CosteaNo, I don't disagree with the point that Ethereum drove the innovation. They put a lot of money into this, we have the ZK roll-ups because of them. And they also have a lot of interesting research where I think, to be honest, like you mentioned Greg Maxwell and the fact that he was able to poke basically into the vulnerability of the GHOST protocol. He also spent many years predicting that Ethereum is going to fail and that it's not possible for it to do stuff. But I think Ethereum survived in spite of all the critics and was able to transition to proof of stake without any major issues and I think is doing pretty good under...

30:14You know, the circumstances of always being under scrutiny and always being attacked by quote-unquote Ethereum killers and having a lot of critics who say it's not the best design, it's not the best way you can do this, the smart contracts are not secure, the scripting language is horrible, but it's still gaining network effects. Right now, another one of my sponsors is Citrea, which basically is trying to create Ethereum on Bitcoin with a ZK roll-up, which I think is interesting. And this could not be possible once again without the research that Ethereum pushed forward.

30:53Yonatan SompolinskyIt's interesting and it's... I'm aware that... I guess everyone's aware that the DeFi on Bitcoin movement has begun maybe two years ago, you know, started gaining traction. I didn't see it yet gaining momentum. I think it partly, of course, is because of the OP_CAT, but also partly because... There's still the highly conservative, extreme conservative... The extreme culture that Bitcoin Core has radiated isn't very conducive to innovation.

31:46And regardless of if you're using for the base token BTC or ETH or another thing, you're still trying to attract people that want to innovate.

31:59And they will want to innovate with their own business models, be it token or revenue. And I don't think it's very... I don't think if you build on Bitcoin, you'll find the atmosphere very conducive to your productivity and exchanging ideas about protocols and non-money-based DeFi or non-DeFi-based use cases. I don't know that this is the atmosphere in Bitcoin. The atmosphere is more about celebrating incremental improvements like Taproot. It's very weird for someone to think about a new AMM idea or a new smart contract.

32:39There's the architecture with the VM that reads other VMs and start having this freedom of mind and then go to Bitcoin and then attach themselves to Bitcoin community. Or the development circles and then to hear about all these arguments about... Or celebrations of one's opcode upgrade. And I'm not mocking it. I'm just saying this is not conducive to the kind of activity you want to migrate to your ecosystem.

33:15Vlad CosteaNo, but I also sympathize with people like Wladimir van der Laan. He was the maintainer for eight, nine years or something. Basically, he's the guy who took the stewardship from Gavin Andresen, the keys to the repo, and has been the maintainer of the project for many years. And he was sort of very wary of changes. He was like, I don't want to mess up. Like, I don't want to be responsible for all of this. It's very difficult. There's a lot of pressure involved. I don't want to be the guy who breaks Bitcoin.

33:48Yonatan SompolinskyCompletely understandable, completely responsible. This is why Bitcoin is very successful in its value proposition. And again, I'm not mocking it. I'm just saying it's very hard to attract innovation to such a system. I'm not saying the culture, the extreme conservative culture of Bitcoin is not merited and it doesn't have its value. Of course, it has a value for what Bitcoin currently aims to be. But pretending that you can attract the same vibes, for the lack of a better word, that Ethereum has had attracted of innovation, of openness to a system where the maintainer is mainly afraid of breaking. Rightly so, breaking or changing anything is incompatible.

34:39Vlad CosteaI agree with this. That's why we needed to have like the anti-Bitcoin, the complete opposite of it, which is moving fast, breaking needs to have sort of a political structure to be able to accomplish this. But otherwise, they did a pretty good job paying developers, academics, challenging people to come up with new ideas. I think Ethereum has had so many failed layer twos and it's beautiful. Like I remember Plasma when it was a thing, Raiden and all of these ideas that went nowhere. And then they said, OK, let's let anyone create their own L2.

35:17And now you have Arbitrum and Polygon and all the others. They do nurture innovation and they do encourage people to build without asking for permission. On the Bitcoin camp, it's sort of the opposite. You do need like the social consent of everyone else to build something. And if you go against the culture, like there is that guy. I don't remember his name, just like you don't remember the guy, the lightning guy. I don't remember the guy, the name of the block. What was it called? Like a Bitcoin L2, which had its own token.

36:04Blockstack. And I think he renamed it to Stacks. Like he got so many insults from core developers. He got called a scammer by Luke Dashjr, just like I guess you did recently. And his only sin is that he actually tried to bring extra use cases and increase the functionality and also had a token. And apparently it's very against the culture to have a token and try to monetize this.

36:40Yonatan SompolinskyThere's definitely an arbitrary timing issue, right? The early discussions, even when Satoshi was around, did mention ZK. People understood the value of being able to separate computation from verification. It was just from their perspective, like years away. And of course, I don't think there's any argument about that. Of course, the best rewrite of history would be if everything was production ready then. And you could have created this separation of layers and keep the backbone highly cypherpunk, trustless, toxic, you know, and requiring everyone's consent, anyone, it's enough that you're

37:41Not liked by one or two or three people or you're not passing the purity test to prevent you from innovating on the backbone and then having other layers that are conducive to innovation. But I argue it's not only a technical separation of layers. It still needs to be a cultural separation because, again, because of the different atmospheres that are needed for these different tasks. Maintaining the repo and the money base merits its own culture and its own maybe aggressive or sometimes toxic culture. And being conducive to people that want to create new ideas requires a different society, a different layer, a different culture.

38:38Vlad CosteaI think the culture sort of evolved to justify the technical limitations. So as soon as it was obvious that there is not going to be a block size increase and there is not going to be any sort of hard fork for it. And the community agreed that this is the real Bitcoin. They needed to build some ideas around it to justify why it's so limited. And they said, well, it's more like gold. And how many transactions do you need for gold? Gold is slow. Gold is hard to move.

39:08You don't need to make payments with gold. Like you just use it for settlements, use it for large transactions. And I think overall, this has been detrimental to the adoption of Bitcoin as money. We haven't seen much of it. If you look back 10 years ago, there is more of it than there is today. We had Dell and Expedia and Microsoft and Steam and I guess many more services that were dipping their toes into Bitcoin. And now we don't have much of this. It's just people. I mean, in some ways it's nice because it's individuals as opposed to companies, but it's much more difficult to accept BTC today than it was 10 years ago.

39:54Not only from a legal point of view, because there are more regulations, but also because it's unpredictable. Like you don't know if you're going to deal with a very small transaction fee or there's going to be some random spike or someone's going to inscribe their ordinal collection and the fees become $50 again like they were last year. And then what? You have to set up some sort of weird L2 which is interactive so it means you have to be connected to the internet to receive payments. And by this I mean lightning.

40:30It's not the optimal user experience. And that's why a lot of these payment coins have benefited from the fact that they offer a similar user experience and also a more instant settlement. Like Litecoin has grown a lot. Monero is very used on dark markets. I think there's also a Dash and Zcash market out there. And also Kaspa which is trying to do payments. I don't think you guys are aiming to become the reserve currency of the world. You're trying to become the payments layer, right?

41:08Yonatan SompolinskyI don't know. I wouldn't limit Kaspa's aspirations to payments. I would say in the context of our previous topic it's an attempt to actually merge these two challenges in one native system. Namely having a solid sound money monetary base with someone that's paranoid about the repo. Admittedly much less than Wladimir but still. And yet facilitating and being very conducive to DeFi and very friendly to DeFi.

41:59While Kaspa today can support payments only, it's not a matter of a long time. It's a matter of maybe 12 months. I don't know exactly how much for Kaspa to support DeFi. But it's not only an upgrade. It's more about aspiration. Aspiration is not limited to payments. It's exactly to have a system and a culture that manages to marry and make peace between these two pillars of crypto. Between money and finance. I can't say that I know which pillar is more important. Of course, Bitcoin came first and this was the main steering.

42:58It forever will remain the primary if anything else just for its precedence. But both value propositions are equally important in my view. And at least my inspiration with Kaspa is to marry these two value propositions and pillars of crypto into one system without any friction.

43:28Vlad CosteaIdeally yes, but today do you think it would be possible to have tokenized Bitcoin on Kaspa which moves with higher velocity?

43:39Yonatan SompolinskyPossible in what sense? Technically or adoption?

43:42Vlad CosteaTechnically. Adoption is mostly social. It's just people agreeing that this is useful. But technically can you have this?

43:52Yonatan SompolinskyAs with any system you need to create a bridge and there's some trust in the bridge. Either the common centralized bridge or a rotating committee.

44:06I'm not sure what the question is. You can do this. What would be the value proposition? Does it go back to the lowercase BTC? Or are you looking for something else?

44:19Vlad CosteaIt goes back to the lowercase BTC and just to have not a layer 2 but a sidechain to Bitcoin. If you want to make payments very fast with something that has a price which is predictably more stable across the bull/bear cycles.

44:38Yonatan SompolinskyMore predictable than Kaspa?

44:42Vlad CosteaIt's just that we have more historic data and people sometimes it's counterfactual to take historic data as evidence but people are psychologically drawn to believing in this and it becomes a self-fulfilling prophecy that Bitcoin is going to be more stable and less volatile. That's what I mean against US dollar or whatever.

45:06Yonatan SompolinskyDefinitely though people don't want Bitcoin to be stable. They want it to be a downward stable. They don't want stability in Bitcoin yet. That's possible. This reminds me of the Bitcoin Ethereum wars when WBTC became highly adopted on Ethereum whether this is a win for Ethereum or Bitcoin. I kind of see the sense in both claims. I would argue that there's always not only psychological but also fundamental, maybe a better word will come up, advantage to the native currency that's running the network.

46:05I don't know that anyone would oppose tokenized Bitcoin running on Kaspa but I guess the flywheel is typically benefiting the native currency.

46:22The reason is because friction is always the main thing that's defining good money. If there's a bridge between Bitcoin and Kaspa then that bridge incurs friction.

46:40And what makes a money system, or one of the things that makes a money system good or not is whether using it requires more friction or not. If you were to compare the advantages of Bitcoin stability or at least downward stability versus nativeness of Kaspa in a system that operates DeFi, for sure at the beginning the stability of Bitcoin would be very appealing. But as the system is adopted and more people gain trust in the system, I think that the nativeness of Kaspa would override the need for predictability. This is what happens I think with Hyperliquid.

47:34There's high usage in the system of its product and then this creates demand by push to the native token. I'm sure you have tokenized BTC on Hyperliquid. It's still Hyperliquid and the HYPE asset is very successful and it's not surprising the fact that you have integrated a tokenized version of Bitcoin is not such a big threat. I guess in a broader perspective money is, and this is kind of obvious, but money is very psychological.

48:26There's no reality where tokenized Bitcoin and Hyperliquid is threatening HYPE. I don't see this as a real threat to HYPE's success, to take another example.

48:42Vlad CosteaYeah, to prove your point, right now there are like 100,000 Bitcoins locked on Ethereum as WBTC, which doesn't use a bridge, it's actually custodied by BitGo, which is not ideal. And for scale, I think the Lightning Network has like 5,000 Bitcoin and Liquid, which is a federation not a sidechain, has like 4,000. And Rootstock, which is super underrated, also has like 3-4,000. All of them are orders of magnitude smaller than WBTC. And this also segues into an ad for the next sponsor of this episode, which is Citrea, which is trying right now to build an optimistically verified ZK rollup on Bitcoin.

49:28That's going to aim directly for this 100,000 WBTC that's locked on Ethereum to bring it to something which has better security guarantees and also is closer to the Bitcoin network. Because essentially it's Bitcoin, but it trades on Ethereum, makes no sense. Why not bring that home? And you can try the software today. It's running on the Bitcoin testnet. It's at citrea.xyz, which is C-I-T-R-E-A dot X-Y-Z. And I think there are also a bunch of other companies which are similar that are trying to do this, like Alpen Labs and...

50:09I think Botanix, if I'm not mistaken. But it's good to see that some of this is coming to Bitcoin. Like, innovation is happening without the permission of the gatekeepers. This is what keeps me going, what keeps me excited.

50:30And just so you know, Yonatan, there are people in the chat who say they're gonna make a meme coin of that turtle.

50:40Yonatan SompolinskyWell, since this turtle is an avid Bitcoiner, he's strongly opposed to that idea. He will rug your meme coin.

50:51Vlad CosteaYeah, then you should send like 50% of the allocation to the turtle so he can dump.

50:59Yonatan SompolinskyI can help him with that. Now that I have too much experience, but for that cause, I'm willing to get my hands dirty. You know, the conversation, it reminds me of two quotes that I want to... I want to try to faithfully quote. But before I quote, I would say I truly understand and identify with the averseness of Bitcoiners to other money or other tokens. This is completely understandable. It's not that at least the lowercase maximalism is not...

51:48It's more than legit. It's very understandable. And perhaps I will say that Kaspa's launch is quite reflective of this understanding and alignment.

52:10But there's a nice quote of Richard Dawkins where he says, we are all atheists with respect to most of the gods that humanity ever believed in. Some of us take it just one god further. And of course, he was preaching atheism, but I think it will be worthwhile for Bitcoiners to, or for lowercase maximalists to recognize that this is also how non-coiners treat Bitcoin. It's not a different sentiment.

52:56And the other quote that comes to mind is, or rather story, it's a very famous story of Milton Friedman. He has a very nice article. I recommend anyone that hadn't come across it yet to read it. It's about the Yap Island, the stone money of the Yap Islands. You know, these stones with a hole in them.

53:18Vlad CosteaThey're called Rai, right? R-A-I. Rai stone.

53:23Yonatan SompolinskyYeah. I don't know how to pronounce it. Ray or Rai. Yeah. So I won't spoil the story. He does a very good job. I don't want to spoil anything here, but he concludes it with saying that every other society's money is very weird to us. Like you walk, you would walk in the Yap Island and you will see their money and you wouldn't understand them. It would look like garbage to you. But they would go to your, they would look at your, so he quotes, he tells the story of the gold reserves in the Bank of America or the Fed, no,

54:19It was in the Bank of America, replacing France bought back the gold from United States. And there's just like the switch in the drawer between the USA and France and the gold and the dollar. And he quotes, he tells the story and he says, someone from Yap would look at this and will say, what is this shit? This doesn't look like a real money system. And so while I identify, I completely identify with the sentiment, I do think there's some lack of awareness that this would be also the sentiment of anyone outside the Bitcoin, to Bitcoin.

55:01So this averseness or the feel that other society's money is just very weird and useless. It's not very unique, but I do want to, I do want to tell you the story of Kaspa's launch, which was very messy, especially compared to Bitcoins. Because I think if you reflect on this now, if I reflect on this now, this is quite telling on the values and how I was really against launching Kaspa almost.

55:51So I did everything that I could to make, to prevent the launch or for a smooth launch.

55:59I'm not sure if you know, the story of Gamenet and all these things were considered by everyone, including myself as very bad ideas that we should get rid of immediately. And rightly so, but it does give you, it gives you a genuine depiction of our mindset back then. So the story was that around the third quarter of 2021, I told the investors that I don't see how proof of work is compatible with fair launch and that we should abandon both the business model and the launch. And to their credit, I believe it was Karthik from Polychain,

56:58He said, you know, the narrative can change, the narrative around proof of work can change, just launch it the way you see fit. And I still had a very, I felt very uncomfortable to launch a new platform.

57:20I didn't want to do so. So I consulted Nick Carter and he was like, you should, if you don't think you want to do this full force, then don't launch. And the creative, very bad idea I had was to create a gamenet, which is sort of an attempt to do a testnet with incentives. And it wasn't a thorough, you know, I wasn't, I was at Harvard at the time, so I didn't give it too much thought. And I thought, I genuinely thought this will be mined by three people.

58:02And I thought it would be very cool to showcase how, showcase the research in live, but without trying to increase its hash rate or adoption. So I came up with a bad idea of doing random rewards in the blocks, but I tried to, I tried to, you know, bake in high level of randomness so that, so that miners will either be disincentivized to join the network or will start double spending and, you know, creating reorgs. And I thought it would be very cool to see the Clash of Clans, to see MEV wars or reorg wars between miners.

58:45And that was the very bad idea of gamenet, but it does give you a sense that it wasn't an attempt to, it wasn't a thought, a careless, conscious-less attempt to launch a new Bitcoin or Namecoin or token cap without any care to the problem of lack of social scalability that Nick Szabo rightly talks about. What happened was that people voted otherwise, and we didn't even bake in a monetary policy yet.

59:31And people, namely miners, most of the vast majority of which I have no idea who they are, they voted otherwise.

59:39They chose to mine it nonetheless, and the hash rate grew very rapidly. And we, as I and one and a half devs that were, that were working on the, on the final runway of the company were not prepared for that. We didn't, we prepared neither mining nor support for this growth. And after two weeks, the network halted and we had to So we had to create, to debug, create a new, a second genesis and cryptographically point to the last block so that everyone can see that nothing shady has happened.

1:00:22And the cryptographic proof is, you know, is available online. But, but you can, you can, you can see this, you can, so compared to Bitcoin's impeccable launch, this is horrendous. But, but I, but I do think this is, this is actually a feature in the sense that you recognize that this wasn't an attempt from a team or from my side to push a new, to create a new token cap. Rather, it was a vote by the people, so to speak. And, and I think if someone launches a new, a new L1, I think this is, this is the best way, you know, to check, to bake, to do it.

1:01:08Do everything you can to make it fail. And then to have the people override your allergies to new tokens.

1:01:17Vlad CosteaThere is actually an article that was published on Substack. I think it's the most quoted one when it comes to criticizing Kaspa or the claim of a fair launch, because they say that there was basically 648 million coins were mined. And you decided to stop you. I mean, in the plural form, not you personally, decided to make a new Genesis block as a result of a bug. This is the claim. And the rest of the article basically explains how people were complaining that they were checking their mined transactions on a blockchain explorer and realized that they're gone.

1:02:00And there were some accusations that this was just a scheme to basically steal some of these early mined coins and redistribute them to the other miners.

1:02:13Yonatan SompolinskyYeah. So these are different. These are three different accusations. Can you, can you repeat them? Like the TLDR?

1:02:23Vlad CosteaThe TLDR is that first of all, the accusation for the team is that you plotted to restart the chain after a lot of coins have been mined, and basically issue a new Genesis block. And also for the miners that they made this plan to take away the coins from the early miners, and some users have lost the coins that they mined in the process.

1:02:49Yonatan SompolinskyYeah. So the users that, the users that don't see this on the block explorer, they're just not technical enough that they need to download a full node with UTXO index flag activated and see this in the UTXO in their full node. The fact that the block explorer doesn't show it could be, is not, is not indicated by the fact that the coins are missing, rather that you need to run your full node by yourself with the correct flag and identify your UTXOs. The UTXO set itself is never discarded.

1:03:28Nothing is pruned and the UTXO set is intact. Nothing has changed. Kaspa operates with UTXO commitments, just FYI. So you can see your UTXOs with running a full node. I do want, I do want to, I do want to say in generally, so two points. One is the more concrete answer is that the network and the community were very large at the time. It wasn't three people. So you can go to the discord and just see everything that happened. You can see that you can see the entire drama unfolding.

1:04:15And, and the second point, the second broader point is that this trust in off, you know, off chain social forums happens in every coin, even in Bitcoin, you have to trust if you are, if you're concerned by breaching of integrity in the past, you can't verify this on chain. You simply can't. People, the miners, the early miners of Bitcoin, Satoshi and Hal Finney could have, could have colluded to do a big, a big reward. They could have colluded. The early miners, and you wouldn't see it.

1:05:06You wouldn't see it in the Bitcoin chain. So the economic integrity of Bitcoin or of any proof of work coin cannot be cryptographically proven. But the main point I want to, I want to, I want to say is that there are many, many ways for a team launching a coin, whether it's a Bitcoin or Kaspa to cheat, to cheat early on. It's through, it's not, you don't have to do pre mining and you don't have to do a bug. Again, you can consider the DNS seeder of Bitcoin failing early on.

1:05:37Letting Satoshi have and Hal have like full autonomy over mining. No one else can connect. The DNS seeder is broken and they, they turn, you know, even when it recovers, they turn off their ports and no one, no one can sync with them. This, this is, and this is not a theoretical point, by the way, this is, this is what happened with Kaspa. If we wanted to cheat, I guarantee you there are many, many other ways that you wouldn't think of for us to ensure that no one else can mine, especially in these early days.

1:06:15We can just create simply a very lazy debugging process of peer to peer network. This would be the perfect undetectable way for us to kick off other competitors, you know, to just have control over the mining. And you, there will be no, no thought about it. No one will know about it. People, people would think this is like people that would not know that it's a broad problem. They would think it's their own local node's problem. And regardless, people would complain about early peer to peer, not being smooth without understanding it's a scheme by the malicious devs to push them out of the competition.

1:07:03So the thought that you need to pull off such a bug in order to cheat is an insult to our intelligence. There are much more sophisticated and smooth ways for us that much less sophisticated and more smooth ways for us to pull this off.

1:07:23Vlad CosteaYeah, I appreciate the fact that you addressed this. You could just say no comment or whatever. But I do want to move on to the next point and try the onboarding of Kaspa so that I can understand what's happening in the background. I have a wallet that I have been running all day and playing with it. And I want to see, to basically learn from you what's happening here, because it's very different from what I'm accustomed with Bitcoin wallets. But before that, and I apologize to the audience, that's going to have to wait two more ads and then we can move on.

1:07:58OK, Bitcoin.com News is a place where you can even read about Kaspa and Yonatan, basically, I think he took a break. On Bitcoin.com News you can read news about Kaspa. And that's how I found out last year when I also started mining with my GPU while it was still slightly profitable. So last year, Kaspa was the most profitable coin, according to a news article by Bitcoin.com News. And I also use it for finding news about what's happening in the world of gold or in the world of Bitcoin, but outside of my usual bubble.

1:08:34And I think it's a very nice resource where they do pretty good journalism and I respect it. And the URL is news.bitcoin.com. Check it out. Personally, I open it every morning and see what's happening in the world. There is also NoOnes.com, which is a peer to peer marketplace for people who want to support the global south. The CEO of this exchange is Ray Youssef. He was previously working on Paxful and he decided he doesn't want any more VCs. He doesn't want to work with anyone from the US anymore.

1:09:11I was telling people about NoOnes, which is, do you know Ray Youssef of Paxful?

1:09:19Yonatan SompolinskyIt rings a bell.

1:09:20Vlad CosteaNow he has this peer to peer exchange where he's trying to empower the global south. And he talks about his African alliance and the fact that he's trying to do more than any African leader was able to do with economics and the fact that he opens up the free market for these people. And if you do like the way this sounds, and it's not just Africa, it's also South America and basically the southern hemisphere. And if you do like this idea, you can provide some liquidity by trading there.

1:09:52It's called NoOnes.com. They also have a debit card for spending your coins. They have a spot exchange. They have a gift card marketplace. If you are into this, if you have some cards that you want to buy or sell, check it out. It's NoOnes.com. Thank you for sponsoring. And now let's talk about the onboarding experience to Kaspa, right? Because in the early days of Bitcoin, it was so simple. You download that Bitcoin QT on your computer, you sent a QR code or address to someone and you received money and it seemed so magical. And I'm curious, what is the wallet that you usually recommend to people? For Kaspa?

1:10:35Yonatan SompolinskyI don't, I use CLI.

1:10:40Vlad CosteaOkay, but let's say that you have a friend who's not very technical, doesn't know how to type command line stuff, inputs. What do you tell him to try?

1:10:52Yonatan SompolinskyThe genuine answer is I don't. I don't, not that I don't have friends also, although that can be the case as well, but I don't have, my answer is I understand how to use the CLI. And other people in the Kaspa community are much more proficient in wallet software. I'm sure, like, I would invite Anton, I guess, to speak about wallets. I'm not the right address for that. I use just CLI.

1:11:25Okay, for the record... And the experience should be the same, as you mentioned with Bitcoin, if you're using CLI.

1:11:34Vlad CosteaFor the record, right now I'm using the kaspa-ng.org wallet. And previously, last year when I tried to mine, I used the KDX, which I tried to sync today and I saw that it's no longer in sync. I guess some sort of hard fork happened in the meantime and it doesn't connect anymore.

1:11:56Yonatan SompolinskyI think they were both developed by Anton. And yeah, KDX is redacted. It's not compatible, at least from what I remember, and it was replaced by Kaspa NG, if I recall correctly. It's not my field of expertise and I'm just using the client.

1:12:27Vlad CosteaI guess you can see my screen now. Let me change this a bit so that you're also visible. Okay, this is my screen. Let me disable this. It's not necessary anymore. I'm just making some adjustments to our livestream view. This is the wallet, which I set up today. It has some stats in the center of the user interface. I do like that it's geeky. It's running a full node in the background that tells me I'm connected to 32 peers. And does it download any blocks to the local computer? Like, is anything stored right now on my SSD?

1:13:13Yonatan SompolinskyI wouldn't know to answer. I didn't develop Kaspa NG.

1:13:18Vlad CosteaOh, no. Anyway, what's the feature that I should try here? Like, what's going to blow my mind? I can go to wallet.

1:13:32Yonatan SompolinskyDo you want to see a transaction? Sure. Let's see how long that takes.

1:13:37Vlad CosteaLet me type my passphrase for the wallet.

1:13:41Yonatan SompolinskyAre you connected to mainchain?

1:13:45Vlad CosteaYeah, I've just decrypted the wallet. I have zero balance, as can be seen here. And I guess I'm running a full node in my browser with this.

1:14:09Yonatan SompolinskyOh, so you can't send a transaction because you don't have Kaspa?

1:14:12Vlad CosteaOh, yeah, I don't have any coins.

1:14:17Yonatan SompolinskySo from my familiar with the community, you should paste the entire address. And I believe the community will send you a coin right away.

1:14:27Vlad CosteaAnd I'll send it back. Or I can reveal my seed phrase at the end of the interview and the first person to read that off the screen is going to get the funds.

1:14:36Yonatan SompolinskyThis happened to me live when I was TAing on Bitcoin. I used the brain wallet. Remember the unsafe brain wallet early days? We had to have like a low entropy. Oh, we're ready to go.

1:14:52Vlad CosteaNo, someone sent me 50 coins and someone sent me one KAS. So it's confirmed already what's happening here.

1:15:02Yonatan SompolinskyLet's see.

1:15:05Vlad CosteaThis is the explorer.

1:15:07Yonatan SompolinskyNow let me enlarge that part of the screen. Okay, so the Kaspa thing, the DAG thing is the block hashes, right? It's the multiple predecessor references. So instead of one predecessor, you have multiple block parents. One of them is the main one in the ordering, the main chain block. But anyways, these are the block hashes. So the accepting block hash is the block that is the, so consider the, so it's useful even in the context of a DAG to think of some chain in the DAG that around which the ordering is solidified and converged.

1:16:00This is, this chain is not chosen according to the, to the longest chain rule, nor according to GHOST. It's the GHOSTDAG protocol. And, and so the accepting block hash is the, in case you're not, so it's the next chain block, which includes your block, the block that includes, included in transaction. So it's very useful in terms of, you know, heuristics for confirmation times to know what, if my transaction, let's say off chain, it's not in the main canonical chain that was ordered according to GHOSTDAG.

1:16:37What's the next block in the chain that's viewed, that saw, or that has a path in the DAG to my transaction? So this is the accepting block hash.

1:16:49Vlad CosteaOkay. I see that I have the input someone sent from a UTXO that had a hundred KAS to my address and received this amount of change. This is pretty cool. Can I see how much money this guy has in his wallet? Yes, I can.

1:17:09Yonatan SompolinskyYou should, you should. We don't have bulletproofs yet. Yeah, just copy the script up here.

1:17:17Vlad CosteaOh, he's smart. He only, I guess this is a mining address or something. What's the mining reward right now?

1:17:28Yonatan SompolinskyLet's see, go to the Kaspa dashboard and let's see what's the mining reward, but you can, I guess you can see it also in the, no, go to the dashboard.

1:17:44Vlad CosteaBlock reward is 4.26 KAS right now. Okay. So I just made...

1:17:53Yonatan SompolinskyThis is a miner that sent to you, sent it to you?

1:17:55Vlad CosteaNo, I just assumed because I saw it's from a UTXO that had exactly 100 and no previous history.

1:18:02Yonatan SompolinskyBut did you check, oh, you checked the public, you saw that there's no public previous history. How do you see it so fast? I don't know.

1:18:15Vlad CosteaJust pushing buttons here. No, he batched. Yeah, there's a lot of history here. I don't want to dox this guy's wallet. Yeah, exactly. That would be very interesting.

1:18:28I do appreciate the fact that you sent me one KAS and right now I have three transactions that got confirmed. What is usually the consensus regarding finality? Like after how many confirmations do you consider the transaction to be final?

1:18:49Yonatan SompolinskyOh, that's a very interesting topic. So I think, so I don't, I think there's a rule, a rule of thumb of 10 or 12 seconds, but I think this should, this should, this should be more, more carefully justified and we should like make the wallet dictate the parameters. So the, like in Bitcoin, the wallet should dictate epsilon and alpha that attacker hash rate, the attacker presumed hash rate and, tolerant tolerance for error for reorg. So in our case, I think the current consensus around 10 or 12 seconds, but, I do intend to publish a more thorough analysis of this, both, most in general, but also in terms of Crescendo understanding how Crescendo.

1:19:43What affects Crescendo had on the upgrade to 10 BPS. That's all this, all this is very, is very, so even if the, let's say we need to, even if for some parameters, people need to, wait 15 seconds or eight seconds, this is for me, very, very slow. So that's why I want to, that's why when I'm thinking Kaspa, I'm thinking DAGKnight. I'm not, I'm not thinking in the current system, but in the next big upgrade, which is, which is, which is really what would enable you to confirm transactions like the internet works, which is, this is the main original aspiration to make transactions confirmed.

1:20:27Like you use the internet. I suppose like you use, proof of stake with what, you know, with its own, drawbacks, but fast finality. So I would, I would want, when I'm thinking Kaspa, I'm thinking, okay, all this is temporary. Let's, so I'm thinking, okay, we need to confirm transactions, internet speed, it should take, lots, a lot of order of magnitude, less than 10 or 15 or 20 seconds.

1:21:01Vlad CosteaYeah, this is really cool. And for the record, am I running a full node right now in the browser? Am I validating my own transactions?

1:21:09Yonatan SompolinskyI think it depends on how you, configured Kaspa NG.

1:21:17Vlad CosteaI haven't spent much time configuring, but in the overview section, it looks like I'm validating and I'm connected to a lot of peers or is this completely SPV?

1:21:32Yonatan SompolinskyI don't think it's SPV mode. I think it's either full node or not. But again, this is not my, software. There's nothing I'm, I'm looking at it, together with you. I'm, I'm more used to the command line. So, for me, it's, I'm learning, I'm learning on it together with you.

1:21:52Vlad CosteaSo now let's try to send a transaction. I would like to thank the one, two, three, four, five, six people who sent me 67 KAS. I want to send some back. And let's say someone from the chat can paste an address or you Yonatan, you can post one too, if you want.

1:22:14Yonatan SompolinskyMy, no, ask someone from the chat will take me, too many passwords.

1:22:22Vlad CosteaOkay. The first address that I see in the chat is going to receive 10 KAS from me. Just waiting.

1:22:39In the meantime, I just want to mention that I remember there was a time when the conversation around Ethereum switching to proof of stake was that they need that for DeFi because it's impossible, physically impossible to do DeFi with proof of work.

1:22:55Yonatan SompolinskyYeah, I heard that too. Enlighten me. What's the connection? I have no idea.

1:23:00Vlad CosteaThe connection is that you basically proved that you can have much faster confirmations and I guess you can do DeFi with Kaspa.

1:23:12Yonatan SompolinskyWell, DeFi, you can do, you can definitely do confirmations, DeFi on Kaspa, but also you could do confirm, DeFi on Ethereum 1.0, right? No one, no one argued that DeFi doesn't work on Ethereum 1.0. And also 12 seconds. Is it eternity in terms of, finance? So, so today Ethereum in terms of finality is also very, very, very slow. Ethereum today is like Kaspa. What's the difference?

1:23:48Vlad CosteaWell, the difference is that Ethereum is proof of stake and Kaspa is proof of work.

1:23:53Yonatan SompolinskyNo, I understood. But what's the implication of this?

1:23:57Vlad CosteaThe implication is just that there was this narrative floating around that it's necessary to do proof of stake when in fact it was disproven.

1:24:09Yonatan SompolinskyYeah, it was disproven, but I also don't, don't think I ever understood it to begin with. So I don't know how to refute it. I don't, I didn't understand the narrative to begin with. It was a very strong narrative and I found myself multiple times on the defense, trying to argue back, but not understanding that the core argument, which I think it was. If I'm, if I'm recovering the initial intuition, it was initially wasn't about finality with just about the scalability. The assumption that you can't do, I can't actually recover the reasoning behind this narrative, this myth.

1:24:58I do, I do. I wasn't involved in several debates, but I wasn't able to participate intelligently in those because I simply didn't understand the other side.

1:25:06I think it's a full mistake, but not the connection to this one, I'm sorry.

1:25:07Vlad CosteaI think it has a lot to do with orphaning of blocks, if I'm not mistaken. That was a concern under the proof of work model for Ethereum.

1:25:21Yonatan SompolinskyThat because of reorgs, you wouldn't, you wouldn't be able to support DeFi because of reorgs?

1:25:28Vlad CosteaYeah. And the block time at the time was around 15 seconds, if I'm not mistaken. There were a lot of, not a lot, but there were some failed transactions that were happening. It was actually possible to pay gas for your transaction and it was not getting through.

1:25:49Yonatan SompolinskyBecause of reorgs or just because of the general EVM dynamics?

1:25:53Vlad CosteaI guess both. I'm not sure. I'm not an expert in Ethereum. I'm not going to make any bold statements right now.

1:26:01Yonatan SompolinskySo, so just for the sake of completing the argument, the reorg thing is not important unless you are, it just has no consequence, unless you are talking about bridging to other layer, other networks. Like if you're in a Cosmos ecosystem, I can understand that you need to have stronger finality, stronger and faster finality guarantees in order to bridge to other layers. But within a cohesive system, within one consensus ecosystem, reorgs can be annoying, but they don't affect the consistency of the state. Simply that this DeFi branch is replaced or, you know, in a DAG is semi replaced by this DeFi branch that the consistency remains just replaced.

1:26:50So, and everyone sees it. It's not like, sorry, especially in the case of a block DAG, everyone sees these two branches. So it's, it is a narrative that was very, very prominent and we are happy to refute it. But I guess since then, people developed a distaste towards proof of work, regardless of whether it can support DeFi or not. Of course, there's the DeFi on Bitcoin camp that is an exception. But other than that, if you show today people on Ethereum, Kaspa, let's say Kaspa with DAGKnight confirmed transactions with a few round trips, they won't, they won't, it won't affect them much.

1:27:50It's still a proof of work engine and they won't support it or they won't be, they could be impressed like as a research project, but not more than that.

1:28:01Vlad CosteaAnd I feel like we need to talk about this because when you launched Kaspa in 2021, it was still sort of politically risky to launch a proof of work blockchain. But before that, just let me send some coins because I need to complete the quest. There was SilentPhilPSX in the chat who was the first person to paste an address. And someone else just asked me if I shouldn't send the funds to the same people who sent them to me. I think that's a bit pointless. Like I regard this as donations.

1:28:34So let me type the address and the amount. You're getting 10. What's the priority fee? Like I can manually type the fee.

1:28:47Yonatan SompolinskyThat's very similar to what's happening in Bitcoin wallets. Yeah.

1:28:52Vlad CosteaI'll just choose low.

1:28:54Yonatan SompolinskyDon't choose low for the podcast. For the podcast you choose priority. Really? Yes, because the way you measure confirmation times theoretically is only by creating, by sending priority transactions in any network. You can't measure confirmation times without taking into account quality of service. It won't make any sense. One network can process blocks within 100 milliseconds, but transactions wait long in the mempool because there's no good quality of service. Which is why, by the way, something that happens in Kaspa in September 2023, or I guess 3, because we didn't implement RBF.

1:29:57Precisely not implementing RBF caused the network to operate normally, but without meaningful, a meaningful priority feature.

1:30:12And then the network processes blocks at the same speed and on-chain transactions at the same speed, but transactions wait long in the mempool. So both conceptually and practically the way you define and measure confirmation times in a network. Proof of Work or Proof of Stake or any other network is assuming you're sending the highest priority transaction, then you're checking how long does it take for either deterministic or probabilistic finality. That's why I'm urging you to do the priority one.

1:30:50Vlad CosteaBut once again, what happens if I choose low? Let's say I'm super cheap and I don't want to pay. It's a very tiny fraction here. It's not even one cent.

1:31:03Yonatan SompolinskyWhat happened to me in the early days where my colleague, Yoan, he was a co-author of SPECTRE, which is our first DAG protocol. We didn't have lots of money, so he wouldn't be willing. He wanted me to pay for his coffee and Bitcoin, and then he gave me shekels, but we would fight over the fees. And then the same thing will happen to you if you do the low, you'll praise the low priority. Your transactions will take a long time to confirm, and Bitcoin a long time is forever. And here, let's check. It won't be informative in my opinion, but I'm happy to check.

1:31:46Vlad CosteaFive seconds. Let's see how fast this is going to be. Oh, I need to type my password.

1:31:53I have this quirk where I basically have to say my passwords out loud when I'm typing. Go ahead. No, I'm not. Send it. Let's see what happens on the blockchain explorer. This is the fee. It's been confirmed already 100 something times. It seems like the low fee did not make any difference.

1:32:27Yonatan SompolinskyThat's because the system is not interesting yet until there's full adoption, in my view.

1:32:33Vlad CosteaBut it did batch. Like, look at the way in which it constructed the 10. It basically took 0.5, a 1, three 1s, a 2, a 3, another 3. Instead of choosing, I think I received a larger UTXO here. And I'm curious why it does this. Someone sent me 100 KAS right here. Why did it not send from the 100 one, as opposed to batching all of these?

1:33:06Yonatan SompolinskyIsn't that good, by the way? The choosing to compound?

1:33:13Vlad CosteaOh, it's nice. For privacy, it's really good. Because if it chose the biggest UTXO, it would expose your stash, right?

1:33:27Yonatan SompolinskySure, it's not the perfect privacy. I know you're a fan of that route. But even just for privacy is great, but also just even for performance of the network for shrinking the UTXO set.

1:33:47Vlad CosteaBut there was also a UTXO that I received prior to this, which was exactly 10. And I guess it needed one to pay the fee, but it could pick the one that's 10 and another one that's a 1. But instead it constructed this situation. How much did you send out? 10. You can see my wallet here, basically. I don't know how I can zoom in. These are the transactions that I received and were confirmed before I sent this one. And I'm curious why it did not use this UTXO with the 10, as opposed to combining the other ones.

1:34:26Yonatan SompolinskyI hope it's intentional, because it's compounding transactions. This is very good. Compounding transactions should be the default choice of the wallet.

1:34:47Vlad CosteaIs it possible? I mean, I'm pretty sure that there are wallets that do this, but can I choose the UTXOs when I make a transaction? Let me try again.

1:34:56Yonatan SompolinskyI don't know if it's in Kaspa NG, but we'll see a lot. Of course you can.

1:35:03Vlad CosteaThat can be really interesting. Would you describe this as being an average experience for a Kaspa user?

1:35:14Yonatan SompolinskyI wouldn't know, but it's not very different. I don't see this as very Kaspa unique. This is just how UTXO works. You have UTXO selection questions in Bitcoin as well. Different wallets or wallet softwares have different policies.

1:35:45I actually read recently, I think Jameson Lopp published an article about UTXO selection in Bitcoin. I don't consider this something, this challenge or this complexity or interesting question particularly Kaspa unique. I would say that the need to compound the way we treat or regulate the growth of the UTXO set is something that is unique in Kaspa. It's both because the need arised and because we invested time in thinking what's the correct way, so to speak, to put a cost on increasing the UTXO set. Which I think is of general interest.

1:36:44I know in Bitcoin there's a lot of wars on the spam transactions. In Ethereum there's all sorts of rent discussions.

1:36:57I'm not sure what features and what are just nice ideas. If I remember correctly, today Ethereum didn't implement the rent ideas. But in Kaspa we do have something unique that is closely related to what you're seeing and doing now. Which is the STORM transaction fee mechanism. I'm not sure if you've heard of that. But either way I can elaborate.

1:37:27Vlad CosteaNo, you should elaborate. I think it's really interesting that I'm counting the inputs that happened in this new wallet. So there are 3, 4, 5, 6, 7, 8, 9, 10, 11, 12 inputs. And right now I have five UTXOs after sending out two transactions. I like the tendency of making it efficient as opposed to generating more UTXOs out of this.

1:37:57Yonatan SompolinskyAnd privacy-wise, you can refresh your address for the same wallet. You don't need to use the same address. You can create a new address with a hierarchical wallet for the same wallet. You can create many, many addresses and use them. It's not perfect privacy, but it does make tracing your transactions more difficult.

1:38:27Vlad CosteaIt's basically how Satoshi described the way you should use Bitcoin in Chapter 10 of the White Paper.

1:38:36Yonatan SompolinskyYes, definitely. Many things in Kaspa are either copied or inspired by Bitcoin. But there are some elements that are novel, other than the actual DAG. And the transaction fee mechanism is, in the context of the UTXO set, cost is something relevant. And I wish other people in crypto heard of it. I think it's quite strange that, like, Kaspa is sort of an island, right?

1:39:28We have our innovations and our research, and it's kind of segregated, almost segregated from the cryptosphere. Because you see, like, the Bitcoin wars on the spam and Ethereum, admittedly older ideas about rent.

1:39:49And I think the STORM mechanism of Kaspa could have or should have been brought to the discussion in these contexts. But it's kind of difficult to tune into these discussions because there's still, admittedly more so in Bitcoin, the strong suspicion that this is a war between, or a marketing war between currencies, rather than the early atmosphere, which was more... More about the actual protocols and regardless of whether it's implemented here or there.

1:40:42Vlad CosteaI just keep on sending small transactions back to people just to see how this works and what's gonna happen. I would like to thank everyone who basically donated KAS to this wallet. I'll find interesting ways to experiment with this. Is there anything interesting that I can do with the KAS right now? In the interface of this wallet, or maybe I can prove how I can swap it or spend it somehow? Or use DeFi with it?

1:41:19Yonatan SompolinskyYou can't use DeFi yet. We're still working on it. We're still working on implementing the ZKP opcode, which is a more ambitious... Or maybe two orders of magnitude more ambitious than the OP_CAT one.

1:41:42So DeFi and Kaspa is not... It requires still more, at least in the native way I'm envisioning it requires more patience.

1:42:00Before going to the very interesting topic of DeFi and Kaspa, and while you are playing with it, I will discuss the protocol of STORM. STORM is a mechanism to penalize or to correctly impose a cost on creating and increasing the UTXO set. Again, this is not a challenge that's unique to Kaspa. This is something that is heatedly debated in Bitcoin circles.

1:42:42And Ethereum has flirted with many protocol ideas in the past, all of them around rent. So rent means you pay transactions. Oh, this is nice.

1:43:02Vlad CosteaWhat is this?

1:43:04Yonatan SompolinskyOh, this is a blockDAG.

1:43:06Vlad CosteaSo this is a visualizer of the blocks that are being mined right now?

1:43:12Yonatan SompolinskyYes, this is like Bitcoin with internet speed. Pardon my cliché, but this is... It's very beautiful. It's clean proof of work. There's nothing that... There's nothing trick... There's no trick, there's no supernodes.

1:43:32Miners are anonymous or pseudonymous. There's no committee running this, there's no rotating committees. No BFT. Everything here, pure proof of work.

1:43:49Vlad CosteaI can adjust the visualizer. Is it still possible to mine from home with GPUs or did it become fully an ASIC chain? ASIC chain, I suppose.

1:44:01Yonatan SompolinskyYeah, now ASIC chain, yeah.

1:44:02Vlad CosteaI remember last year, and I'm sorry that I interrupted you from what you were saying, but there was a piece of news about someone who went to a hotel room and mined Kaspa and made like $5,000 or something in a few days. Wow, not me. It was sort of a scandal because the conversation was around the fact that someone basically went and consumed a lot of electricity from a hotel room for which he paid like $100.

1:44:36Yonatan SompolinskyI feel some regret to not have tuned into Kaspa News as much as you did.

1:44:45It does remind me... There were many such questions early on in Bitcoin. There were many ethical questions about people using university clusters, even genuinely, just to track the network, but on the side to have it mined or people using public facilities to mine. I did not know this happened with Kaspa, but I told you Kaspa is like Bitcoin, so I guess it's just a proof of that, right?

1:45:24Vlad CosteaI guess we all know at least one university professor who's mining Bitcoin in his lab, or at least used to. But anyway, last year, Kaspa was the most profitable proof-of-work coin, and I don't know exactly what happened to that. Maybe there was a halving of rewards or something, but it was quite a big deal. Even the guy from the Bitcoin university who gets a lot of views on YouTube decided to make a video that was called Kaspa the new Bitcoin. I feel like the arguments that he picked were sort of low blows or the lowest hanging fruits that he could, but still it was...

1:46:09Yonatan SompolinskyWas the answer yes or no?

1:46:11Vlad CosteaHis answer was no, because it's impossible, right? You can't have the same immaculate conception that surrounded Bitcoin in its infancy.

1:46:21Yonatan SompolinskyI tend to agree, yeah.

1:46:23Vlad CosteaI do agree with that, but at the same time it doesn't mean you shouldn't try. You shouldn't stop people from attempting to do something different.

1:46:36Yonatan SompolinskyI mean, again, it's okay that there's a cultural prohibition or press against launching endless or useless L1 tokens, L1 platforms or tokens. I'm not very critical of that sentiment. It has its role. I'm not turning the other cheek as well. As you see, I'm less informed about it. I'm not that tuned into these kinds of media at all. I typically don't know about it. There's a room for this cultural press against it, but again, it's not an attempt to do a new Bitcoin in the sense that, again, Bitcoin doesn't have lots of aspirations these days.

1:47:35So it's not a competition even in the sense that, again, Bitcoin, other than a few teams that are trying to push the L2 DeFi narrative on Bitcoin and trying to get approval for OP_CAT or trying to circumvent the script to implement sort of a hacky way OP_CAT-like functionalities. This is not the same kind of aspirations. Bitcoin has a very strong value proposition, and for that value proposition, you don't need 10 blocks per second.

1:48:24You don't need even one block per 10 minutes. You can have one block per hour. It's not the same kind of system.

1:48:34It's more about implementing the proof-of-work values or architecture or principled architecture for a broader contribution, for a broader mission. Bitcoin is not attempting to free finance from centralized choke points and is not interested in particularly speaking in Bitcoin as a singular entity, just for further simplification, for the sake of conversation and clarity. But Bitcoin is not today aspiring to be anything other than store of value, as you mentioned.

1:49:23And so I don't think it's reasonable to compare it in the sense that, OK, Kaspa is trying to be the new Bitcoin.

1:49:30Bitcoin is, it is trying to do things that we thought, that everyone thought Bitcoin would do, would want to do. This is maybe more a better interpretation. As you mentioned yourself, if the ZK roll-ups or even not only roll-ups, even on the tech ideas, the experimentation, the insight that the community gained, if all this was ready back in 2011, 2012, then the whole ecosystem, the whole discussion was different. And then Bitcoin could have been, you know, could have brought in its aspirations or, sorry, or at least wouldn't narrow its aspirations.

1:50:19And then you could have described in that rewrite of history, you could have described this blockDAG that you saw as an attempt to... Mimic or outpace Bitcoin, become the new Bitcoin. But since this is not the case and history had its own, chose its own path, then I don't see it even at competition. Bitcoin's aspirations and value proposition are very valuable. And they don't require more than a slow, more than my friend here, my Bitcoin friend here.

1:51:04Vlad CosteaI wanted to ask you something based on the reply that I got in the chat. Someone sent that he sent 101 KAS from Brazil. And in the transaction ID, there's a message that he attached. Is it possible to send messages using the blockchain and broadcast them publicly?

1:51:30Yonatan SompolinskyWhich block explorer shows the search-based... You do have the transaction hash? And the block explorer does not show that? Let's see. Let's learn together. I'm not...

1:51:47Vlad CosteaIt is here, so I do have a hash.

1:51:52Yonatan SompolinskyOkay, the payload says nothing.

1:51:55Vlad CosteaNo, it doesn't.

1:52:02Yonatan SompolinskyLet me ask in the group. I'll ask the folks, if anyone knows of a tool.

1:52:10Vlad CosteaBut there is also someone mentioning that I should try some kind of messaging app that uses Kaspa, Kasia or whatever.

1:52:18Yonatan SompolinskyYeah, I heard of that. I even installed or connected to it and it will take me time to get to it because there's lots on the plate. But yeah, I like this experimentation, whether it's a good use of blockchain or not. At this stage, I like these experimentations regardless of their principal merit. Let me just ask in the chat.

1:52:58Vlad CosteaAlso, I have one last sponsor to plug in this episode and that's SideShift, which is the kind of exchange where you swap stuff. So if someone offers to pay you in stablecoins and you don't want to hold on to that or some friend says, well, you should try my coin. And it's this silly meme coin on Solana. You can actually swap that for something that you want. It might be Bitcoin. It might be Ethereum. It might be something else. Unfortunately, for now, they don't support Kaspa and actually tried before this interview to see if they listed it.

1:53:33I'm not sure what the listing requirements are, but it would be cool if I could use this Kaspa to swap it or whatever, just to provide more liquidity. And the URL is SideShift.ai. SideShift.ai. They have a lot of liquidity and pretty low fees. Unfortunately, I don't think they operate in the US. But then again, I'm not going to tell you exactly how or that you should do it, but there are ways you can circumvent the IP filtering.

1:54:12Yonatan SompolinskyWhat's the transaction? Oh, sorry. Did you finish the last question?

1:54:14Vlad CosteaIt's fine.

1:54:16Yonatan SompolinskySo if you can post the transaction hash.

1:54:25Vlad CosteaJust a second. No, I clicked on address. Let me go back. Transaction ID. And this is the hash. And where do I post this?

1:54:39Yonatan SompolinskyI'm posting this on Twitter.

1:54:42Vlad CosteaYou mean in a DM on Twitter?

1:54:45Yonatan SompolinskyNo, I'm posting it as a reply to Vlad hosts the best podcast in Bitcoin.

1:54:51Vlad CosteaOh, OK. So let me put this in the chat here on Google Meet. This is the hash.

1:55:17So I do like when I look at this Blockdag visualizer. It's really amazing. Is there any pruning involved? I forgot to ask you about this. How do you make sure that the blockchain doesn't bloat and doesn't just become very data hungry?

1:55:35Yonatan SompolinskyWell, there's lots of pruning involved.

1:55:40Vlad CosteaIs it sort of like Mimblewimble or different?

1:55:46Yonatan SompolinskyNo, Mimblewimble is not compatible with DAG. At least not the vanilla protocol. Because in DAG, you combine any block, the accepting block, as you saw, the block in a DAG. It references all predecessor blocks, even if they have conflicts. But then this will, it will be impossible to create a cryptographically sound kernel in Mimblewimble terminology, if I recall correctly, that proves the correctness of the new transactions if some of them conflict, some of them are double spends of another.

1:56:40So Mimblewimble DAG, we tried it. Ethan Heilman, back in the day, from Boston University. We tried to create a version of this.

1:56:53We called it Azkaban. But this was actually, interestingly, the first attempt at Kaspa to create, this was Guy Corem's idea to create a synthesis of DAG, of SPECTRE at the time, with Mimblewimble. And that became the Azkaban attempts. But nothing serious came out of it because it was either, so it's impossible with the vanilla protocol. And I don't think there was enough interest from enough people to merit launching these things together.

1:57:44Although, it is cool, I admit. It is how we started. Also, Mimblewimble has lost traction, sadly, very fast.

1:57:54I don't know why, but I guess the launch was too fair in that sense. So Grin, at least the Grin version, lost traction. I don't know if it's active today, the community. But I didn't hear, I at least didn't hear any activity from that area, unfortunately.

1:58:16Vlad CosteaWell, what happened is that one of the main developers from Grin ported that onto an extension block to Litecoin. So now Litecoin basically has Mimblewimble on the extension block. And it's the most popular, or at least the implementation that has most users and most liquidity right now. But also Beam is pretty interesting, the way that they were able to evolve past the limitations of the interactive design. And they added something that's called Lelantus to be able to enable users to receive offline payments and also do all sorts of DeFi.

1:58:59I mean, the ways in which they were able to extend that basic design was interesting. It's just that for some reason it did not get a lot of traction.

1:59:10Maybe it's marketing, maybe it's just... I remember in the Bitcoin maximalist circles there was a quasi scandal involving Trace Mayer. I think this was January of 2020 when he went to a Las Vegas conference and he was handing pieces of paper promoting his Mimblewimble coin MWC. And he was promising that it's sound money which is fungible and programmable and with fast transactions and instant settlements. And it had a huge pre-mine, like 60% or something, and people were very critical of that. And it was because of that incident that some people stopped paying attention to anything Mimblewimble, even though Grin was the fairly launched one and Beam was the one that was trying to do more interesting token kind of stuff in DeFi.

2:00:09Maybe that was the moment. But there was also a reveal at one point where an article pointed out that you can destroy the privacy assumptions of Grin just by having a listening node that does not prune any of the data. So you're running a full node.

2:00:27Yonatan SompolinskyYeah, that was known in advance, that if you listen to the peer-to-peer network and you listen to all transactions, then there's no privacy in the mempool. No one argued otherwise. I guess this article, I recall this paper, but this wasn't, this was more of an implementation or, I don't recall if it was a simulation or an actual breaching of the protocol, but that wasn't, that protocol didn't pretend to have that level of privacy.

2:01:11Vlad CosteaThere are people in the chat who really want me to ask you about vProgs and DAGKnight.

2:01:20Yonatan SompolinskySure, but I do want to, if it's okay, ask you, what do you make of the Grin launch? To me, this is a very interesting experiment because it was both GPU-friendly, not only fairly launched, but I would argue too fair in the sense that the minting curve was, if I recall correctly, constant. So something in the minting curve and in the GPU friendliness just didn't work out and very early on, like within a few days or weeks, there was already a big disappointment. I'm just wondering if you, if it's okay to get your take on that.

2:02:08Vlad CosteaI think the tokenomics, quote unquote, or the pumpamentals or whatever you want to call them, disappointed a lot of people because there was no advantage to joining the network early. Like everyone who discovers something on day one, they're like, yeah, I'm going to be one of the insiders. This is my chance to be Hal Finney or something, right? But in the case of Grin, the issuance is constant and you get basically 60 coins per minute, so one coin per second, and it just keeps going in perpetuity.

2:02:39It inflates at this rate. There is no halving cycle. There is no nothing. It doesn't matter when you get involved because there is no advantage for people who discovered this early. And I guess those who did discover Grin early were sort of disappointed and said, well, what's the point? Because right now it's half baked. Maybe in a few years it's going to be a lot better. So what's the point of trying to use this now as opposed to waiting for this to get developed? And there is no incentive to hold it.

2:03:12It's supposed to be used. I think this sort of killed it. But as an experiment with a fair launch, I think it's genius.

2:03:24Yonatan SompolinskyIt's genius in what sense? If it's launch specifically, if it's failed specifically because of the fairness.

2:03:33Vlad CosteaI wouldn't say it failed. It just turned away the people who are willing to invest in it.

2:03:40Yonatan SompolinskyIt turned away also the devs themselves. They were outmined immediately by all the hype. There was too much hype around it. And they immediately, after a few days or weeks, they lost interest or they said they need some extra incentive to participate. So I'm not sure in what sense. It was a genuine launch, but it's an experiment that's very telling to us what's the cost of overly fair minting curve.

2:04:12Vlad CosteaI think also the kind of people who are working on Grin are very disinterested in making money. They just want to contribute to this from a purely mathematical and cryptographic point of view. And they don't care about anything else. And I've met a bunch of people who do this. But of course they invest in Bitcoin and other stuff too, just to be able to support their work.

2:04:39Yonatan SompolinskyI think the sentiment is sustainable as long as other people don't make loads of money off the project. I think that sort of asymmetry is less sustainable, maybe less motivating. It's okay to have, similar to my original creative horrible game net idea, it's okay to have this project that is not aspiring to be profitable or to have an active growing effort. But if other people are making financial business over on top of it, I think this is the point where your motivation, your energy drains. And I think this happened very quickly with these Grin devs.

2:05:41They were happy to do it as a fun experiment with a supposedly anonymous author. But it became less fun once all VCs hired, sorry, rented lots of GPU farms to mine, to outmine them. I think they lost the interest and right, I think right before, at least, very, very understandable.

2:06:13Vlad CosteaI remember the day when Grin launched very clearly. And I remember there were some people trading one Bitcoin for one Grin. It was surrounded by a lot of hype and there was a lot of expectation from it that it will, basically it was the second Bitcoin in the way that some people were thinking about it. There were even guys who were concerned that it's so good that it might just replace Bitcoin. Because it scales much better, it has better privacy. If you can figure out how to do transactions without being online all the time, then you can build some very interesting stuff with it.

2:06:55It's just that I guess people lost faith very fast. It was mostly economic, it was not technical.

2:07:07Yonatan SompolinskyYeah, I see no one answered. I'm just trying to check if someone answered them.

2:07:17Yeah, I think it was very clever. Okay, I find it personally difficult to believe that the protocol was authored by an anonymous, by someone that was not part of the original group. It was clever. I'm not saying in a cynical way. It was clever or maybe not only clever. It was nice and fun to have the Mimblewimble author pseudonymous.

2:07:57I think this is the aspect that, this one aspect that created lots of hype and of course the fact that Andrew Poelstra, if I remember correctly, was endorsing and actively working on it. It kind of kosherized the project. And VCs back then were, you know, as a Bitcoin Core dev, back then and I guess even today, if you are actively working and developing a new L1, there's no chance VCs won't launch their predatory strategies on your protocol. But I think the point I'm trying to push is that this anonymity of the originator was, to me, less credible than others, I guess.

2:08:58And others have pulled off such anonymity in other protocols. I prefer not to mention names, but you're helping an anonymous author or team of authors create an anonymous paper and then out of the blue, you found this paper and you're interested in it and you want to contribute to it. I'm not saying Andrew is part of it. I'm just saying I found it. I'm not at all claiming Andrew is part, has conspired against anyone. I'm just saying I found it very weird. I didn't think the author was truly detached from the scene.

2:09:49Vlad CosteaWell, allegedly, so I know this from YouTube presentation from 2016, Andrew Poelstra heard about Grin from Bryan Bishop and they were on IRC and someone by the name of Tom Elvis Jedusor just casually logged in and dropped this paper and never came back. And Andrew Poelstra sort of refined it and also came up with this proposal that this could be added to Bitcoin as an extension block feature. So it was him who came up with the Mimblewimble extension block, which is right now being used on Litecoin.

2:10:26But interestingly, when I interviewed him last week and I asked him about this implementation, he said it's a bad idea and he was naive and he doesn't believe in extension blocks anymore and he doesn't think Mimblewimble is the best way to do privacy on Bitcoin.

2:10:46Yonatan SompolinskyOkay, and Bryan Bishop was active on Grin, I didn't know that he's part of the founding group there.

2:10:53Vlad CosteaNo, I think Bryan Bishop is just the guy who told Andrew that this paper dropped and it's interesting. I'm not aware of any other involvement after this.

2:11:08Yonatan SompolinskyAnd what would be the best way to do privacy on Bitcoin or on Kaspa?

2:11:13Vlad CosteaI think there's a lot of merit to zk-SNARKs. They have become very lightweight over the last decade compared to what they were like when they were introduced as Zerocoin and Zerocash.

2:11:36I mean, I remember just a few years ago you were not able to run a Zcash wallet on your phone. It was not computationally possible. But now, thanks to phones getting better and the math getting better, we can get that. And there's this have you read the Tachyon paper by Sean Bowe from Zcash. Basically, he is describing how he was able to figure out how to scale Zcash to billions of people. And if that happens, and now that everything is open source without any sort of asterisk that you have to ask permission or reference this author, I think this can be used on Bitcoin or Kaspa or anything else, like something off the shelf. You just take the libraries and you figure out how to implement that.

2:12:28Yonatan SompolinskySo with the ZK opcode that we are developing, this will be very possible. I do think from an adoption perspective, you can't afford or you don't want to implement privacy natively on layer one. It has its appeal, but the price you're paying is it will be much harder to be listed and be used by ordinary people. Of course, Zcash has its appeal.

2:13:15I'm sure Zooko sees it as completely worth it, the privacy versus the difficulty in being listed and being integrated into the normie user base. But it's a choice that I think Kaspa couldn't afford in terms of adoption and integration.

2:13:46But once we implement the ZKP opcode, people would be able to choose to use it. It won't be the default address, if I recall correctly. Even in Zcash, the shielded transactions weren't default. But it will still have its merit.

2:14:07Vlad CosteaIt's really nice if the adoption of Kaspa happens organically with users setting up their own wallets, running nodes and being their own bank. Because in Bitcoin, I see the narrative shifting with even the people who are very adamant that you should hold your own keys. Now they're saying, ah, maybe that the average person is going to have some difficulties in doing this and maybe it's better to just buy ETFs or whatever. I don't like it because the whole point of this movement was to get people to hold their own keys.

2:14:44If you don't hold your own keys and you don't use your own wallet and you don't validate your transactions, we go back to PayPal, basically. And honestly, that's really not the reason why we're here. I'm happy that there are a bunch of people in the chat who are enthusiastic and they're recommending wallets. I think this is the nicest part of this whole industry, the fact that people can hold their own keys and manage their own coins and do whatever they want without asking for anyone's permission.

2:15:19Yonatan SompolinskyDefinitely. And specifically in the case of Kaspa community, there are lots of the spread of information about wallets and how to use this. As you saw, I'm not the best candidate to explain how to use these stuff. And it's very nice to see that people have lots of experience in doing these things on their own. I wanted to double down on what you said. I strongly agree with you. I think that the lack of user-friendly wallets for normies, not for people that are tech-savvy, is a big barrier.

2:16:10And I really like the smart wallets. Now they are called account abstraction. Now they're under the umbrella of account abstraction, which is social recovery, but not through signing the ledger and sending it. It's not recorded off-chain, but it's recorded rather in consensus. So you can do things like in consensus, declare things, declare policies of the sort. If this other address wants to take all of my claims that I lost my keys and wants to take my coins, then after maybe six months of me not responding, allow them to take 2% per day.

2:17:12You can do these policies, and of course you can generalize it to multi-sigs and to K out of N. But the point is that you implement these general policies on-chain in a way that doesn't require your grandma to supervise the whole process. And I argue that this is a strong prerequisite for self-sovereignty. You can't be your own bank. The whole not your coin, not your keys slogan is very elitistic for the tech-savvy. It's not yet implementable for ordinary people. And if you do implement in consensus, and I'm not necessarily promising this or promoting this as an L1 feature, although I might in the future, but I am saying that this should be the default on-chain feature, and this guarantees self-sovereignty.

2:18:18And there's another reason why this is self-sovereignty, because there's the basic hammer attack where someone steals, you know, just, you know, goes to your, knows you and goes to your home and forces you to empty your wallet. And if you have on-chain rather than off-chain social recovery, then you can simply, you can actually prove to your captor that you can't release the coins. Say here's my private key, log in on-chain, log into the wallet and see that you can drain my wallet only 2% per day, starting from six months from now.

2:19:01It's a feature for self-sovereignty that provides resilience against these sort of, these sort of more common attacks. So I view this as very important, both in the context of normies, but also just to protect the sovereignty of others.

2:19:26Vlad CosteaAnd I also want to add something which to me is interesting. Jameson Lopp has published a series of articles about this. There are situations where your electronics are just not good enough for long-term storage of coins, and you should have some sort of backup. Of course, you're encouraged to write down your seed on a piece of paper, but if your house catches fire or there's some sort of earthquake or flood, and you need something stronger that's going to withstand all of these hazards, you should look into metal backups, which ideally you can also, if you want to hide them or something, you can just drill a hole in the wall of your apartment and put that inside, or you can bury it in the garden or whatever you think is interesting.

2:20:18The one that I usually recommend is Cryptosteel, because they have been around for like a decade or something, and they're like the ones who started this whole movement of metal backups in Bitcoin. And if you use, I actually have a code for discount, so if you use BTCTKVR, which is Bitcoin Takeover, you get like 10% discount, so use that if you can, if you're interested. I have a bunch of them around the house. I'm not gonna say exactly what they are and where I keep them.

2:20:51It's just that they're very convenient and you can put any password. It doesn't have to be a seed phrase or passphrase or whatever. Even your email password if you think you're gonna forget that one day.

2:21:08Yonatan SompolinskySo this is actually the divide between the normies and Jameson Lopp or the Bitcoin elitists. Because Jameson, and again, he's a positive person. He influenced my Bitcoin learning and conversion immensely. But his image is precisely, his way of protecting the sovereignty of his wealth was living in the forest in an unknown, undisclosed location.

2:21:47Disconnected and secretive. And this is very, I love the story, but it's exactly not a movement that can scale to the masses. It's a very elitist religion or movement. And this is what I said that I think in one of the early blog posts for the lack of white paper. In the inception of Bitcoin, of Kaspa, I said it's sort of a Protestant version of Bitcoin. Because this very idea, the Casa nodes and the metal inscription of security of your keys. These are all things that I love. They are interesting, they are fitting.

2:22:49It's also important to have in your hardcore a group of people that are adamant about these things and that don't need any auxiliary solutions and don't use any solution on chain, etc. But for regular people, I argue that you have to scale these values and you have to compromise, you have to introduce some compromise. I remember the SPV wars. I remember in Bitcoin, if you ran an SPV node, you were evil. That was the level of fundamentalism. And again, I like it to the extent that it's the bedrock of the movement, but then it's also limited in its influence.

2:23:38And my final point, by the way, what I'm doing now is pitching to the other core builders why I will ask them to implement this feature in a KIP. That's why I'm kind of belaboring the point. But I argue that whether through account abstraction or smart wallet methods that I mentioned previously or through something else that you or Jameson can come up with, you should have the ability to prove to your attacker that you can't release the coins. This is a necessary feature. And then you don't need to live in the forest.

2:24:23I remember when I invited Jameson to scaling Bitcoin in Israel, I signed in the email a verse from Jeremiah, from the Old Testament about God seeing you wherever you are present. So that's kind of the divide between the more technical elitist values of Bitcoin Core versus the more Protestant atmosphere in Kaspa.

2:25:02Vlad CosteaIs there anything elitist about telling people to use their own wallet and secure their own seed? I don't think there's anything elitist about it. It's just like your email account where you have to secure your own password. The difference is that when people set up emails, they have the expectations that they can reset their password and they just call someone or send a message to some sort of support desk and they get a password reset, which doesn't happen in this double key situation where you have a public key and the private key. The private key is sort of permanent to your wallet. There is no way you can rotate that to unlock the same funds.

2:25:50So you have to understand responsibility. And ironically, if you explain this to boomers, they're going to be much more receptive and get it as opposed to millennials or zoomers who are so used to forgetting stuff. And they're like, yeah, I'm just going to skip this. I'm not going to write down the 12 words.

2:26:10If anything, I'm going to copy them to clipboard and put them in some notes app in my phone to which like 100 apps have access. And yeah.

2:26:23Yonatan SompolinskyWhat do you lose by, so you can only benefit by additionally adding smart wallet governance in consensus. This is what, this was the Blockstream. I remember, the acquired GreenAddress, I think the Italian, project and, the product there was precisely that, sorry. That wasn't precisely, it was one feature that was explained to me in length by Adam, by Adam.

2:27:10He, he invested a lot of time educating me or not, like taking time one-on-one to educate me on his, on his, on this product.

2:27:27And the GreenAddress had this sort of policy, right? Where it's a one out of two, a multisig. And if, sorry, I think it was two out of two multisig. And then if one of them is inactive, if one of them is inactive, like confiscated by the FBI, then after six months you can release it. So what would you bet? What would you lose by generalizing this and implementing this sort of logic on chain?

2:28:05Vlad CosteaSo can you repeat the question? What would you lose if?

2:28:09Yonatan SompolinskyI'm saying that I understand that you can expect people, you can understand the educative value of, or the cultural value rather of educating people about responsibility and taking care of their own, passwords.

2:28:35Okay. So, so with, and assuming you did this, you know, you converted everyone to this response, to self sovereignty, self sustainment, you can enhance the security of everyone by additionally encoding the logic of what happens with their keys, with their on-chain UTXOs by implementing more expressive logic on chain.

2:29:04Vlad CosteaOh yeah, I agree with that.

2:29:06Yonatan SompolinskyIt doesn't come in instead of this educative movement, of reviving boomer values, but rather it's, it's just enhancing the security of everyone. And I think there, in that sense, you lose a lot if you're, if you're freezing your scripting language, in the way Bitcoin does without any, without any willingness to open, to implement new functionality, even in, even just, even for the mere use case of store of value, there's lots of more, features you can implement.

2:29:44And if I if I believe, and I believe this is what Jeremy Rubin has tried to push for like with his CTV and other and other opcodes, he, he tried to enrich the scripting language because exchanges and HODLers had a difficult time of implementing by their, by their own the involved scripts securing their funds to just push for all this, all these features that are still real, are still in relation to either payments but also to just for a value into L1 consensus and so far has failed to do so.

2:30:28Vlad CosteaYeah, I think the reason why we didn't get the two of two multisig to become mass adopted is that Bitcoin doesn't scale to incorporate that. It would become way too expensive for the average user to sign a multisig every time they send a transaction from their wallet. And in the case of Jeremy Rubin, he promotes CTV, which is Check Template Verify. It's an opcode that he's been trying to push with varying levels of success over the last four or five years. And initially it was called Secure the Bag.

2:31:06That's sort of like an Easter egg about his work. He is trying to make it scale with something which enables vaults and a lot of interesting use cases. It's just that for some reason, the community has so far rejected his proposals. I agree.

2:31:30Yonatan SompolinskyAnd I know that he was, I think he was trying to implement more than Check Template Verify. I think it was a bunch of opcodes or scripts that he wanted to turn into opcodes. I don't recall the entire, the entire evolution of this, but it didn't start just with, Check Template Verify. But in any case, yes, I think even two out of two multisig is not, is not, it could be, it's not sufficient or at least that could be highly, the security of this could be highly improved if you, if you, if you implement, more, more logic, vault logic or some, I guess, I guess covenants can cover most of the things that I'm thinking of, to be honest.

2:32:19I think, yeah, maybe it's, maybe covenants should be, maybe it's, it's good to do these things simply through covenants, which are, which are by the way, implemented in KIP-10 in Kaspa. So this is something, maybe we should just write these scripts, but it's easier with vProg than zk, and this, the layer, the so-called layer two. So, I'm not sure, I guess I'm not sure if it makes sense to do these things through the native script of Kaspa, which is similar to Bitcoin, just with covenant or wait for the, or the vProgs to be implemented.

2:33:05And then you have like, you can, you can, you can boot bootstrap your own mini mini zk VM, just for your own UTXOs to lock your own UTXO there. And then you can add also, some privacy. I do think that if we do it in layer one, then, or through a covenant, then there's, yeah, arguably it's, it's more secure or more, it's a harder route to take. And therefore it's kind of, paradoxically is more secure in the sense that you'll do things more raw.

2:33:51You'll, you will treat the stack more seriously when you're implementing these things, rather than the more Turing-complete free software kind of practice that you're doing when you, when you're implementing mini ZK VMs. Anyways, this is, this is a more of, a rambling about how to do so, but, I just, I just really want to see this feature. Implemented not only in Kaspa, but also adopted as the default, the only legitimate way or default way to hodl your coins.

2:34:35Vlad CosteaIt is sort of like a multi-billion dollar question. Like how do you get people to self custody in a way that scales and prevents them from shooting themselves in the foot? I think social recovery was popularized by Vitalik and I think that he himself has set up a scheme like that to make sure that in the event that he gets abducted or he gets held at gunpoint or whatever, there is nothing that he can send except for maybe the tiny amount that he holds in a spending wallet.

2:35:11And there are lots of these ideas about vaults and multi-sigs and the ways in which you can manage funds. And it would be interesting for me to see how Kaspa can implement some of these to prevent average users from getting rugged from basically sending their seed phrase by accident to a third party. But also to have varying levels of security so that if you do really want to withdraw all the coins from your wallet, you can basically do that reasonably fast.

2:35:55Yonatan SompolinskyYou can do, you can implement this latter feature by encoding in advance what public, what addresses of other people that you trust can drain, can drain your funds quickly. I think the social recovery, yeah, I think the social recovery that Vitalik, promoted at the time at least was, or initially was off chain. It wasn't, it was not, it was not, encoding things in consensus. But maybe with the advent of account abstraction, the EIPs, it makes more sense to do it off-chain.

2:36:40But you have to do it on-chain, and in any case, I will, I will, I will forever be in favor of doing it on chain. I think it's, again, also easier to prove, to prove your logic so that, so that everyone knows that there's, that there's no, there's no benefit of abducting you. Like, I don't know if the protocol of Vitalik is provable to his abductors.

2:37:10Vlad CosteaI don't know either, but I remember at first when I read about that, I was thinking it sort of defeats the purpose of the whole movement. The fact that you get to choose who you trust in a system that's supposed to be trust minimized or even trustless, but in his case, I sort of get it. Like he's under a whole different kind of scrutiny. I think he posted an article on his blog recently and he described that he cannot travel without being photographed in the airplane, in the airport, everywhere he goes. I guess it's sort of frightening.

2:37:54Yonatan SompolinskyHe doesn't look particularly frightened. I do want to mention, I will come back to me, I forgot what I wanted to say about this. No. Yeah. I want to, I wanted to ask why you are combining the trustless or the minimize trust in the system and you're extending it. And I'm wondering why to the social layer. And I'm not sure I agree with that extension of the trustlessness of the trustless values. Why, why would you not benefit from existing trust, not in the system backbone, the system backbone should be completely trustless, but between, in the social layer, if I trust you.

2:38:46So this is this, I mean, ideally an efficient system would encode this trust. Critically, you know, in consensus, either, you know, either in plain text, meaning plain pseudonymous, or just in a, in a commitment in a cryptographic commitment, but ideally the system should encode the trust relationships we have. I think this is one of them. This is a conflation that I think should be avoided. Like, and this goes back again to what I was alluding to when discussing Jameson's, admirable, lifestyle.

2:39:34Like normal people have trust relationships and this relationship is from an information theoretic, perspective or analogy is information. And by pushing against this, encoding or using this information, you are by definition creating or suboptimal inefficient systems. Like if I know that I trust, let's say, Vlad, okay, and your public address, why not encode this information? For instance, in a, in a smart wallet transaction that says that no one can drain my funds. To that, to this and this extent, other than Vlad and Alice and Bob. Why not encode this?

2:40:25And I think that this kind of, rejection of trust is also, it drives, I think Amir Taaki in his criticism of the Bitcoin culture has also alluded, several times to this point. I'm not sure if you're familiar with his project, DarkFi.

2:40:55Vlad CosteaOh, he was on my show, like six or seven months ago, we spoke for seven hours.

2:41:02Yonatan SompolinskyOh, okay. So, didn't he pitch you the concept of trust? Of course, it's, it's, it seems like contradictory because he's, he's doing an anonymous DAO, but the point of this, of his, of his network and his, protocols is to facilitate encoding of trust, admittedly through trust, between pseudonymous members, but still it's not, it's not rejecting the, value and information that stems and emerges from people's trust from pseudo people's trust.

2:41:40Vlad CosteaHe did mention the DAO, but he focused mostly on explaining why it took him so long and how many redesigns he did along the way. He said he started with Mimblewimble and ring signatures, and then he decided to switch to zk-SNARKs and Halo 2. And he should be launching this year. Hopefully Amir is very good at coming up with ideas, but historically speaking, he hasn't finished too many of them. And people really expect to see DarkFi because it's going to be basically the first baby that he delivers all the way.

2:42:26Yonatan SompolinskyYeah, I'm also i'm anticipating it, not necessarily as a, as an avid user, but just, again, from the personal, yeah, the celebration of his own achievement and delivering something alive. But bad, but back to the point, I think the rejection of trust it should not, there's, there should be a strong rejection of trust in the in the infra, in the backbone protocol. There should be rejection of trust in the devs and in the, in just in the in the infra and anything that's in the, you know, in changing the repo, as you as you mentioned at the beginning. But if you are, if you are rejecting trust altogether, trust in people that you trust, not that the protocol trust, then you are by necessity creating an inherently suboptimal ecosystem.

2:43:24And I think this is, this is, this could be tied to the state, to this distaste towards finance, but, that would be, a farfetched or a long argument or a longer argument to me.

2:43:45Vlad CosteaWell, I do like the fact that you can have the option to not trust. Like, I think this is part of the revolution that we are organizing right here. It's that people don't need to trust in what other people tell them. They can verify the code. And if they cannot verify the code, they do need to trust someone else. And it's their choice who they're going to trust. But the point is to minimize politics and go back to what Nick Szabo was writing in terms of social scalability and also God Protocols, or at least this is the way in which I see it.

2:44:27Yonatan SompolinskyUm, but you did, but you did extend it. Maybe I miss, I misread it, but I, someone can extend this trust. This, this, is this opposition to trust voluntary trust as well, right? To, to a cultural prohibition on voluntary trust, like in, like in the case of smart, of social recovery. Right. And this is not, this is not a mouth slip. I think, right, crypto made the word trustless a positive, right?

2:45:12It's it, before Bitcoin trustless was synonymous with untrustworthy, but now trustless became, it became a positive and I completely agree. I'm removing any anchors of trust from the backbone protocol, but at the same time in the social layer, I think we have to encode trust because trust exists. It's not, and anyone should be not only able to trust other people, but also to make use of this trust through interesting, I don't know, through interesting protocols. There's, there's various ideas, whether financial or other ideas that you can, develop once you have a trust network.

2:46:00I believe Amir is building his dark DAO, again paradoxically, because it's, it's sound like the ultimate anonymous network, but he declared the, implements it to facilitate trust networks.

2:46:08Vlad CosteaNow, once again, it goes back to the argument that trust scales very poorly, and it's very difficult to identify someone you can trust. And it's useful to have tools to prevent this from being necessary. Of course, you're going to trust people, but you don't have to trust them for everything. And you don't have to establish political bodies that mitigate trust and basically reinvent the wheel.

2:46:55Yonatan SompolinskyI'm with you on that.

2:46:57Vlad CosteaI remember when people in the Ethereum camp were saying that you can enforce constitutions on the Ethereum blockchain. And I said that to one of my PhD supervisors at the time, and she was like, this is scandalous. We fought hundreds of years to have our constitution and these tech bros are coming along and saying, well, you can put your constitution on the blockchain and we can verify whether or not something is constitutional or not. All of these quirky ideas that were used to get more attention to the potential that this technology can have was really exciting at the time.

2:47:39And I still think that there are some situations or contexts where you don't want to trust other people. And it's nice to have tools that make you not need to trust someone else with this.

2:47:54Yonatan SompolinskyIn most situations, you don't want to trust people, other people, and I definitely don't want people to trust me. The vast majority of interactions do not merit general trust. As you said, even when you trust someone, you trust them maybe not to steal your coins, but maybe they can lose your coins. So you trust them for one thing and not for another thing. So one-dimensional trust is not very telling and I think this is why, or not very informative, and I think this is why Webs of Trust didn't catch momentum.

2:48:36They were very limited to PGP or the recursive signature. It wasn't really useful in any scalable sense, as you mentioned. And I agree, trust is not highly scalable and even if it were who to trust and what to trust, it makes more sense to trust a protocol or a code than people. But when you look at the system, when you zoom out and look at the system, maybe at civilization from a bird's eye view, you see that there is trust in institutions. I'm not advocating for top-bottom institutions for government or authoritative institutions, just to trust grassroots institutions or people.

2:49:36It's not that I trust people or institutions necessarily more than you do or more than Jameson Lopp does, not at all. But it just doesn't make sense, even just from an information theoretic perspective or from an aesthetic or inherent view on the world, you see people exert or interact through trust. So in these cases, the fact that your system cannot encode it is a testament to the fact that your system didn't achieve the Holy Grail yet, didn't embody...

2:50:28I didn't allow people to embody their actual activity in a trustless and reliable system.

2:50:40I'm not advocating trusting more than you do, I am advocating for a system to embody and encode whatever trust exists, to allow people to encode whatever trust exists in the system in their own UTXO, in their own sovereign UTXO, in their own sovereign vProg or ZKVM. If you have a protocol where... Let's choose another example that's not a smart wallet. If I have a protocol that wants to follow Vlad's financial investments, okay? It's...

2:51:21Vlad CosteaNo, that's a bad idea. Don't do that.

2:51:24Yonatan SompolinskyLikewise. But let's say... But the fact that it's a bad idea, it's my own problem. It's my poor judgment, and it's not infringing on anyone else's sovereignty. But I should be able to encode this, right? To encode this distrust in your judgment in the system, and the system should serve me in the sense that it can follow my judgment, my encoded judgment, and route my investments through your activity. Now, of course, you can hide it if you want, but if you want, you can also reveal it for altruistic or for financial incentives.

2:52:08This is just an example. I'm not saying that I want it or this is necessarily a feature that I'm missing, but it is from a research or from a fundamental perspective. I would be... I would feel uncomfortable if a system, a trustless infrastructure, does not know how to encode and utilize voluntary trust relationships that people do have in whatever context of trust they have.

2:52:39Vlad CosteaYeah, I get that. But once again, I'm still very much in love when I read something from Nick Szabo about God protocol, social scalability. I love that. And all of these stories about how money basically enabled us to trade with complete strangers. When you think about it, this is so revolutionary that people who knew nothing about each other and didn't care or maybe even hated each other were able to trade without any other barriers just because they agreed that this neutral element has value.

2:53:23Yonatan SompolinskyYou think this same principle extends to finance?

2:53:34Vlad CosteaWell, how do you define finance as in DeFi or...

2:53:42Yonatan SompolinskyNo, it's traditional finance, whatever, generally.

2:53:45Vlad CosteaOh, that one requires a lot of trust.

2:53:49Yonatan SompolinskyTrust either in people or institutions or, again, I'm not necessarily alluding to government enforced trust, but finance does require trust and capital allocation and you can focus on money and reject the value proposition of finance altogether, which is arguably the hidden or not sentiment in Bitcoin circles. But if you do not, if you either embrace or you do not wish to fight against the concept of finance, which is, again, it's an embodiment of information. It will be suboptimal for a system to not be able to encode my judgment, whether it's bad or whether it's a good judgment or a bad one. Let's say I believe in your podcast.

2:54:52So I want to, so I have like a private or asymmetric information that others do not have, namely that Vlad's podcast is very, is highly valuable and will continue being so. So, you see, now that you would be frustrated if this information, so to speak, that I have, or prior on the world that I have, wouldn't be no system could expressively encode it. This would be very useless, not useless because you still can trade with you can still trade with me, maybe after every episode, like transaction every minute that your podcast is running, I will send you something but more seriously, if I trust the value of your podcast, then the finance is just a word an umbrella term for more expressive contracts that are based on information or trust, but definitely asymmetric and not protocol enforced information or trust.

2:56:00Vlad CosteaYeah, I get that. And also, how are you able to finance some very basic elements from everyday life? If you don't trust the person building or doing any of the work, like there has to be some kind of risk somewhere along the system. But what I'm saying is not that it's not necessary. I'm just saying it's nice to have the option to not trust and not engage with this. And manage your money however you want, not however, like if you keep your money in a bank, you can be sure that your bank is going to do stuff with your money.

2:56:41Yonatan SompolinskyI'm not a Representative of banks in this episode. I'm not that, I'm not the opposite, the diametrically opposite of you for your views on that. I do want you to fall in love also with extracurricular material other than examples.

2:57:04He's, his views, his blogs were very influential on me as well. I love his writing and his ideas very early on. And I had the privilege of meeting him and discussing with him and learning from him. I'm jealous by the way. I don't know that he remembers me, but it wasn't a long, very long conversations, but he was definitely an embodiment of the early, the livest embodiment of early crypto values, early Bitcoin values. But you should also fall in love or at least acknowledge the literature and the existence and the literature on finance, not only on money.

2:57:48It's not that I don't, I think money is related to the infra, right? So that we have to have the money layer, the infra layer completely, not only trustless, but culturally trustless, right? We have to have the Jameson Lopps that live in the undisclosed locations. We have to have this atmosphere around the infra layer. And that's why I also wouldn't want finance and DeFi implemented in Kaspa's base layer. Above all the scalability and performance considerations, it's also just not the cultural fit. You don't want finance running on the base layer because the base layer should perfect, should secure the money properties and the sovereignty properties of the money and the payments and all the things that you fell in love with.

2:58:50And then you would want, but you still would want a vibrant and vital ecosystem that embodies, again, voluntary trust or voluntary judgment and asymmetric information, which is called finance, but maybe DeFi or finance doesn't sound enough fundamental, so I would just call it embodying asymmetric peer-to-peer information in the system, not in layer one, or at least not natively, not in any way that compromises on the basic properties of the trustless infra and the money properties, but it should extend beyond.

2:59:40Vlad CosteaI 100% agree with you. And it's very interesting because a lot of people, when they think about these L1s like Kaspa, and I guess there are others like Alephium and what was the other one, Quai, they think that they optimize for everything to be on the base layer. If they have on-chain scaling, which is reasonably good, then they assume that everything concerning DeFi or all the other applications will get added to this base layer. So it's interesting to me that you're saying there's going to be a layer two for the DeFi and Kaspa, and I think I read something about a ZK roll-up.

3:00:25Yonatan SompolinskySo I'm not saying there will be a layer two. I want to invent new terminology. Because I think the current terminology is forcing the conversation to a vision that although separate from layer one is still not something I would consider nearly optimal, and to drive it more concretely, the challenge or the threat from layer two entities is that – or maybe the threat from – it depends how you define things.

3:01:19So let me invent my own definitions for the sake of this podcast, and then if it catches, then we'll continue with these forever.

3:01:29So a roll-up is a VM that – or it's a VM structure that is optimized for attracting as much smart contracts or as much dApps into its own state. So it's a sovereign entity that's connected to L1 in the same manner you're all familiar with, but the defining, the distinction, definition of a roll-up is that it's not just one smart contract.

3:02:17It's many people – it's the roll-up entity or the team wants as many people, as many devs as possible to create their smart contracts within the sovereign state of the roll-up.

3:02:27And then this roll-up entity becomes the de facto centralized entity that has a separation from layer one. And if it were one roll-up, okay, but then what prevents someone else from implementing – deploying their own roll-up? And then you arrive at what happened with Ethereum, where you have these different roll-up entities that are playing a zero sum game, and all of them not only try to attract to healthy competition smart contracts developers – sorry, to attract smart contract devs to their own roll-up, but they also have very little incentives to interop between them.

3:03:25So many people in Ethereum try to solve, so to speak, the fragmentation problem or the interop challenge in Ethereum, but I saw this – I saw a few months ago this very nice or very articulate tweet that articulated a feeling that I had for many years, but this was articulated very nicely, that it's not a matter of technology. It's a matter of incentives. Roll-up entities do not have incentives to interop. And so while they do optimize for the separation of concerns and separation of layers, and from the pure L1 maximalist perspective, you would be satisfied with it and would be less sensitive to the problems that happen with a financial layer.

3:04:24If you look at the system as a whole, you get these several, like, I don't know, typically less than a dozen entities that fragment the entire activity in the network. And what I hoped will happen with Kaspa and what we are working on currently is designing for more Solana-like experience or broadly a monolithic – from the user's perspective including the dev, it should be a monolithic experience. Even though it's not executed by layer 1, it's still a monolithic experience against layer 1, And I will go – I will happily go into the details, but still broadly from the user, – from the dev, but mainly from the user's perspective, this is – this looks like L1, although L1 does not execute transactions, L1 does not store the state of contracts.

3:05:27L1 is implementing – is maintaining some thin skeleton commitments to the data structures or to the payload, but it does not – it's still verification-oriented, right? It's still thin in the sense that L1 does not care about any logic of transactions. It cares about metadata, but it doesn't care about any logic. You can do whatever you want there. The miners do not need to run – to execute your transactions. Kuno do not need to. It's just some metadata processing that enables this vision where you would not have these roll-up entities that try to suck as much activity and siphon the network effect of Layer 1.

3:06:17Rather, you would have a native way for people to deploy mini-ZKVMs, which we recently called vProgs, or verifiable programs, on-chain and commit them – and commit their state to L1. Again, through commitment, it's not – it's not the plain text. It's not the – there's no state bloat. It's still separating the concerns in the sense that L1 consensus is unaffected by transaction logic or by the state size, the respective state size that any mini-ZKVM decides to implement or vProg decides to implement.

3:07:02So any vProg can define – can specify and configure their own parameterization about their gas scale, the state size, regulation, parameters, and whatnot.

3:07:16And the L1 sequencer is facilitating this through some thin operations in the ZK opcode. I won't pretend it's merely a concatenation, OP_CAT. No, one trick. It is a heavy lifting for core devs. From the research and development perspective, it's a heavy lifting. But from the perspective of miners and full nodes, this is – there will be – it will be hard for it, but other than that, it will be largely oblivious to the activity. And for the ecosystem as a whole, the aspiration is that this vision will be – will embody – will look more like a monolithic, a cohesive building ecosystem and building tools rather than the roll-up zero-sum game.

3:08:16I can – I'm happy to go into the details, but I'm not sure which part you would prefer elaborating on.

3:08:26Vlad CosteaLet's put it this way. Five years from now, how do you envision that Kaspa is going to function? And how are you going to implement the smart contracts and the features that will be ZK layer 2s and everything else? And also, how is this going to look? I know you're basically a command line kind of user, but how do you envision this from the perspective of the average user? Will they know that they're using maybe a layer 2 or will it be optimized so that they don't need to know anything and the software makes the best decisions for them?

3:09:12Yonatan SompolinskyPart of your questions are outside the responsibility of core, and it should be outside the responsibility of core.

3:09:18Vlad CosteaOkay, no user experience.

3:09:20Yonatan SompolinskyNo, no. I didn't say there's no – it's like Kaspa NG. It's not the responsibility of – or maybe the inner parts of core. I refuse to dictate the system to that extent. It shouldn't be dictated by – to the explicit extent that you are alluding to. There is, again, a separation of – there is – okay, so the way you phrase the question touches on the sensitive – on a sensitive core because – it hits on a sensitive core because there is this – there's the responsibility of core to design a system that is inclusive.

3:10:17And, in particular, that doesn't dictate too much of how things will happen, right? And, again, the main concern is keeping L1 lean and thin. And, as a secondary value, to keep the system cohesive. But the extent that you elaborated the question, I feel it's slightly – at least it's outside the scope of responsibility of core, at least of me. Maybe others – maybe other people in core will go to – will have their own vision and have their – you know, and have the vision and implement it.

3:11:02But let me take it back to the more concrete language. So five years, it's forever. Let's talk five months. What I think will happen, what I want, what we are working on. So the aspiration is to develop the spec – to just, you know, release a yellow paper and specify the two layers or the canonical bridge between them. So one aspect is the opcode itself on a layer one. And this opcode needs – and we need to specify what's the responsibilities – what are the responsibilities of these opcode or this module.

3:11:51So the responsibilities are to verify transactions – sorry, to verify ZKP – to implement a verifier, a ZKP verifier in the L1 code. And to sort of facilitate – or to maintain the skeleton of the DeFi activity or the VM. Or the vProgs activity. So what would this look like in five months? I suppose it would be deep into the implementation of this. And after the implementation of this, here's what I think – here's what I envision this would look like. You will have a similar SDK or similar experience to Solana.

3:12:51You will write code in Rust. You will choose – you will choose – I suppose initially we'll need to whitelist VMs. You can't be inclusive – sorry, in the general – you can't be – for now we don't see a way to be absolutely inclusive. Like without knowing anything about your ZKVM. If not, for the MER needs to know that your program is available. That I can execute your program in case you discover. So we want the programs to be completely trustless. So you would go – as a developer, you would have a very similar experience to a Solana smart contract developer.

3:13:39So you would write a code, you would choose a ZKVM. The smart contract or the program that you are writing will be able to do cross-program invocation similar to Solana. But differently than Solana, you would – this cross-program invocation won't necessarily be to the same VM or to the same – yeah, to the same sovereign – it could be cross-sovereign states. So let's say you have your – let's say you have – you implemented some program. I know, so smart wallet ZKVM. And I implemented a Uniswap VM.

3:14:22And I deliberately combine a Uniswap, which is an app with a ZKVM. And I use the word VM to make it clear that the intention is that natively one app per VM. That's the vision. Because apps want to interop. And rollups don't want to interop. So that's why I'm calling it a Uniswap VM and Vlad's smart wallet VM. And my SDK – and when I'm writing my Uniswap, I will be able to encode this cross-program. Invocation to your smart wallet, even though you are a sovereign program.

3:15:07And this is the challenge to make sense of what I just said. Because if I – if I claim that I have a read – that I can create a read-write dependency with your smart wallet, so how does this – so why am I at the same time claiming you are sovereign and that you are – your state is independent of mine? These two seems like – seem contradictory, right? I'm able to create synchronous composability transactions with you. I'm able to create in-transaction dependency with your VM.

3:15:46And then – but I – at the same time, we claim that we maintain sovereignty. So the trick is to simply – maybe not – maybe not simply, but the trick is to – to maintain a correct regulation of your dependencies. And concretely, the dependency that I create with your program should be regulated according to your own throughput and state size parameters. And should be fully available for you to compute in case my program is not available. So your program has the guarantee – the formal guarantee.

3:16:34That you are always able to execute and prove and update the – and release your covenant in L1 at will. You don't need anyone else to do so. Of course, optimistically, I will prove my own parts of the transaction. Or, you know, the provers – my provers will prove their own part. My parts of the transaction, you'll need to prove only yours. And the thin ZKP opcode will know how to concatenate these proofs and stitch them together. And update the covenant. But that's only the optimistic case.

3:17:08You will always be able to prove and advance your own state. And you will always have the data available to prove even my part of the state in case my provers are not live. So the challenge is to formalize everything that I said now. And I think from a broader perspective, it's – it sounds complicated. It is complicated. But the fact that this part is complicated is what allows the very simple user experience. And while maintaining the – while standing behind these claims of sovereignty.

3:17:54So let me wrap it up because it was quite a long pitch to this design.

3:18:00User experience and dev experience should be easy. Especially if you are proficient with Rust. Or you're familiar with Solana smart contract. Practice, you will write in Rust. And other than – from your experience, it will be rather similar. From L1 perspective, it will be quite simple. There's some maintenance of a skeleton of the metadata transactions. So these two are simple. The core research and development will need to – are absorbing this complexity by defining these gas fee mechanisms and gas regulation mechanisms so that your vProg is fully sovereign.

3:18:52And you have – you are ensured that you have all the data needed. In the pessimistic case that everyone is live, everyone tried to screw you over, but you still can prove and advance the state of your program.

3:19:12Vlad CosteaI was going to ask you about this because obviously when people think about smart contracts in this space, they assume some sort of EVM or compatibility with that. The fact that you mentioned Rust and Solana proves that you're not following that path with Kaspa.

3:19:30Yonatan SompolinskyNo, no, no. So Solana is not a – is not a provable – is not a ZKVM. I'm talking about the – I'm talking about the experience of the developer. You would be able to – if someone implements a zkEVM in this system, okay, this will be – this could be integrated into the system easily. You would – as I said, a developer needs to choose a VM to work with. So they can choose to work with a ZKVM in the system.

3:20:11Vlad CosteaThat's beautiful. And I was thinking, what are some of the first applications that you think are going to get built? Something like Uniswap.

3:20:26I'm just trying to basically encourage developers listening to this to consider maybe using Kaspa to build their applications, or port something that's already available somewhere else.

3:20:41Yonatan SompolinskyI think porting both Uniswap and Aave is very important. And the reason I'm mentioning Aave is because, it's for two reasons. One short-term reason is because it allows the project to reach escape velocity in the sense that once you have a robust, and again, robust not only in terms of just importing Aave, but also in terms of liquidity. Once you have a robust Aave product, or lending product, then people that have accumulated Kaspa can foster the projects, can contribute to developments, either L1 or other projects, without losing exposure to Kaspa.

3:21:37So today, if you are a whale hodler of Kaspa, and you want to support development, you will contribute. You will need to sell some of your Kaspa in order to support development, either of L1, either altruistically, or for some other project that you believe can make a change. But if you have a good lending system implemented with, and again, with liquidity, with sufficient liquidity, then this allows whales and hodlers to keep their exposure to Kaspa while still funding projects. So you would do it with posting Kaspa as collateral and borrowing stablecoin.

3:22:31I think this feature is, from a systemic point of view, is the difference. It's a game changer, because now, I believe there are many such entities in Kaspa that will be able to contribute lots of more fundings to anything that happens. Of course, they exercise their own judgment, as we spoke previously, but it's a big difference whether you need to get rid of your Kaspa or Bitcoin to develop, to fund things, or you don't. You can just borrow. Again, not for free. Borrowing has its own risks.

3:23:19But anyways, this is what I don't anticipate, but I would like to see. This is the first reading of this.

3:23:29Vlad CosteaI'm basically going to provide a long explanation and then tell you how I got to this. In this space, there is a disproportionate amount of power that centralized exchanges have. So that you can imagine a situation where, for example, if during the block size wars, these exchanges decided that Bitcoin Cash is going to receive the BTC ticker, they could effectively revert the course of history. And take us to a place where Bitcoin Cash, the large block version of Bitcoin, could have been the one that people bought as Bitcoin across a majority of exchanges, and there was consensus that that one is BTC.

3:24:12And the question that stems from this is, do you believe that this idea to make DeFi possible on Kaspa is a response to the fact that there's a lot of politics involved in getting listed across exchanges and having a certain requirement for liquidity that you need to provide and so on. So that for new projects that emerge in this space, it's a lot more difficult to fulfill all of these criteria. And it's just better to work around it.

3:24:48Yonatan SompolinskyI would say that my immediate answer is that even if this weren't the case, and even if we didn't despise centralized exchanges and their ability to manipulate or exercise disproportionate authority, even if this weren't the case, I would still prefer DeFi over these centralized entities. And the reason is because you have this atomicity. It's very important to have the ability for people to build contracts to encode programs that are atomically composable.

3:25:40And even if centralized exchanges were righteous and played the game very nicely, which isn't the case, but nonetheless, we still wouldn't recover that atomicity.

3:25:51We would still have the same experience as in traditional finance, where you can't create atomic transactions. And while atomic transactions seem like a niche feature, it really embodies the essence that the other party, the counterparty, can halt and can decide. Other contract developers either need permission or their users have a risk of their transactions not flowing. So decentralized finance or permissionless finance is just superior due to its atomicity. Even centralized exchanges were very well behaved.

3:26:46Vlad CosteaIt seems like the DAG visualizer sort of crashed. It might be from my browser. Pretty sure that the Kaspa network is still mining blocks and we didn't run into a crisis situation on live video.

3:27:02Yonatan SompolinskyIt would be funny if it did happen.

3:27:06Vlad CosteaYeah, I tried to refresh the page and still I don't get anything here. Oh, it's back. OK, no idea what happened there.

3:27:23Yonatan SompolinskyWell, you know, it's actually interesting because I think I can constantly think about this, what can happen when we spoke about trust. Let's say the protocol is trustless. Let's say you read the code base, you read the academic paper, you read the code base, you're convinced it works. You connected your DNS. Everything in the infra is trustless. But then there's only one block explorer and the vast majority of people don't know how to use CLI or command line, you know, querying the full node. Would you consider it trustless?

3:28:06It's easy and similar to your example with the BTC ticker scenario, it's easy for the block explorer to dismantle the network or to cause chaos and essentially take down the network. And it will be very hard for people to trust the network if they think that it crashed.

3:28:32So I think this realization, this is this is in a in a micro level. This was what is what happened with RBF, with RBF bug or without potentially a lack of RBF and CASBUB prior to September. Again, I think 23. The network itself was operational, the mining, the mining continued, but it was very hard to know what's happening if you're only using block explorer. So I the reason I think about it constantly because I realized and I I real I increasingly realize it in more broader and increasingly broader senses that it's not enough to have just the base layer trustless.

3:29:24You have to make any effort you can to have all layers trustless and avoid single points of failure. And this this does go back to the question why I'm slightly allergic to roll ups, because let's say someone develops a successful roll up entity gains the entire traction and is is assumed by everyone to be Kaspa, right? So assumed to be Kaspa L2, but has but but it's a separate layer. Then again, it can exercise control and you don't and this is not hypothetical. Hypothetical. You you know that the amount of activity that Base had in DeFi Ethereum.

3:30:13It's it's it's it had it's had lots of control. I don't know if it's more than 50 percent, but I think it's not far from that, the DeFi activity in Base versus Ethereum.

3:30:30Maybe I'm exaggerating, but I don't I don't think to a large extent. So so you have a central entity. And then what happens if if Coinbase decides to to censor many, many users or or to increase simply increase the increase the fees? So the fact that increase the fee, I mean, in a way that will feel to you that you went back to traditional finance. So what will you do then? So it's not it's not enough to to speak of the decentralization of your of your base layer.

3:31:04And I think this is something that I learned in the years that passed. I was of the view that it is enough to have the infrastructure trustless and anything else can be done in a more carefree manner. And now I'm more I'm more cautious or paranoid about other other layers being co-opted or centralized.

3:31:38Vlad CosteaYeah, and I believe that. So I'm going to say this now because we have almost 10,000 viewers on the live stream. It's useful for everyone who cares about the project and wants to contribute to also run this sort of infrastructure. Like there are other ways to do a block explorer. You shouldn't wait for the Kaspa team to build something. You can basically use open source libraries to generate some sort of graphics from the node that you're operating. And put that on a Web page and people are going to be able to visualize what's happening in the block explorer.

3:32:15I see a lot of expectation from people that the team that fundraised for the project should build all the infrastructure. But if you really care about the future of any project, you should make sure that you're decentralizing the project and the way in which you do it is to contribute. If you can't personally contribute or you can't vibe code something. Just look up to community members who are able to do it and just remove this single point of failure, which can be like the quote unquote official block explorer or wallets or whatever.

3:32:56Just seek to integrate into as many already available tools as possible. Maybe lobby to Blockchair to also add Kaspa. If it's not there already, I'm not sure. Blockchair or whatever. The more of these tools that you have and the more nodes are being scrutinized to verify the data, the better. Of course, you can always run your own node and verify all the data against it. But it's not the nicest experience from an interface point of view. You have to type some commands and you have to know exactly what you're looking for.

3:33:41Yonatan SompolinskyEven if you run a full node, even an SPV node, the root of all evil, even that would be very beneficial for a crisis where no one knows what's happening and you won't know what's happening. Maybe the two block explorers will experience some bug or some DDoS and you will be able to both know what's happening and to prove it to everyone up to the SPV assumption. I do want to just correct the impression that I'm not sure you have, but maybe others do have that there's like an organized team.

3:34:23And I'm not saying this in order to read off the material, the marketing material, but really these versus these visualizers and these wallets. I know some of the people, some of the people I know, they're pseudonyms. I'm very detached from, I have no context on this. I'm sure Michael has much more context on them, but even he's not responsible for them. He didn't build them. These things are built by developers and they have their own, you know, they get the credit and the blame. And don't assume this is something their team, the core is very, so there's a few layers of core, but definitely not the team in the organized sense that at least I don't know many of these people in person.

3:35:26And I don't know who built what tool at all. I don't know the wallets. I don't know the block explorers. I use them. I see the block explorer I use, but also the CLI. And anyways, it's important to just from an expectation perspective, it's not even about decentralization, although it is implementing decentralization. I'm also mentioning it for mentioning it for the mere reason of aligning expectation. There are different people with different contributions and don't assume that there's an organized team or that I know what the organized team is or that I know the set of good tools that are implemented on Kaspa.

3:36:15Do your own thing, build your own things. Don't assume I know about them and I don't need to, or Michael doesn't need to and other core members don't necessarily need to. And definitely, yeah, the more people that know how to operate natively the full node, or at least an SPV node, the more resilient the network is, especially in times of chaos. And in times of chaos, it's too late for you to download a tutorial to know what's happening. You need to know, you need to prepare in advance.

3:36:49And I think especially when we implement DAGKnight will be the embodiment of chaos preparation of crypto. Like DAGKnight is a protocol that will be, I think of it as World War III resilient, like with a cyber war and everything, the internet is experiencing real chaos, then I believe Kaspa and DAGKnight can withstand this, but not if you're using Kaspa through block explorers. You need to use it much more natively in order to, you know, you need to have direct access to the inner workings to the backend, of course, if you can, if you have the technical knowledge.

3:37:39Vlad CosteaRight, so what is DAGKnight? I guess we alluded to the term a few times. It's, if I understand correctly, it's the next big upgrade in Kaspa.

3:37:56Yonatan SompolinskySo whether it's bigger than the vProgs, the ZKP opcode or not depends if you prefer the money or the finance use case. But let's not belabor that. So we hope this will be implemented together as a hard fork, but these are indeed two separate things. DAGKnight is the final evolution of the protocol. So if you think of, so let's start with Bitcoin, let's go back to Bitcoin protocol and call it in the more academic name Nakamoto consensus. So Nakamoto consensus is the longest chain rule. You have a slowly growing tree where usually it's a chain, one chain in the tree, and you always pick the longest chain and you discard any other branches.

3:38:51The GHOSTDAG generalization of this dictates the following structure. You have a graph, a DAG of blocks growing in one direction. And in order to make sense of this graph and to recover consistency and consensus, you order it according, roughly according to the heaviest K cluster. Of course, I'm oversimplifying it, but the intuition is, I think, much simpler than people think it is.

3:39:23I argue that if you, Vlad, look at this, maybe not this block DAG visualizer, but a block DAG that I will draw to you with your naked eyes will be able to identify what is the honest group of blocks.

3:39:43And this is what enabled us to come up with a protocol. It's not so much abstract as looking at the DAG and the whiteboard and seeing many DAG manipulations and saying, wait, but in any of these manipulations, I can identify the set of honest blocks. And I argue anyone can do so provided they are given a sufficient number of examples. But for you to be able to do so, you will immediately ask me, hey, how do I know whether this is an attacker or not? You will see the DAG, the graph of blocks, and it will say, hey, I only see a mess of things.

3:40:30How am I supposed to know if this block is an attacker block or not? And I will tell you, well, I'm giving you another information. I'm giving you information that the latency in the network is upper bounded by something, and the implication of this, I'm giving you the implied information is that honest miners, wherever they are hidden in the graph, are able to create blocks that are not too disconnected. So, if I tell you that honest blocks, the delay in the network is such that honest blocks are only one or two blocks away from the main chain, and other than that, they are compounded on one another, then you will look for not a chain, but a 2-3 heavy, or 2-3 thick chain, and you'll see, hey, I see here in the middle, a 2-3 blocks thick chain, and it's always connected.

3:41:31I'm connecting at most 2-3 rounds to the back, and so I know that this is the 2-3 chain. This represents, with high probability, the honest chain, similar to the practice that you would do when you look at forks in Bitcoin. You would say, hey, I know that this one chain is probably the honest chain. This one is too short, right? So, you do the same trick. You look at, you assume 51% honest, and you say, okay, the chances, the probability that this 2-3 heavy chain, 2-3 thick chain, 2-3 block thick chain that is heavier than other candidates, the chances that it is not the honest chain is exponentially low as I wait a few more rounds.

3:42:20So, you embody the same idea. You just use your eyes. Of course, it translates to a protocol, but the intuition is you just look at the mess that you see, and if someone tells you that honest blocks are created with a 2-3 DAG latency, then you'll know how to pick with your naked eye the honest chain with high probability. And the GHOSTDAG protocol simply or more convolutedly implements this idea with many complications that I won't get into here. But this is still, with all the benefits it brings to the table, it's still a synchronous protocol.

3:43:06It still assumes an upper bound on a latency, and it still fails to be secure if this upper bound is compromised. So, if for some reason, I don't know, an electromagnetic pulse, the internet breaks down for, or at least there's a large disconnect, and now blocks are propagating with a much larger K or much larger latency than you will have anticipated, then the protocol ceases to be secure similarly to Bitcoin, which ceases to be secure under the same circumstance. So, Bitcoin and currently, Kaspa are synchronous in the sense that their security relies on an upper bound on the latency.

3:43:48The DAGKnight protocol, which I co-authored with Michael, is a result of the question that I posed to Michael, which is, how can we get rid of this parameter K and allow the network to be secure regardless? So, I don't want to have this need to rely on the dev's good estimation of K. Rather, I want a protocol to be secure, whatever K is, whatever the parameter is. So, formally, it's called partially synchronous, but I won't get into the debate whether it's legitimate to call it partial synchronous because distributed systems guys hate this name.

3:44:33I meant to say they hate the claim that DAGKnight achieves it. But for the sake of this podcast let's call it partially synchronous. And the idea that Michael came up with is. So we started with an idea that Aviv came up with many years before that which is to recursively look at that block that is most disconnected from the graph and trim it, and then iteratively trim the just take from the graph the most outlier, the most prominent outlier, iteratively until you see a K-connected graph.

3:45:12And then this protocol is only secure, we showed it secure only up to 33% Byzantine faults, and we can improve it to 40%. But then the subsequent idea that Michael came up with was very, very neat. And it is, instead of deciding in advance what K is and choosing an upper bound on the latency, choose, okay, so let me state it clearly, instead of maximizing the K cluster, instead of choosing the heaviest K chain, so Bitcoin maximizes the one chain, right? Bitcoin chooses the maximum length of one block in chain, right? That's the Bitcoin protocol. Are you still with me?

3:46:06Vlad CosteaYeah, sorry, I get distracted by the chat, but I'm still trying to listen.

3:46:12Yonatan SompolinskyOkay, so Bitcoin, let's start again. So Bitcoin maximizes the weight or the, it chooses the heaviest one chain. GHOSTDAG chooses the heaviest K chain or K cluster, let's call it K chain, KY chain.

3:46:34DAGKnight, before choosing the maximum K chain or K cluster, it chooses the minimal K for which the maximal K cluster is at least 50%. So let me repeat that. If Bitcoin chooses the maximum one cluster or one chain and GHOSTDAG chooses the maximum K cluster, DAGKnight before, DAGKnight does not know what K is, right? We want to get rid of that assumption. So before maximizing the K cluster, DAGKnight searches, DAGKnight solves a minimax problem of finding the minimal K for which the maximum K cluster is heavy, is capturing the majority of the graph, is heavier than 51%.

3:47:22And to convince yourself that it is sound, you should simply look at the case where there's no latency in the network at all or negligible latency, like in the Bitcoin case. Both protocols will converge with Bitcoin, coincide with the same protocol, up to edge cases that I won't get into. They implement the same intuition of choosing the heaviest chain, but they just generalize it to the case where there's meaningful latency or high block rate, the K chain protocol that's GHOSTDAG. And then DAGKnight generalizes it by, instead of choosing the K cluster, first searching for the minimal K for which this maximum K cluster is higher than 51%.

3:48:08And with this, I understand that for many listeners, this sounds overly convoluted, but if you write it as an optimization problem, this is less than 30 words. It's a very, very clean presentation. Bitcoin maximizes the one chain, GHOSTDAG maximizes the K chain, DAGKnight finds the minimal K for which the maximization problem of K reaches 51% or 50%. It's a very clear optimization of a paradigm upgrade. Now, the implication of this, even if you didn't understand anything, even if the listener understood anything or nothing of this, is that you no longer need to assume that the latency in the network and the internet is less than something predetermined.

3:49:10And so if the internet is very smooth, the confirmation times of transactions are very smooth, very fast. It can be like very few seconds. The same order of magnitude of a stake coin. It's O of RTT, network RTT, round trip time. And if the internet is very slow because there's some systemic collapse and lots of network hiccups, then the network will still be secure. And it will be the only network that's secure against 50% attacks because it's the only protocol that's secure against up to 50% attacks without assuming any latency.

3:50:00So this is not just a nice feature. If you want to prepare this monetary system to a global all-out war or some other catastrophe like electromagnetic pulse where the internet really suffers from real breakdowns, then I don't see other protocols that will survive this other than DAGKnight.

3:50:36Vlad CosteaAre there any known trade-offs to this DAGKnight protocol or any assumptions that can potentially break?

3:50:48Like, I'm trying to be Greg Maxwell but I don't have the math knowledge or the cryptography knowledge to do this, so I'm...

3:50:54Yonatan SompolinskyNo, Greg Maxwell will approve. Greg will also approve of GHOSTDAG being only an improvement over Bitcoin. I will get to the implementation trade-off shortly, but from a from a conceptual framework perspective, GHOSTDAG itself, even if in the sequential setup, is a pure generalization of Bitcoin. There's no trade-off whatsoever.

3:51:27And then also DAGKnight is the generalization of GHOSTDAG, which is in turn again the generalization of Nakamoto consensus. So I don't see any trade-off in the underlying principles. Of course, if you want to use this and we use it in the live system, then the trade-off is that you need to prune data in order to support high throughput. And then you have this... And while the pruning is completely kosher and it's actually using the kosherizing blocks in the protocol, it still creates complexities. And you mentioned again Gregory, there's a very nice, again, something that influenced me or my thinking on this system, a very nice excerpt from Gregory early on where he discusses trust.

3:52:29And he says something along the lines of, not that trust is unmerited, it's frequently merited, but the problem is that trust-based systems are more brittle. And I think if Gregory saw the pruning protocol, he would say this works, but this creates a more brittle system. And I prefer a system where nothing is pruned and I can run the plain ledger from beginning to end. And I don't want to use SPV proofs, although I understand the cryptography and I agree it's unbreakable, but I still argue that such complexities are not healthy for the system.

3:53:15And this is a trade-off. I don't think we should dismiss this sentiment. Of course, creating, introducing complexity into the system is a trade-off from the user experience, from the builder's experience shouldn't suffer from this, but of course, for people in core and for people that are helping audit the protocols and then make sure nothing breaks, this is of course a greater challenge. And I completely understand people that will reject this, at least I understand the sentiment of being paranoid about introducing complexities into a monetary system.

3:54:03But if you, again, it goes back to the question, what are you willing to give away in order for this to be usable and useful? So we're giving away our tranquility and we're willing to be more, to work harder to make the protocols, to audit the protocols and make sure we're doing things correctly. But I think you see this visualizer justifies it because it recovers the aesthetic. It gives you a reward for working, for implementing something complex because it does, to me, this doesn't look very different than Bitcoin. It looks like an aesthetic enhancement of this system. To me, this induces trust. For me, this visualizer, what do you think?

3:55:05Vlad CosteaI think it's hypnotizing in some way. It looks like a living organism of sorts, which is always interesting. I used to spend a lot of time looking at the Bitcoin mempool, which is mempool.space. It's a very nice visualizer and it just shows every 10 minutes when blocks get mined and you can look inside to see what's there. This one is a bit too fast for me because I would want to just stop and zoom in to see what's happening there beyond this visualization. And of course I can do that in another section of this.

3:55:41It just makes me wonder how I can just randomly click on a block and see what kind of transactions are inside. Let me see if I can catch one. So it does show the number if I, or the hash, I'm not sure exactly what this is. I also want to ask you, it seems like there is a green line that transcends this sine wave that's being displayed here. Is the green line basically an indicator of consensus or the longest something?

3:56:23Yonatan SompolinskyYeah, it's the longest something. There's also a graph inspector, which you might find more to your taste. The URL is kgi.kaspad.net.

3:56:48Vlad CosteaOh, this is the one. Oh, so I can click on blocks and I see the transactions inside. I like this.

3:56:56Yonatan SompolinskyYeah, kudos to the developers. I'm sorry, I don't remember the names and I don't know them or either don't know or don't remember who created these things. But shout out, yeah, both of these visualizers are pleasing. One is pleasing to me, one is pleasing to Vlad.

3:57:16Vlad CosteaI mean, this one feels more functional. You can even see some arrows that point to connections between, I guess these are transactions.

3:57:26Yonatan SompolinskyNo, these are blocks, these are block references. It's like Bitcoin, I told you, they're just block references.

3:57:36Vlad CosteaThis is so cool, to be honest.

3:57:40Yonatan SompolinskyI think he wanted to go live again to refresh or something.

3:57:42Vlad CosteaOh, I need to. Yeah, there's a pause button and you can basically look at the current state of the DAG. It's not, can you call this a blockchain? Because it's not the traditional blockchain design, it's a DAG.

3:58:05Yonatan SompolinskyI think it depends on the context. There's some context where the word blockchain would be fitting because you're describing, because you're describing, it depends on the context. If you're comparing this to some, I don't know, regular BFT, yes, it's a blockchain. If you want to have, in that context, you would call it a blockchain. But in the context of proof-of-work protocols, you would call it a blockDAG. I think the suiting term depends on the context of the listener or what they're trying to get. I wouldn't say it's not a blockchain if I'm speaking to someone in a loose, broad context.

3:59:00Vlad CosteaI was going to ask you, what's the bottleneck for Kaspa? Is it bandwidth and internet speed?

3:59:07Yonatan SompolinskyIf it's okay, I want to just answer your question about these lines, these blue and previously green line that you saw. Yeah, so most of the arrows are regular arrows. But, so I said I'm oversimplifying GHOSTDAG. The actual implementation or the actual proven protocol is a greedy algorithm implementation of the heaviest K chain or K cluster. It's easier both to prove and to maintain a code base where the ordering of blocks is dictated by a selected chain. This chain is by no means the longest one. It's just the center, so to speak, of the K heaviest one.

4:00:04You could call it the heaviest K chain, and that's what you're seeing. And a similar concept would appear in DAGKnight, but since DAGKnight has no K, so we would just call the mega chain or the DAGKnight chain, whatever. So that's on that. The bottleneck for Kaspa, I mean it's hard to say because there's no there isn't yet activity. So are you referring to the conceptual bottleneck? Or the actual bottleneck now?

4:00:39Vlad CosteaWell, the bottleneck now and also the one conceptually that you're going to encounter in the future.

4:00:48Yonatan SompolinskyTheoretically, so theoretically when you scale up the block rate, then the DAG in principle can become quadratic, right? The information in the DAG becomes quadratic because every round you have all blocks, in theory I'm not talking about the actual DAG, but all blocks point to other predecessors and you have this redundant but still intense computation to do when you're increasing the block DAG, right? So the width of the DAG represents the latency of the network. So if the DAG is 10 blocks wide, you know that there are roughly 10 blocks created in every network RTT.

4:01:43And if you scale the network, right, because blocks that are created afterwards are already pointing, so they will accumulate and compound to that direction. They won't be created in parallel. Right, so the anti-chain or the DAG width represents how many blocks are created in parallel. Which you can think is just proportional to the block rate multiplied by the latency. So block rate multiplied by latency equals DAG width under no attack. Under attack, it's one minus alpha. It's like the honest ones are creating what I said and then the malicious node can do whatever it wants.

4:02:20And we have a messy DAG, but still in it there's a sub DAG that has the property that I mentioned. And this is identified by the protocol. So if you increase the block rate, at a certain point you reach a state where the computation of the DAG, the DAG has too much information. The DAG's complexity has too much information and then you need to, and then you hit a bottleneck. I think around currently around 32 or 42 blocks per second. Now we're running on 10 blocks per second, but I think around 32 or 42 blocks per second.

4:03:00We hit the bottleneck of even without transactions, just the block computation becomes too heavy. And this imposes a problem on me because I mentioned that I'm building 100 blocks per second blockchain or block DAG. But Michael showed me that on the current Rust code base, which is still a marvel of an achievement, even just to achieve this, even to reach a 32, 42 blocks per second. This is a marvel of an achievement. It's not something that anyone can achieve, even if they know, even if they understand the protocol, this is not something that can go without mention.

4:03:48And of course, Michael was not alone in that. There was a full team of developers. Again, I don't know many of them. So, but back to the point. So the nice, the funny story about the 100 blocks per second aspiration is that it really stemmed, it started with, so there are two people who steered me into doing this. One is Tom from DFINITY, who sort of bragged about DFINITY being able to process 100 blocks per second. Of course they can, because it's proof of stake. I was slightly pissed by it.

4:04:33So I said, okay, so we're going to do it with 100 blocks per second proof of work. So we're going to, we're going to give the word of practice. And this was at the height of the myth that proof of work can't scale. So I sort of want DAGKnight Kaspa to be the fastest among all chains, not only proof of work chains. I just want it to be the fastest consensus protocol. And it can do so because proof of work has this local property that every miner can progress the state alone.

4:05:11So there's no leader, there's no, it's not leader-based consensus. Anyone can progress the state locally, just mining through a memoryless process. It has this very neat dynamic, which translates to its ability in DAGKnight, not in previous protocols, to outpace other consensus. Let's hope we can reverse the myth and then people will say, okay, proof of stake can't reach these levels of speed. And maybe sometimes in the future, someone would be able to match the pace of proof of work with DAGKnight. But until then, we would reverse the myth.

4:05:52But another reason, or not too serious reason, but I was asking Elichai, which is a Bitcoin core dev, so he worked at DAG Labs, but he was still a Bitcoiner, a true Bitcoiner, Elichai Turkel. And I said, you know, you're not impressed with anything that's happening here. We're building a block DAG. So he said, yeah, you know, he was more impressed with, he was still impressed with Bitcoin only. And I kind of pushed him to the corner and asked, what would impress you? What kind of achievement would impress you?

4:06:32And he said, if you're doing 100 blocks per second, then that, in a secure manner, that would impress. And I decided to impress him. So that's the, so going full circle, that's what we need to achieve. But on a more serious note, we're hitting the current bottleneck is hitting around 30 or 40 blocks per second. But this is because you are adding to, you're adding redundant information in the DAG. So consider like, let's say there's a clean structure where every two seconds internet round trip completes.

4:07:10And in this, in this duration, 10 blocks, we're creating an anti-chain in parallel. And then two seconds pass, and again, 10 blocks per second, and then again, per two seconds, and then 10 blocks again. Let's say this is a structure just for the sake of conversation. So you can see that it will be 100 edges, 100 blocks pointers every round, right? Each of the 10 block blocks mined in this round will point to each of the 10 blocks mined in the previous round.

4:07:38This causes the DAG to have quadratic complexity and is one reason why you hit a bottleneck. And you would, but you can see that this is, embodies, this represents redundant data or information on the DAG, because you don't really need all 10 blocks to tell you about all 10 blocks. It's simpler, it's great, it's natural, but it creates redundant data. And if you want to scale it to 100 blocks per second, then you'll have to give up on the simplicity of the mining rule, which states that you point at all predecessor blocks that you see, and you'll need to create some, to replace it with some probabilistic rule that will choose K out of N or L out of K in this case, block predecessors to point at so that the block still has the same level information, but you don't have that many edges or block DAG data to process and reason about.

4:08:45And what's the benefit of 100 blocks per second other than impressing Elichai is, it's just, it implies that if you have 1% hash rate, you can mine one block per second. So I do envision that with these, even without 100 blocks per second, even if you scale the system further, even today with 10 blocks per second, it means that someone that has, that an entity that has, let's say 5% of the hash rate can mine a block every two seconds. So let's say your user's attention span or patience window is two seconds, then it's enough for you to have 5% of the hash rate in order to mine your own user transactions.

4:09:28And the more you scale the block rate and increase the block rate and stretch the system, you lower the bar for mining your own user transactions. And this is, I can't spell out exactly what use cases this opens up, but I'm sure this is a big leap towards a significant leap towards like the sovereignty of the system because people can You don't have this, all these, you'll. Still, you don't have this layering of you need to operate your transaction through third -party miners, which limits your ability to negotiate your position in the block space.

4:10:16Consider a situation where you want to secure the block space for your own use cases. Let's say you want to run a STARK VM on top of, on top of Kaspa and STARK proofs are large. So you don't, you don't want to need to negotiate your STARK proofs always on, you know, with transaction fee market. You prefer having the, you know, the autonomy to publish your own, your own STARK proofs in large blocks, you would be able to do so if you have a miner or to guarantee yourself to not compete in the, in the transaction fee market.

4:10:50If you're, if you have a miner. So it's just one example of what I think could happen if entities can run their own miners. I'm not saying this is necessary. I'm just saying it's interesting. What will the dynamics that it enables are interesting. But, but yeah, I'm not, I'm not, I'm not portraying this as maybe the top, the top. It's not a top three priority.

4:11:18Vlad CosteaThere's a question in the chat, which I guess I told you before we started, but people picked that up. It's from Dr. K, who is one of the creators of a network that's called Quai. And that project does something similar to what Kaspa is doing, except that they also have sharding. And his question is, wouldn't Kaspa have faster finalization if you maximize the graph anti-chain?

4:11:47Yonatan SompolinskyYeah. So I suppose the question is if you increase the block width in our, in the context of our conversation, no, you will have as long as the round does not complete. So you need to confirm transaction. You need a one round to complete. And if you maximize the anti-chain at least in GHOSTDAG, then you can't really confirm transaction, at least for meaningful sizes of assumed attackers before the round completes. Now there is a, there's a point about confirmation times that I want to mention in general.

4:12:32It's not related to this question, but in general, right in Bitcoin, you need to assume the size of the attacker. When you set your client, you say, okay, I have a tolerance of Epsilon and I'm assuming attacker of size Alpha. Let's say 35%. And, and in the nice thing about DAGs is that if the attacker does not attack you explicitly, the attacker can double spend you only if they can revert your transaction only if they double spend you explicitly. Right? So if I'm a merchant in DAG and Kaspa, and I don't, in my, in my own threat model, I don't have the threat of the, of the counterparty double spending with more than 10% of the hash rate or 5% of the hash.

4:13:18But then I can set these parameters very low, but in Bitcoin, I need to assume the size of a global attack on the network because an attacker can revert the entire chain regardless of their interaction or lack of interaction with me. So merchants and transaction recipients in DAG, it's broadly most reasonable DAGs have the property that you can set the size of attacker in your local client to a low value, according to your own threat model. Whereas in Bitcoin, because an attacker can revert the entire chain in one shot, then you don't have this property.

4:13:57So you need to set Alpha, the attacker size to your assumed, to the, to the assumed attacker size on Bitcoin, not on your, on your own transaction. Is that clear? Or, or I should repeat this in more concrete ways.

4:14:16Vlad CosteaI'm not sure there's someone called Dr. West in the chat who has been asking this question for the past four hours, and I finally had the right context to ask it, right? Because it's about how you can optimize the DAG to have faster finality for transactions.

4:14:36Yonatan SompolinskySo again, so the intuition of anyone that's thinking about DAGs should be, as long as we're still creating transactions in blocks in parallel, sorry, as long as we're creating blocks in parallel, there isn't meaningful security above the one confirmation security. Okay. It's not, it's not meaningful. I, we can, I can optimize it and I can, I can get into the nuances of this, you know, if someone wants the most tight confirmation policy, then we can work towards that. But, but the broad intuition should be that as long as the round, the, as long as honest blocks are still created in parallel, so in internet round trip time hasn't completed, you don't get meaningful security above one confirmation.

4:15:25And then, but then after it completed the confirmation, the security increases exponentially fast. So the, so the, so the raw, so the plain answer to this question, to the previous question, at least you mentioned is that as long as you create more, more and more blocks in parallel, you will, you won't be able to confirm transaction. And then this is actually not a property that you need to Kaspa or to, or to go with DAGlabs or DAGKnight, it's a property of any consensus network.

4:16:04There's, there's theoretical limits on the, on the speed of confirmation that you can't, you can't go be below the network's latency. This is not, this is not even a property of either Kaspa or any, no consensus can break these down.

4:16:26Vlad CosteaRight. I was going to ask you if, for example, there's going to be a node on the network that is very slow or slow in terms of internet connection. Is it possible for a majority of miners or nodes that are slow to slow down the network or are they going to be left behind? And if they are left behind, isn't this a bit centralizing for people around the world who don't have access to very fast internet?

4:16:54Yonatan SompolinskyThat's probably a nice, a nice pushback of Gregory. The different from a researcher and definition, from a research definition perspective, we define, and again, this is not unique to GHOSTDAG or Kaspa, any consensus system, any synchronous consensus system. You define the honest nodes according to your latency bounds. So if you assume an upper latency bound of two and a half minutes, you know, roughly as in the case of Litecoin, then any mining node that suffers a latency in that order of magnitude is considered faulty in the, in the consensus jargon, whether it's Byzantine.

4:17:46So the Byzantine adjective is more about the behavior or the intentions of this malicious of this node, but it's called faulty, which is a subset of Byzantine in case it's not connected sufficiently connected to the network. So any consensus network, if you assume a synchronous network, you assume that you define any node that is slower than assume latency as part of the Byzantine faction. Now, of course, if you, if you assume that everyone is connected within a hundred milliseconds, then this is a centralized force, but note that creating 10 blocks per second doesn't, doesn't imply that we are assuming that you can, you can create a hundred blocks per second.

4:18:35And it's still assumed that the network latency is three seconds or one minute if you want the trade-off the trade-off here is not. So if you want to be safe and you want to say, I want to be, I want to be open to, I want to be inclusive to miners that are behind firewalls or slow regions. And I assume that they're, and I want to be inclusive to them. And I define the latency to be not five seconds, but five minutes.

4:19:07Okay. Then the implication of this in GHOSTDAG would be that the parameter, the K parameter that you will define in the network would be order of magnitude larger. And the implication of this in turn is that you will suffer a much longer confirmation times because you will need to wait O of K meaning you'll need to wait at least the assume latency multiplied by the block rate. And if you assume latency is five minutes rather than five seconds, then the confirmation times explode.

4:19:45So this is the trade-off the more you are inclusive to late to low latency miners, the more you are, the more you are, the more you are postponing the confirmation times of transactions. This is not a trade-off of Kaspa. This is a trade-off of any synchronous consensus protocol, any consensus protocol, which embodies a hard-coded latency bound. For instance, Kaspa, for instance Ethereum, for instance Bitcoin suffers from the same phenomenon, of course, with different safety margins, according to the, to the principles and practicality of, you know, trading off the principles and what's practical.

4:20:30DAGKnight does not suffer from that again, because DAGKnight doesn't require any assumption on K. So, so there isn't, there isn't, this is, this is just.

4:20:36There isn't any need to assume these things. And you see, this is a very natural advantage of DAGKnight over these protocols. There's no need to assume or to explicitly make the mining inclusive only to a subset of miners. Anyone can mine. Of course, the more, the more, the slower your internet is, the less competitive you'll be with transaction fees. You can't, you can't run away from that. Users will want to be mined by fat miners. But anyway, so I think, I think this is, this goes back to why DAG, why, when I think about Kaspa and DAG and DAGs, I really think of this, the DAG that we are seeing as a precursor to DAGKnight Kaspa.

4:21:35Vlad CosteaYeah. And on the topic of mining, I think it was a year ago or a couple of years ago that the transition was made completely to ASICs, meaning that graphics cards are no longer profitable. Do you think that this is the desired outcome for mining to become professionalized? And if so, how many manufacturers of ASICs are out there and how easy is it for the average user to start mining?

4:22:08Yonatan SompolinskyFor the average user who purchases ASICs?

4:22:11Vlad CosteaYeah. How easy is it to get ASICs in the first place? Because I know in Bitcoin, it's quite a challenge to get your hands on the latest ASIC from Bitmain.

4:22:27Yonatan SompolinskyI don't think I know the answers to the two part of these questions. I think that I presume Kaspa didn't achieve yet the competitive scale of Bitcoin, where if you mine with a, with a lower, with slightly lower, slightly less efficient ASIC, you are pushed out of the network. This goes back again to the stability and instability of the price. So I assume in a Kaspa, it's still easier to mine with less efficient ASICs. I'm not an expert on this. You did ask another question that I did want to add. Yeah. So, so I do think this is desirable.

4:23:19Yeah. I do think ASICs are desirable. You should, you can, you can either have a ASICs or a, or a very good flywheel mechanism that will ensure you a very, a very high hash rate. You can't, you saw what happened with Monero. They were the dream of every ASIC hater. And it's easy to rally. I don't know why it took so many years to do this, to pull off such attacks, but, or such transparent attacks, I would say. But it's easy to, for non ASIC friendly networks to be under attacks, that happened multiple times in crypto, with NiceHash facilitating GPU rental for attacks.

4:24:16Yeah. Then definitely I see ASICs, I don't know if the desired outcome more, more so it's, it's, it's definitely much more secure than GPU friendly.

4:24:29Vlad CosteaNow I was alluding to that because last week, Monero had this situation where someone conducted the selfish mining attack. And by someone, I mean the Qubic people who basically incentivize the miners to join their mining pool and promise to them that they're going to get much higher rewards because of their token or something like that. It was interesting to see that play out. It was not a 51% attack as some of us assumed. It turned out to not reach that threshold, but it was 30 something, which was enough for them to do some, I think there were two reorgs and there was some selfish mining involved.

4:25:18Which once again is interesting and it begs the question whether or not it's worthy to chase this dream of not supporting ASICs and changing the algorithm every time ASICs are being invented. The advantage of ASICs is that you can't really print them out of thin air or you can't spin up more of them than the ones that were produced. It's a very specialized production process. There are only a few of them, and I mean thousands of them that can get produced per year. And this means that from the perspective of any kind of attacker, it's difficult to get the means to try to attack.

4:26:00But it can also be centralizing in the sense that you're going to have this elite group of users who can mine with the latest machines and the rest of them will have to buy pre-owned devices or whatever as it became in Bitcoin.

4:26:23Are you aware of how many manufacturers of ASICs are working on Kaspa miners right now? No. But is there more than one? I know that there's Bitmain, but I'm not aware of any other.

4:26:37Yonatan SompolinskyYeah, I think there's Bitmain and IceRiver at least, but I don't know any details on this. I'm sure many people in the community do know details on this. I just didn't read about it or read and forgot, but I don't know much about this. I would say that the reason I'm hesitant to answer is because crypto doesn't have a better solution than ASICs. And innovating here, so I'm coming from a research perspective and motivation, and the ability to innovate here, innovating in this layer, really the hardcore rails of the system, you should have a very good idea, you should have a very, very good reason.

4:27:46You need to have a very good reason in order to propose a change of that layer of the proof of work dynamics. I'm not saying Kaspa will never undergo modifications there, but there needs to be a very, very good reason and a very large super consensus in the network, in the community for such a change. And for this reason, thinking about how to improve what has been for now the state of the art security in crypto, how to further improve it, is not occupying me too much.

4:28:39I did come across this proofs of useful work of a brilliant team that invented, I think it's the first invention in proof of work that I've seen in recent years. They invented a very cool mathematical trick to recover the same properties that you'd expect from proof of work, while allowing the algorithm to do any matrix multiplication of your will. So in contrast to any, so there are many, many proposals and there have been for a decade at least proposals for useful proof of work.

4:29:25But I think almost all or all of these proposals defined in advance, defined in advance what useful means.

4:29:36And these protocols and these proposals were not, were either not useful or not proof of work. And this is the only work that I've seen that is both useful and proof of work. And the way they do so is by leveraging them. So they're building on the fact that matrix multiplication has a canonical way to be executed. And there's a consensus in the hardware industry that it can't be improved, at least not in the generic case.

4:30:22So they build on the gap, on this gap to create, they build on the gap between doing this algorithm and they introduce, okay, so they introduce artificial noise into the matrix of multiplication in a way that you can easily show to someone that you committed to the right transaction, you multiply the matrices correctly, and the verification costs very little for that person, and you chose the original matrices yourself.

4:31:07So consider you are choosing two matrices, A and B, and you need them. Their multiplication is useful for you. Why? AI training, AI inference. So you chose it. So then the protocol adds artificial noise to these matrices. You show to the network the noisy version of their multiplication, but you can easily, in your own backend, remove the noise and get your own original multiplication without the noise. So this allows you to train data or do inference in AI without any, while securing the network.

4:31:54And this paper is the first, the only innovation that I think is ever worth considering in the context of your question.

4:32:04So going back to if I would ever think of proposing to the community a change from the ASIC path, it won't be in the near future. But if in the long, if in the long term future, I would be the one proposing anything, then it would be something in that direction. Why? Because the ASIC, the fact that it's not ASIC, it's GPU friendly, is weighed, is weighed, is weighed off by the fact that you have enormous market to secure your network. So it's not feasible to attack the network.

4:32:45So again, there's two ways to secure your network, either by being ASIC friendly or by having a very, very good flywheel of GPU friendly, which is arguably what happened with Ethereum and Ethereum 1.0. No one thought Ethereum 1.0 was not secure. So it has a gigantic, enormous hash rate, and it was considered secure, even though it's not ASIC friendly.

4:33:08Vlad CosteaI remember the ASICs came later, but maybe one year before they went proof of stake. I also want to pick your brain on the issue of proof of work versus proof of stake. I want to know your personal opinion on the differences.

4:33:26Yonatan SompolinskyWow, this is an old entire topic.

4:33:29Vlad CosteaIt is. My oversimplification is that proof of work tends to be the more honest form of money, where there is an expenditure of energy that justifies the price of the currency. And also, you cannot say that you created that out of thin air. And in the case of Bitcoin, for example, I think that proof of work is the one element of the whole equation that makes it not look like a Ponzi. Like Satoshi, this unknown person just created a monetary system. If it wasn't for the fair launch and the fact that you can mine your own coins and basically be your own Ponzi organizer, I think that's the wrong way to put it.

4:34:16But in this metaphor, it's what makes sense. Like it's not a Ponzi anymore. In the case of proof of stake, if you have one entity that issues all the coins as a pre-mine and maybe allows them to get discovered by validators, there is a huge gap between early adopters and the ones who will basically buy the coin above the $1 price. And it's unfair. And it also incentivizes early adopters who get a lot of coins to perpetually stake and earn the largest chunk of rewards. And there is no real cost to that except for maybe just keeping a computer running.

4:35:09Yonatan SompolinskyThere's no continuous cost to pushing you out of your current relative position. If you have 20% of the network, there's nothing compelling you to invest more in order to keep that. I guess that's what you're describing.

4:35:23Vlad CosteaMaybe the price action can get you to sell and someone else is going to pick that up. But it's still very oligarchic in this way. And I think proof of work is more dynamic. And proof of work, if you don't keep up and you're not competitive, then you're going to basically become obsolete. You constantly need to find cheaper energy, more efficient devices to mine. And once you cannot find this anymore, you're not going to be profitable and someone else maybe will figure out a way. And you can argue that also in proof of work this can become centralizing because you need a lot of capital to be able to have the most efficient form of energy and also the best machines.

4:36:13But it's dynamic. If you look at proof of work and Bitcoin over time, there are so many phases and eras of this happening. When it first started with CPUs and then GPUs and Satoshi sent an email in 2010 to Laszlo and told him, well, slow down with your GPUs because you're taking too many rewards. And Laszlo decided to do that moment when he bought two pizzas for 10,000 Bitcoin as a way of giving some of the coins back to the community. And then there are the FPGAs and then the first ASICs by Butterfly Labs.

4:36:52And it just kept growing. And then you had a bunch of people discovering that there's a lot of electricity that's affordable in China. And that's also where the ASICs are being manufactured. So a huge concentration of mining happened there until 2021, when basically the Chinese government said they're going to kick out the miners. And now I guess more than 50 percent of the mining is in the US. The European Union still hates proof of work for some reason, so there's not much happening there. But even in countries neighboring China, like Taiwan and Singapore, there's a lot of activity happening right now with mining.

4:37:38And many of the Chinese miners moved in, even in Hong Kong, which we can debate all day whether or not it's China. But there is still some activity in there. And the point of this whole story is to say that proof of work is dynamic. And there are events that happen over time which break any monopoly or perception of dominance. And in the case of proof of stake, most of the actors tend to be exchanges, especially if they are investors, as it happens in the case of Ethereum where Coinbase and Binance are some of the biggest stakers.

4:38:20I don't see any way or any scenario where this is going to change unless the users are going to decide to migrate out of these large centralized exchanges.

4:38:35Yonatan SompolinskyI don't feel I can add much to this great description. It's an interesting piece of data. I didn't know more than 50 percent of mining is in America. Is it North America?

4:38:50Vlad CosteaYeah, there's Foundry and there is also, let me look, there is Luxor, a bunch of others. If you add them up, I think. Let me take a look right now. I don't want to speak wrongly. So Foundry has 31.83 percent. Luxor has 4 percent, almost. MARA Pool also 4 percent, so 8 plus 31. This is 39, almost 40 percent. And there's also SECPOOL, I think is Chinese. Ocean has 1 percent. I don't know how you're going to label this one. SpiderPool, I think is Chinese.

4:39:38F2Pool, Chinese. You have all of these. So AntPool 17 percent, ViaBTC 11. I think the Chinese ones still have more than 40 something percent if you add up. But still, it's interesting. There's also 3 percent of unknown hashrate. It's a lot more balanced than it used to be. This is where I'm alluding.

4:40:12Yonatan SompolinskyI will add, though, that there's another parallel set of arguments for proof of work that is unrelated to what you mentioned. I think historically there were two camps, there were two camps within the proof of work camp advocating for, advocating for work for two different reasons. The first one is what you mentioned. But the other one, so the first one, it's the camp that's treating proof of work as the superior or the only legitimate way to print new money, which I completely agree with.

4:40:58And the second camp emphasizes the superiority of proof of work in terms of consensus.

4:41:09So even if, you know, even if emission winded down, I'd still promote proof of work or argue that proof of work have properties that are unachievable in proof of stake. And the reason is because proof of, or the mechanism is because proof of work have, introduces into consensus an external clock, an external verifiable clock, the proof of work latency. And this really simplifies the mechanism, the consensus mechanism. And there are many properties that proof of work can achieve due to this, due to this external source of randomness or external clock that proof of stake can inherently cannot.

4:42:03To name one thing that I'm, to exemplify this through one thing that I'm working on, one research I'm working on. With another Aviv, from today's postdoc at Yale, is MEV auctions. I know it's kind of a different topic, but consider you want to introduce a transaction to the network that, and you want to, and you want to auction the right to sequence it between the miners, between the leaders of consensus. Let's say, let's say we're thinking with entire, all the consensus possible consensus configurations. Then in order to even to think about such an auction, so you can do this with such auctions, with sham auctions, with sort of tricks and Ethereum kind of proposer-builder separation, which I never proposed a builder separation, which I don't think can be justified.

4:43:10But if you think about the entire, the fundamental challenge here, so you want to have the, you want to send your transaction to the network and auction the right to sequence it between all the participants of consensus. Now, for this to happen, you can't have a leader, you can't have a leader based consensus, because if you have a leader based consensus, then that leader can choose whether to, whether to mine your transaction or not. So from, so if you think, if you see this DAG, this DAG has the property that every round, there's no monopoly on sequencing.

4:43:53So if you want to take a transaction and auction it across the miners, you can, you can design such an auction. And you can't do this with proof of stake because in proof of stake, again, I know there's lots of MEV auctions in proof of stake, but they're all hiding, they're all using some centralized intermediary or other tricks. But if you want a clean auction or a clean way to, a real way to auction the right to sequence your transactions, then you can use it, you can implement such a thing only in proof of work.

4:44:29This is just an example of the property that proof of work has, that there's competition on sequencing in every round. I guess the broader point is also, or the broader exemplification of this is that, let's say someone wants to censor a transaction. Okay, so there are many protocols, there are many different protocols in the consensus landscape, but almost all of them that you know of are leader based. So Bitcoin is leader based, right? The next block is mined by one miner.

4:45:17Of course anyone can compete and try to be that miner, but eventually you know that if you mined successfully that block, you're the monopolistic entity on that block.

4:45:28So you can censor transactions or otherwise financially manipulate them, what's known as MEV. Now of course, across multiple Bitcoin blocks, you get the censorship resistant guarantee because one honest miner is sufficient to introduce you into the block. But with this system that we see, you have a property that immediately, you have this leaderless consensus, immediately in every round you have censorship resistance. I want my transaction immediately to be processed by someone, assuming an honest majority or even one honest entity, assuming even one honest entity, mine is a block in that round, I entered in that.

4:46:17And this kind of guarantee, whether you use it for censorship resistance or whether you think about it in terms of censorship resistance or use it for MEV auctions is only possible in proof of work, at least without using other tricks or in centralized intermediary.

4:46:36Vlad CosteaThat's interesting. I never considered that. There's also someone in the chat who says, so his name is AFL underscore AFL. Can the externality of block production and proof of work be replicated in a weak subjectivity way in proof of stake to create the same mechanisms of pruning?

4:46:59Yonatan SompolinskyI need you to repeat that. I understood the words, not the sentence.

4:47:03Vlad CosteaSo can the externality of block production and proof of work be replicated in a weak subjectivity way in proof of stake to create the same mechanisms of pruning?

4:47:22Yonatan SompolinskyThat's quite convoluted. The pruning creates the weak subjectivity. It's not created by it. Once you have pruning or finality, then you already have weak subjectivity. So I guess the question is whether you can, I guess I don't understand the question. It's too convoluted because there's multiple unrelated concepts and I don't understand how to tie them together. The externality of block production. So implementing, mimicking externality of, mimicking proof of work with proof of stake is something that has been proposed. I think Aviv Zohar even published a paper on this, if I'm not mistaken.

4:48:04Many people tried, or at least several projects have tried to come up with ideas.

4:48:14So the idea would be like, you want to mimic an ASIC, an ASIC purchase. So you mimic, you stake a very large amount and this amount allows you to, and then you burn every block, you burn coins. I think this is what like a subset of this transaction, of this question is alluding to. There are many interesting proposals. There was a researcher named Sarah, I don't know how to pronounce her name. She created a cool protocol called Fantome, which is, she said is inspired by Phantom, by Phantom GHOSTDAG, which is a DAG version of, which is a proof of stake version of GHOSTDAG, where it's called betting and consensus.

4:49:08It's actually a very neat idea, but I don't, it still suffers from proof of stake problems, but in and of itself, it's worth mentioning. It's the idea that you, instead of burning electricity or other resources in order to produce a block, you bet, you put money inside the block that is burnt in case your block is not in the main chain or the main K cluster. I remember hearing her lecture about it and I found it a nice proof of stake version of GHOSTDAG, even if we, if we don't have Phantom GHOSTDAG, even if we don't really want, even if we don't, just ignoring the problems of proof of stake, it's a nice research idea. But other than that, I don't think I answered the question.

4:50:04Vlad CosteaWell, I'm not sure exactly what the person was trying to find out because it's a very long question. There's also Ace Footage in the chat and he says, what was the original intent behind Kaspa's rapid emissions?

4:50:22Yonatan SompolinskyI don't think Kaspa was overly rapid. If you compare it, I think whenever it was much more rapid, if I'm not mistaken. I don't think it was rapid. It was, it was, there's a trade off, right? You need exactly to avoid the problem of Grin and Grin was the last fair launch that I witnessed before Kaspa. And what happened with Grin sort of deterred both me and other early adopters from considering flat minting curves where early adopters have no incentive. So at that time, I thought that if I want to err, it's better to err on the opposite direction than Grin due to their failure.

4:51:14So, so that's what, that's the context of what I was wary of when designing the minting curve. And again, the minting curve can be, it's not, it's not the only parameter, right? In Monero, there was, it was hard for normies to mine Monero, right? To, to answer this. It was only you had to be part of the in group, at least in the beginning, in order to join mining, not because of a protocol barrier, just a practical barrier. And they didn't really, as far as I recall, they weren't very, they didn't go out of their skin to help people mine.

4:51:58And this is what I, this is what I alluded or was referring to in the beginning of this conversation, when I said there are many ways to either cheat or behave in a way that you don't want, behave in a way that compromises on a fair launch, even without, even without explicit manipulations. So, so I guess the bottom line is, I was wary of Grin's flat curve, and there had to be some incentive for early adopters. And you can watch the hash rate and the Discord history to understand that it was very, very successfully decentralized from the outset.

4:52:45Of course, I would personally wish that I and others in the inner circle would prepare better for the explosion of hash rate, because we, I really didn't prepare anything in advance. You can actually look in the Discord a week before the launch. Elichai and I and Ori were still debating what proof-of-work function to use. We're still consulting with the community. So there was, this was really raw, fair launch, for the better or for the worse. And so I don't think the rapid emission hurt Kaspa in the sense of centralizing the holding.

4:53:32I think, I think the hash rate was very, very diverse, diversely spread. Of course, it's hard to have verifiable data on this, but from what I remember and from what I, from the feeling I got early on and from the current dynamics in the community, it seems to me very fairly distributed. And I don't, I don't think it's, I don't think it hurt Kaspa in any way to the contrary.

4:54:12Vlad CosteaJust for the record, this is very spontaneous. Like I'm just reading questions from people in the chat. I appreciate the fact that you're answering them. I guess this also dispels any critique that we might receive after this one, that maybe something was rehearsed or there was some planning. There was none of that. Like our interactions were minimal. We just set the date and the time of this interview. I don't think you told me anything about what we should discuss and it's very free flow. Then there's also Joe2305 in the chat who says, you think that improving the frequency of hash rate sampling makes Kaspa less susceptible to selfish mining attacks or other sub 51% attacks?

4:55:04Also, do you think hash rate sampling that uses the anti-chain independent samples can more accurately measure hash rate and Fisher information?

4:55:18Yonatan SompolinskyI didn't understand the first part of the second question, but to answer the first question, so DAG has the nice property that selfish mining is much less, like order of magnitude, much less profitable. The reason is because selfish mining is an awkward attack strategy because let's take Bitcoin. In the first two weeks of Bitcoin, I mean in the first two weeks of attack selfish mining on Bitcoin, you would not benefit anything. You would simply sabotage others and you yourself will not benefit anything. So how does selfish mining work?

4:56:12You withhold your blocks, your successfully mined blocks and selectively publish them in order to outrun, outpace others and sabotage them. So selfish mining is really a sabotage manipulation. You're not gaining anything directly from selfish mining. The benefit from selfish mining, at least according to the original model, is that after the difficulty adjustment rate, this sabotage implies that if you look backwards on the window and you see how much rewards you received versus others, then you increase your own ratio in the rewards and difficulty is supposed to normalize it back to the regular model where you increased your benefits.

4:57:06So this explanation, the reason I detoured to explain selfish mining on Bitcoin is because if you understand that sabotaging other blocks is the core primitive of selfish mining attacks, sabotaging them by orphaning them out from the chain. So if you understand that this is the only thing, this is the only profitable primitive in selfish mining, then you immediately understand that in order of magnitude more difficult, because by selectively publishing two blocks, when I published one, you can't push me away from the DAG. I'm still part of the DAG.

4:57:51You may be earlier to me, from me, than me to the chain, but I'm still part of the DAG. So pushing me, pushing a miner, a sabotaging a miner is much more difficult in DAG. And that's why selfish mining is much less profitable, exponentially less possible. When I said exponentially, I mean in K, like the K parameter controls, if you are able to push a miner outside the K window, then you have sabotaged their block. But then you have to do a reorg or you have to mine K plus one blocks faster than the network.

4:58:34This is exponentially unlikely. So that's the good news. The technically accurate bad news is that such manipulations still require lower hash rate than in a chain structure. So here's the good news and the bad news. The good news again is that profits from selfish mining in DAG are exponentially low. And the bad news is that you need a much slower hash rate in order to carry out these unprofitable attacks or manipulations. So this is the story with selfish mining in DAGs. There is, I describe now something that is related to block rewards only.

4:59:21There is of course the competition on chain blocks, which is, you know, and chain blocks do get additional rewards. If not for the mere fact that they mine transactions, you know, case of duplicate transaction fees, they receive the fees. So the model that I described in my previous answer is slightly contaminated by the presence of transaction fees that do provide preference to chain blocks. But that's, yeah, that will require a longer treatment, but I don't, but I think that the core, the core dynamic is around the block rewards and I hope I described the implications for DAG.

5:00:11Vlad CosteaYeah, so you said you did not understand the first part of the second question, which was, do you think hash rate sampling that uses the anti-chain can more accurately measure hash rate and Fisher information?

5:00:26Yonatan SompolinskyBut what's the context for hash rate sampling? If the context is difficulty when the difficulty adjustment algorithm, then we already use the, we already use all data. We use that entire block DAG. Even, even red blocks to make, to adjust the, to sample and adjust the difficulty. So it's already, it's already baked in. And by the way, even in Bitcoin, if Bitcoin protocol would sample as we do off-chain blocks, then selfish mining in Bitcoin wouldn't happen as well, because by sabotaging other, others, you're not increasing your profit even after two weeks.

5:01:09So, so I think that's the question assumes that we're not sampling the hash rate from the anti-chain, but this is not correct. And if alternatively it's about the proof of work being dependent on an anti-chain blocks, then no, that wouldn't be a good idea. But I think, yeah, I think the, I think the question assumed difficulty is only based on the chain.

5:01:38Vlad CosteaJust for the record, I love this. We have been going for five hours. I'm still getting questions, but I feel like the quality of question is getting so much better as some people are dropping out. Like it's no longer about when does Kaspa get listed on this exchange or one moon or whatever, it's just some very technical high signal stuff, which I appreciate. And I'm going to ask you the next question, but in the meantime, I'm going to ask people who are still around to put their Kaspa addresses in the comments.

5:02:11And I'm going to redistribute what I received earlier in the stream to people who are still around, which I think is fair because you guys are heroes for sticking around. Yeah, I feel like these are some of the best questions that can possibly be asked. There's someone named, so put your addresses in the comments section. There's someone named Conviction. And his question is, you said kHeavyHash is oPoW friendly. I guess O stands for optical. Will Kaspa adopt kLightHash long-term to optimize the capabilities of optical miners?

5:02:56Yonatan SompolinskyKLightHash? I have no idea what kLightHash is. So the story with, the story with kHeavyHash. So yeah, I'm not sure if I should describe the detail of the story about kHeavyHash or just admit that I don't know what kLightHash is. What's the second part? What's the actual question? Will Kaspa adopt what?

5:03:26Vlad CosteaAdopt kLightHash long-term to optimize the capabilities of optical miners.

5:03:32Yonatan SompolinskyI don't know what kLightHash is, but I will answer nonetheless. The current, so the current protocol is already friendly for optical miners, but the optical chips industry is much slower than we anticipated, and therefore it's currently mined by digital miners.

5:03:53It doesn't seem to me realistic to expect in the near future that optical miners would mine with kHeavyHash or kLightHash or whatever that is. But that's, but Kaspa doesn't need to do, but for kHeavyHash they can just do it without informing anyone, and if kLightHash is worth looking into then educate us on what that is.

5:04:18Vlad CosteaOkay, I just sent a hundred KAS to Burke who put his address there. Next up is Kaspa Silver, and Kaspa Silver actually has a question here. He says if it hasn't been discussed yet, ask about the missing transactions that exist in Kaspa. This is the biggest FUD that Kaspa gets.

5:04:54Yonatan SompolinskySo again, all transactions, almost all transactions are missing. All the spent TXO transactions are not kept in Kaspa, only the UTXO. So some operators of block explorers have generously decided to keep that data voluntarily, or most of that data, and apparently they did so only for the vast majority of Kaspa's history and not for the entire history, and that's the context for which, that's the context for this, what you call the missing transaction, but there's no missing UTXO transaction, and if some archival nodes decided not to keep all data, it's their prerogative.

5:05:46I would mention again that if a block explorer does not show someone's UTXO transaction, this doesn't mean the UTXO transaction doesn't exist in the UTXO state, just run your full node with UTXO index and find your UTXO. There's no UTXOs missing and there's no systemic guarantee, there's no consensus guarantee or systemic guarantee that anyone will keep the spent transactions, the STXO. So from my perspective, the vast majority of transactions are missing, are deliberately missing.

5:06:36Vlad CosteaOkay, let me get to the next question, which if I recall was, okay, there is kang9393 who says their proposals to upgrade were rejected because, oh no, no, not this one. Wait, this is actually an answer.

5:06:54Ace footage says props to Yonatan for staying around. Why is it not an issue for each validator node not to hold full transaction data since Genesis? And the question is about pruned versus full node.

5:07:14Yonatan SompolinskyAll full nodes and data in Kaspa are pruned.

5:07:22Vlad CosteaYour crypto crew says if this wasn't asked yet, can Yonatan disclose why his and Aviv's proposals for scaling Bitcoin in the early to mid 2010s were rejected or ignored?

5:07:36Yonatan SompolinskyNo one seriously considered proposing to Bitcoin core devs, updating Bitcoin protocol back then. It was the heat of the big block debates and no one really, no one, it wasn't the right, but the question doesn't begin. That's the true depiction. The fact that Bitcoin core were aware of these ideas doesn't mean this entered their top priority. And I guess rightly so in those days, again, the big block debates were the main thing that occupied everyone's mind in terms of scaling Bitcoin's current crypto in general.

5:08:40And initially I was just going to be in the position of learning, not in the position of proposing changes. And also, by the way, by the time that GHOST protocol was cited in Ethereum white paper, we already knew that it doesn't achieve what we thought it achieved. We already were not deep, but in the journey of DAG protocol, DAG protocols. So we were kind of surprised that GHOST appeared in a white paper.

5:09:14Without consulting us for the mere fact that we already had insights about its properties. But it takes years to prove, even if Bitcoin Core were more attentive, it took me years, for instance, to prove SPECTRE protocol. The protocol was invented like maybe two and a half years before I completed its proof.

5:09:47So the timescales are just different. Both the timescales were different than people imagine. And the priorities of Bitcoin Core were different at the time. And I guess for good reasons.

5:10:05Vlad CosteaIt's funny how different your answer is from the conspiracy theories that are being posted in some of the comments.

5:10:19Anyway, Joe Apple says, what about quantum resistance? Do you think it's worth integrating early?

5:10:28Early, and in a parenthesis, he says via PQC slash ECC replacements. So PQC is post quantum cryptography. ECC is elliptic curve cryptography replacements. Or does it add more risk than it solves right now?

5:10:51Yonatan SompolinskyI don't think for Kaspa it adds a lot of risk because we take the more flexible approach towards innovation and upgrades and hard forks. But I don't know the quantum cryptography in any expert level, but the people that do know don't seem to take it too seriously in the short term. And in the longer term, again, Kaspa is deliberately less conservative than Bitcoin regarding updates.

5:11:40I don't think it will be risky, such upgrades will be risky to Kaspa. Admittedly, the aspiration is to start decreasing the hard fork frequency as time develops.

5:11:55We're not in the other extreme of making difficulty bombs inside the protocol every six months like Ethereum did.

5:12:17In general, I don't see these things as, these things shouldn't take years in Kaspa like it takes in Bitcoin. So as more information gets out and is verified by people that do understand post quantum cryptography, Kaspa will have more time than say Bitcoin to upgrade and less time say than Ethereum to upgrade. I think this is where we're standing on the upgrade hard fork policy, which is not set in stone, but I think we are in the range of like closer to Bitcoin, but not very close.

5:13:01Vlad CosteaBitcoin recipe in the chat wants to know if you wish that you would have stayed anonymous.

5:13:09Yonatan SompolinskySo this goes back to, this is actually tied to the previous question, to my previous answer, which is about innovation. So personally, of course, but it wouldn't be genuine as long as it's declared a platform that embraces innovation. Like as long as we still have, and I guess, I mean, I'm talking about myself, have innovation, as long as there's innovation that I'm heavily involved with, involved in, to perfect Kaspa's sequencing properties and consensus properties, then being anonymous would be less genuine, unless the point is pseudonymous, but then since pseudonymous has the same, almost the same, if not more control.

5:14:16For me personally, of course, I would prefer to be anonymous and I aspire to not, I aspire in the long term to not be around. I'm actually waiting for this more than some would suspect. I don't like the burden of being around and, you know, I want to have the freedom to research and innovate in other areas, you know, and go back to my more tranquil lifestyle. So from my personal perspective, definitely this was not a good move, but from the protocol perspective, while it's a definitely negative for Kaspa that I'm not anonymous, currently, it's not a net negative because I'm still involved, heavily involved in innovation.

5:15:15And as long as, and at the moment, this will no longer be the case. I'll be the first, the first to run out of the window. No one, I'll need no help in that.

5:15:31Vlad CosteaYeah, this one comes from me, actually, and is a follow up. You envision that there will be a moment when Kaspa is complete and there is nothing else that you want to add to it?

5:15:42Yonatan SompolinskyYes. For me, there's DAGKnight, there's DAGKnight and the vProg design. And then afterwards, I want the main thing I was working, the main research I was working at the postdoc at Harvard and then currently as a research associate in Harvard is on a new design of, a new design of smart contracts or maybe a new flow for VM and smart contract transactions. Sometimes it sounds, I'm not sure how to describe it yet in the best way, but I see this as a very meaningful upgrade to the way transactions are sequenced in crypto.

5:16:40And once, but the paper hasn't been released yet and the community hasn't, I mean, it hasn't been peer reviewed yet, neither by the academic community nor by Kaspa community. And if this proposal is accepted, then there will be like, it will keep me, it will be another meaningful upgrade. I guess less to L1 will be less affected by it, but the entire ecosystem flow, but it can be more major or less major. I don't have enough prospect on that yet.

5:17:27I mean, I don't know if it's a huge upgrade or a minor upgrade with a large, very large potential.

5:17:34I think it's more the latter, like it's a moderate upgrade. It doesn't affect L1 almost at all, but it does affect the flow of L2. And if this proposal, if this research is solidified and the proposal is accepted, then this does, this does create another big, another heavy lifting project for us. But for my own personal aspirations, both for me and for the project, I view that as completing or perfecting the sequence. The sequencer is then perfect and then, or at least sufficiently perfect. And for me, it's complete.

5:18:26It's not to say that Kaspa will prohibit, will forbid hard forks, but it will become more rare and require more justifications. And I guess I will aspire to be much less involved, if not at all in the move.

5:18:46Vlad CosteaThat's part of decentralizing the project. There is this quote from Satoshi from 2010 under the thread where ZK proofs were being discussed. And he said that, well, if this becomes possible and someone figures out how to do this, a much simpler implementation, simpler and more efficient implementation of Bitcoin can be written. And I want to ask you, do you think that there is some sort of cutting edge technology around or that can emerge that can make a much simpler and more efficient and overall better implementation of Kaspa to be written?

5:19:27Or is it state of the art and it's going to stay state of the art for many years?

5:19:31Yonatan SompolinskyIt's still hard for me to hear the word state of the art with regard to proof of work. There's so little, there's just a small camp in crypto that would be accepting for this phrase, right? For mentioning state of the art next to proof of work.

5:20:18Whether there is cutting edge technology, no, there isn't. There isn't anything in horizon. I mean, you see what Satoshi knew about ZKP. So it wasn't, you're asking me to imagine something that I never heard of. Whether, and I can't think of any.

5:20:46Layer where you would want to innovate. The only thing that I would that I found a cutting edge technology that improves proof of work and in extension can potentially improve Kaspa is what I mentioned about the matrix multiplication protocol which I'm not yet advocating so much as I'm reflecting that this is one of the only instances of something that I just saw that is like wow this is an upgrade, a cutting edge upgrade of a cutting edge technology that can can materially upgrade the concepts that that Kaspa is running on.

5:21:31So I guess, I guess we found one that's the proof of work function. But this is not for the, this is not for the short or medium or this is not something that I'm thinking of in practical terms. This is something that I'm mentioning. It begins and ends and mentioning it. If someday it will become, or if someone will, I think this team is trying to launch a proofs of useful work platform. If this experiment will be successful and it will be easy to adapt to DAG, then this, I think it will be worth bringing it into the table and discussing it with the community.

5:22:23I'm not sure I will be very, very involved in Kaspa at that time. Like, ideally, no, but I hope, nonetheless, I hope this will be, this or similar tech will be incorporated.

5:22:42Vlad CosteaI've also asked Vitalik the same question in the context of Ethereum. If you could travel back in time and relaunch Kaspa once again, based on the learnings that you have acquired over the years, what would you do different?

5:22:57Yonatan SompolinskyI don't think there's anything I wouldn't do different. I would do everything differently. There isn't one thing that I did correctly. I think some things happened to turn out good for Kaspa, but the whole process and my whole way of thinking about it would be, so I don't think about the launch itself, but maybe like years before I will just launch.

5:23:43I would just go about the R&D in a much more spontaneous, unorganized manner, rather than the burden of trying to CEO a company and failing and then abandoning and launching.

5:24:00Nonetheless, I wouldn't go through this route. I think Kaspa could have launched earlier, much earlier if I had chosen this path, but within this path, I would do everything differently. I can't single out one decision that I think I would have reversed more than others.

5:24:32Vlad CosteaThat's interesting. Let me get back to questions from the chat for a second. Even if some of them feel redundant, I feel like they're formulated in a different way. Like Kurt Shacks says, Can you explain why proof of stake is undesirable? This is another common fud. Why proof of work?

5:24:58Yonatan SompolinskySo refer to Vlad's answer. He gave a very long answer. Again, there are two reasons. One is the right way to print money. And the other is that proof of work enjoys properties that cannot be achieved with proof of stake. Namely, the use of external source of randomness and external clock for consensus that enable you to create very fast leaderless consensus and to so you're able to create very fast. So you achieve leaderless consensus in each round. So you get in each round, you get, I guess it's more about fast proof of work rather than proof of work, but it's enabled by proof of work,

5:25:43Meaning it's not possible with proof of stake. You get immediate censorship resistance. You get the ability to auction your transaction sequencing rights, which is very relevant, very desirable for DeFi transactions. There's a myriad of properties that are only achievable in proof of work, and in proof of stake, you suffer from from other than the decentralization, the centralization dynamics that Vlad elaborated on. You just also suffer from the regular limitations of just BFT, Byzantine fault tolerance, traditional consensus protocols. For instance, DAGKnight, which achieves 49% Byzantine tolerance, Byzantine fault tolerance.

5:26:32And without assuming any latency bound has been shown to be impossible to achieve, to be achieved in BFT. So in the proof of stake or BFT model, you can't achieve what DAGKnight achieves. It's a very strong and clear exemplification of what you can achieve when you have an external source of time or external clock baked into consensus.

5:27:14Vlad CosteaI hope that he's happy with the answer. I'm pretty sure that even though I said some of that earlier, it's going to be sound bites of what you said that will get cut down into clips and posted on Twitter later on. So it holds more value if you say it. There's also Altcoin Bonanza, who says, What did Adam Back say about your efforts to upgrade Bitcoin? Did you ever get any feedback from Adam Back?

5:27:47Yonatan SompolinskyI never made efforts to upgrade Bitcoin. The fact that we spoke about DAGs with Bitcoin Core Devs wasn't, again, wasn't in the context of, we weren't in this, it wasn't the context of now Bitcoin should adapt it. Not because we didn't want to, but because it was very clear that Bitcoin wasn't going to adopt consensus changes to this level when SegWit was that heatedly debated. It's just delusional to think that anyone would consider a different chain selection protocol when there was such a clash around, a year long, a years long clash around the block size.

5:28:46For us to explicitly propose our protocols to Bitcoin would be completely out of touch with culture and priorities set by Bitcoin Core, whether they have set the correct bar or not. It's a discussion for another day, but given the conservative, the extremely conservative bar that they have set, no one ever would have considered, seriously considered proposing incorporating any such thing in Bitcoin. It's just that it wasn't the right atmosphere. It still isn't. I think, by the way, that the only scenario where Bitcoin would consider adopting the DAG protocols is if after DAGKnight is implemented, some, God forbid, catastrophe happens and then DAGKnight is the only one to remain alive.

5:29:53And then I think this would be appealing to Bitcoin Core circles. If you can imagine even a satellite powered network that is able to just be resilient to internet hiccups and to just an all out cyber war. And if DAGKnight is the only one that's remaining operational in this context, I think such an event can lead Bitcoiners to look differently on DAGs. Until then, I don't think it's interesting to them.

5:30:46Vlad CosteaRight, so can you elaborate on why you think DAGKnight is going to be the only protocol that survives cyber war or catastrophic events? What is it about DAGKnight that makes it so resilient?

5:31:03Yonatan SompolinskyThe fact that it has no security parameter baked into it, so that even if the internet suffers serious breakdowns. Again, if the entire world is completely disconnected and we're back to the stone age, then it's not interesting, of course. But in the middle phase, where we're not back to the stone age yet, but the internet is still somewhat operational in the sense that you can communicate. You can communicate, but not with the assumed latency of other networks and you can't assume any good bound on the latency.

5:31:50Then all the synchronous protocol will break and Kaspa is the only 50%, like 49% Byzantine fault tolerance protocol that does not assume a bound. So let's say the internet suffers an on and off state where the network has a latency of 10 seconds and then 10 minutes. Then 10 seconds and then again 20 minutes. Then 10 seconds and 10 minutes. Anything in that range, okay?

5:32:27I'm mentioning this range because this is still a range where civilization is intact and you're interested in money.

5:32:40In this scenario, Bitcoin and Ethereum and any synchronous consensus would fail, would break. Just the tree of Bitcoin would fragment. Of course, it depends on geographical centralization. If 51% is in the USA and there's no fragmentation therein, then there's only two factions in the network. But even two factions in the network means it's unclear which one wins and then an attack where if there's three factions in this global war, then one third can override.

5:33:27It's enough for one third to override the two factions. So the security threshold of Bitcoin deteriorates very fast if the internet delay starts going below five, six, seven, eight minutes.

5:33:45And for other proof of stake networks and for Kaspa today, which is also synchronous network, it deteriorates very fast after 10, 20 seconds latency. It's already breaking their latency assumptions. But with DAGKnight, you have no latency assumption. Of course, confirmation times would be much longer, assuming you're interested in making sure your transactions are stored and there's no backwards attack. You can't reverse history. You're willing to wait for confirmation times, but you just want to know that your previous transaction and previous UTXOs are safe. Then DAGKnight or any hypothetical protocol that partially synchronous protocol that assumes no latency bound and that is resilient to 49% attacks will be the only one that survive.

5:34:45Vlad CosteaInteresting. There's also Abu Jia, I hope I pronounced that correctly, who says, Do you like the idea of stablecoins on the Kaspa network?

5:34:58Yonatan SompolinskyWhether I like it or not, the fact is that stablecoins is one of the main, if not the top, product-market fit of crypto. So currently adoption of crypto is through stablecoins. We could have dreamt of different paths in the past or envisioned different future, but for now, it is what it is. People are using crypto and it's definitely improving both their lives and minimizing trust, not minimizing, but decreasing the need to trust local institutions. I would say that among stablecoins, of course, I would prefer algorithmic ones over centrally issued ones.

5:35:55I think this goes back to the protocol that I'm trying to develop that has the ability to formally condition the flow of transactions in smart contracts. The connection is that algorithmic stablecoins need a very smooth way to adjust collateral and prevent liquidation or act upon liquidation.

5:36:41So algorithmic stablecoins, of course, are preferable from a decentralization perspective, much more preferable. But I wouldn't think even to oppose the main product market that crypto has found, even if it's not what we envisioned in the early years.

5:37:12Vlad CosteaThere's a remark from your crypto crew who says, wow, these questions up the quality of this session again. Shout out to Yonatan and Bitcoin Takeover for being here for over five hours. And there's Altcoin Bonanza, who says, what was the vision behind the verse in the Old Testament Ezra 7:18 for the Kaspa Genesis block? Oh, wow.

5:38:01Yonatan SompolinskySo first of all, the easy answer is that it's a verse where the word Kaspa in Aramaic appears in the Old Testament. That's the easy, that's the immediate answer.

5:38:19The context of this verse is the declaration of Cyrus the Great, the emancipations of the Israelis, but not only Israelis. He was a great ruler and he was sort of very, I think, pluralistic or one of a noble ruler in the world. And he was considered, was praised in the Old Testament and the Persian Cyrus the Great. So I guess that's an aspiration for I think it's an aspiration for a world where which which many people, many from across the aisle can identify with.

5:39:14Vlad CosteaWhat does across the aisle mean in this context? Across the political aisle. Oh, that's a good aspiration.

5:39:29Cape says, without archival nodes, does that make Kaspa a privacy chain? I don't see any issue with this, since there will always be someone who wants to view and maintain majority of transaction history. But it's an interesting theoretical point to make while on this topic.

5:39:48Yonatan SompolinskyHire him for marketing.

5:39:55Vlad CosteaBecause it's unrealistic, right?

5:39:57Yonatan SompolinskyNo, it's a great talking point. Let's take the lemon and turn it into a lemonade.

5:40:08It's like we spoke about Mimblewimble, right? The model is always in privacy. If you want to achieve real privacy, it can't be based on an assumption that there's no one listening to all transactions and recording them. Especially in the case of Kaspa, that you are broadcasting them all the time. So no, you can't make me say this talking point. But all power to you for making this. For these creative proposals.

5:40:56Vlad CosteaMan from Bogota says, are you Satoshi? Honestly, I don't think Satoshi left Bitcoin and just disappeared. And you're the only one obsessed with Nakamoto consensus.

5:41:14Yonatan SompolinskySo I think we addressed this, but I'm reflecting on this question a lot. How can people, how to, how to, just like a generational gap. People that are early in Bitcoin, in crypto, would never even fathom such ideas. I think the people that are in this third wave or fourth wave of crypto, so much has happened. And the information in crypto is so unclear and the context is everything.

5:42:00So people that came, I think in the third or fourth wave, just don't have any, for none of their fault, they don't have enough context on what happened in the early days of crypto.

5:42:13So yeah, so I'd say that it's not even close. I do. I mentioned Alex Chepurnoy half jokingly, because he's, of course, he's not, no one is, you know, it's a joke, but I'm saying like, there are people that are more legitimate suspects. What I mean, there are more legitimate suspects than myself. I wouldn't be able to create a production ready code. The highest piece of code that I wrote was the selfish mining optimization simulation in MATLAB. And it's not that I won't be able to contribute to any piece of code, but definitely not something production ready. So it is what it is.

5:43:08Vlad CosteaThere's ace footage who says, would this upgraded L2 require another token slash coin or just use KAS? Use KAS.

5:43:16Kurt Cech, I hope I pronounced that correctly, says, is it possible that Bitcoin could adopt Kaspa technology?

5:43:39Yonatan SompolinskyI think I addressed it like three questions ago. There's only one scenario where I think Bitcoin, where I think some Bitcoiners would consider as a DAG route. There's only, it's an extreme one, but the world is becoming ever more extreme. So it's not completely unlikely.

5:44:05Vlad CosteaVaruna says, I think some of the other big FUD is fees amount, which is a few hundred daily versus block reward, which is about 344,000 K daily. I said thousand and K. Sorry, it's 344K daily. How will the spread close over the years? So I think the argument is that the transaction fees are small relative to the block reward in US dollars.

5:44:36Yonatan SompolinskyYes, that's the case with any proof of work coin in the beginning before adoption kicks in. It's not a wider gap than what happened with Bitcoin, but in any case, the strong hope or the social, the realistic hope is that this will close, this gap will close when we have DeFi rails implementing the main chain. As we implement the vProg design and people start creating DeFi apps, I'm pretty sure that that transaction fee activity will rise.

5:45:23I can't guarantee it depends more on you than on me, but that's a plan.

5:45:30So in one sense, the the lack of DeFi activity in Bitcoin is is, you know, is harming the chances of recovering the long term security budget. Whether whether and in what pace will Kaspa guarantee the security budget depends on on dApp builders, on liquidity providers, like it's not only about dApps, it's also about liquidity, having enough enough real activity. So but in this sense that the layer, what is called layer two is supporting the security of layer one, which is another argument why you wouldn't want to ignore that.

5:46:21Vlad CosteaKaspa Knight says, how does DAGKnight protocol's dynamic oracle adjustment mechanism improve upon GHOSTDAG's liveness and security guarantees in the presence of adaptive adversaries controlling up to one third of the hash rate, particularly in optimizing confirmation latencies for Kaspa's target of 100 blocks per second under variable network propagation delays? This is quite a long one.

5:46:52Yonatan SompolinskyI don't understand the word Oracle. I don't think it's related here, the word Oracle. But but and let's put the 100 blocks dream aside for now and talk about 10 or two dozen blocks per second. So so even with 10 blocks per second in GHOSTDAG, the security depends on the parameter K. And like any synchronous consensus protocol, you can't confirm transactions faster than the parameter, then that latency, some latency bound that you assumed there is some asterisk here because for smaller attackers, you can.

5:47:42But but but but but but but okay, it's not it's not a meaningless asterisk.

5:47:46I can dive to it and I better write a blog post on this research post on this. But but but for this medium, the answer is that for large attackers, at least you can't in any synchronous consensus, including Kaspa today, you can't confirm transactions faster than O of K. The parameter, the latency parameter, whereas in DAGKnight, there's no a priori assumption what K is. So you can confirm transactions in the order of magnitude of the network's actual latency. There's another asterisk here, but I'll put it aside.

5:48:31It's the worst case, assumed latency, but with the current network, not with not with a network that the devs had to envision in the past. So the implication is it can improve confirmation times from 15 to five seconds or from or even to three seconds. I don't know that it's it's the same order of magnitude of proof of stake or any BFT. Whether what exactly the multiple multiplicative factor there is is yet to be seen. We need to to implement to implement this and check. But but it can be it can be one order of magnitude faster.

5:49:18Vlad CosteaHere's an interesting one from Dark Lord, and the K is basically the reverse Kaspa logo. What is the theoretical market cap we need to see within the next one to two years to maintain a healthy security budget? I saw you mentioned early in the Discord server that we'd want to aim for 10 to 50 billion dollars within 2025.

5:49:47Yonatan SompolinskyI'm not sure if this is a recent Discord message that I wrote. I doubt it. I don't have them. I can't think of the numbers off the top of my head. Not now, but it is definitely a topic that I'm we need to address. It does depend on the pace again of the DeFi adoption. Right. Because the more the more adoption there is, the less the less the market cap needs to grow. So that's why the more secure path for a security budget for now is to increase the activity in DeFi and increase the increased the transaction fees.

5:50:30I had some idea about this. I'm not sure if it's the right forum, but when you see if you saw in the Ethereum.

5:50:41It's even suffers like a year's long tension between L2 and L1 and it became more expressed in the recent year and it talks about taxing the L2s and are you familiar with this discussion or about the taxing the parasitic roll-ups?

5:51:03Vlad CosteaNo, I don't think so.

5:51:07Yonatan SompolinskyAnyways, some people proposed this in more concrete terms. Some people proposed it and were just pissed by the roll-ups claimed like a siphoning of Ethereum. So some people proposed a sort of social consensus to tax roll-ups rather than forcing this in social consensus. Having some social consensus have like a taxation of tokens like say 20% of a DeFi token of this standard is printed out to miners.

5:51:54Again, not forced by consensus but having them not forced by anything else than social consensus. I thought this is a good idea in case Kaspa ends up needing a faster security budget assurance to think about having this socially mandatory taxation.

5:52:21And another route is what I mentioned earlier, the proof of useful work which funds proof of work which funds the mining. This is its main premise. Its main value proposition is that while you mine, you secure the network, you're also doing highly useful things on your own private data. But here, such a change requires a very, very long process. And so the other routes I mentioned are more realistic at least in the scale of one or two years. Like in advance, you can't do a change proof of work faster than that.

5:53:28Vlad CosteaThere is also a question which is a bit redundant because we discussed this, but the answer is very short. So I guess I can ask it just for the people who tuned in later. Ariel Pina says, when can we expect DAGKnight to be implemented?

5:53:47Yonatan SompolinskyI'm wary of answering this. It depends on how much we can scale the core efforts. We are currently stretched thin and we currently prioritize the people designing. And the reason, by the way, we decided to prioritize the smart contract design is because it's a bottleneck for other builders and to scaling up the other activity in Kaspa. If we focused on DAGKnight, then other people wouldn't have a way to build things on top of Kaspa. And so we had this intuition that it's better to postpone DAGKnight, even though it's very late, extremely late on the timeline in order to release first the bottleneck to allow other people to contribute to interesting activity on top of Kaspa.

5:54:49So that's the context why DAGKnight is taking very, very slow. So our aspirations are still to launch these as one hard fork, together with the vProgs, although, not although, because these are two orthogonal modules, then there shouldn't be, it shouldn't be complicated to unite them. I don't, I'm wary of saying exact timelines. We did commit to releasing a yellow paper at the end of August about the vProg design.

5:55:35And so I'm not sure, like Michael forgive me for this, for committing like that early, but giving that we have this soft deadline, then we are focused on that.

5:55:55But we are also in parallel, there are some people that are working on DAGKnight still in slow pace, and I'm sure that once they ramp up the speed and have more like progress to show, then the timeline will be clear on this, be able to give more concrete numbers.

5:56:21Vlad CosteaDark Lord says, how much did you study game theory while conceiving Kaspa and how often do you review Kaspa's trajectory in this area at the present time? How much, what's the second part? How often do you review Kaspa's trajectory in this area, which is game theory?

5:56:44Yonatan SompolinskyI really don't like the term game theory. I think it's a fancy name for common sense. I know, I'm not saying it doesn't have value. And of course, it's something that I am actively researching, but I feel like it adds an aura of understanding that is not necessarily, or predictability that is not necessarily merited. There's not a jab at the person who asked it. It's more like, it's more of a general frustration that discussions that were once about incentives became framed, received the framing of game theory.

5:57:33And it's not, game theory is not necessarily a highly, a much more predictive framework than common sense. I remember in the Bitcoin, when I witnessed the Bitcoin split and the ticker that you mentioned, like I think two weeks before the split, Jihan Wu decided that he won't call it Bitcoin. Do you remember that? It wasn't, it was supposed to be a war of tickers. And then I was naively assuming, at that time, I was very emotionally attached to my game theory models of what will happen. And nothing of that sort materialized. Game theory is just a nice, abstract way to describe good designs or bad designs.

5:58:29And that's the merit of it. Like if you think of a mechanism, then you can use game theory or specifically mechanism design to understand whether it's, whether it's robust or flawed. And in contrast to use it to analyze reality has a much more limited scope of success or predictive power. And my rant here is not, it's not very original. Like people in research have proposed much more like Nash equilibrium was a very sexy term and everyone knows it, but there are much more mundane notions of equilibrium that are less known, but are more of a more predictive nature, like learnable equilibrium.

5:59:23There's equilibrium that equilibrium that is, can be shown to be learnable by convergence when the agents are regular learning agents and there's no, there's too much assumptions. It was, I think I'm, I'm not answering the question, but I guess my answer is I don't pretend, and I don't think others can pretend to use game theory in a strong, predictive manner. It is a framework for good versus bad designs, but not a strong framework, in my opinion of predict for predictive predictive path.

6:00:16Vlad CosteaCoderofstuff says, how do you two have so much energy? This podcast is coming up close to six hours. We just surpassed six hours, by the way. And he also adds, wow.

6:00:29Yonatan SompolinskyI have energy, but I do need to go for like 30 seconds break, bathroom break sometime soon.

6:00:38Vlad CosteaThat's fair. We should have that just for hydration and bathroom. Chat, are you going to allow us after the six hour mark to take a break?

6:00:51I'm just going to put something to cover this. Let's take, let's say, five minute break and then we come back. If you have any more questions, post them in the chat. And before I go into the break, I just want to remind everyone that this podcast is sponsored by SideShift.ai, Bitcoin.com News, Citrea, LayerTwo Labs. And also, I hope I didn't forget anyone, but of course I did. It's NoOnes.com, which is the peer to peer exchange. I have to mention this. And if you guys want to help me get a real Toyota Supra, which is the sports car by Toyota, then I did the math and I need about 6,000 people to buy Cryptosteel metal backups.

6:01:39And I'm pretty sure that we had more than 6,000 people watching this. So maybe like half of the people, if they go to cryptosteel.com and use promo code BTCTKVR to get 10% off when you purchase your metal backup. Then you can actually help me get a sports car, which is insane. Anyway, bathroom break. I'll go get something to drink and keep asking your questions and we will be back in about three minutes from now. So let me put something in the background and we will be back.

6:08:47[Break with background music, about 06:02-06:08.] And we're back. Thank you for waiting for us. I guess we took like seven minutes break, which is fine, but it's important to write down that it was seven just in case we end up shooting for the absolute record of the show, which was eight hours, 21 minutes. I have no idea if we're going to get there, but let's take another question from the chat. Can you have tokenized assets like stocks, bonds, etc. On Kaspa?

6:09:18Yonatan SompolinskyMaybe on an L2, of course. Of course, with an L2, yes.

6:09:27Vlad CosteaKaspa Silver says, would tail emission ever be a thought? Personally, I would consider this admitting failure in a sense.

6:09:37Yonatan SompolinskyI mean, whether we will change the tail, the minting tail, like increase the tail.

6:09:49Vlad CosteaI guess it means because Kaspa has a fixed supply right now, right?

6:09:54Yonatan SompolinskyYeah. So the question is that we're keeping the supply the same, but changing the relaxing the curve.

6:10:00Vlad CosteaI think it's about the monetary policy to add something in perpetuity, like a guaranteed mining reward after a certain point. This is a proposal which is inspired by Monero and Peter Todd came up with this also in Bitcoin. And he published a paper that's called tail emission is not inflation and assumes that a lot of coins are lost.

6:10:30Yonatan SompolinskyEven without that's the main argument of Peter that many coins are lost.

6:10:38Vlad CosteaIn the case of Kaspa I don't think that applies to the same extent.

6:10:45Yonatan SompolinskyI mean you could argue in principle that tail emission is not inflation because it's printing an ever decreasing fraction of the supply. But I don't think this is something that we will consider or I will consider. Again there's no formal we.

6:11:09There isn't even very informal we. But it is a good question and not only with respect to Kaspa but what will happen with Bitcoin's 21 million cap. I know many people are claiming this won't happen because Bitcoin will hit record high market cap before it happens.

6:11:40In this set I think that the Bitcoin the 21 million cap itself in Bitcoin even if already now has lower. Like if it was in the past 100% consensus now it's like I don't know 95% consensus about it. Always never breaking and then increasingly people would propose other ideas. For Kaspa I don't think this will be the approach. I mentioned the three approaches I will take. One is not a very formal approach but working very hard to bring DeFi activity and transaction fee activity. The other is social consensus taxation of activity, of DeFi activity.

6:12:41I wouldn't propose it in consensus but I would see that a good idea or a desired idea to have any token in the Kaspa DeFi ecosystem print a significant, not a majority, but significant amount back to the network to the miners in a manner. The third approach would be to consider proofs of useful work. This depends on the success or failure or anything between the projects that will use it.

6:13:32We won't experiment. We won't be the first to experiment with it. That wouldn't be helpful or responsible. These are the three rules that I think should be considered and not the Not the monetary, not the tail emission. I don't know if it's a...

6:14:01Think even if someone put forth Peter Todd's claim even if they are correct in principle. Money also has a very strong psychological layer of confidence and trust. I don't think it's worth it.

6:14:37Vlad CosteaThat's a good answer. Scott Link says idiot interviewer. Thank you very much Scott Link for tuning in and listening.

6:14:45Yonatan SompolinskyThis reminds me of Mark Twain. He had a good response when someone sent him a letter with the word idiot. So we answered, he replied, I get harassed with letters and many times people forget to sign with their name. But this is the first time I get a letter with someone forgetting to write the letter. You can ChatGPT this and maybe it will be told better.

6:15:20Vlad CosteaNo, it's fine. I just like to poke fun at all of these reactions. I feel flattered that this guy thinks I'm an idiot. By the way, I look forward to your interview Scott. If you don't like this one, do your own. I think this is the right approach. People are asking when Coinbase, when Binance and all of these questions which have nothing to do with development.

6:15:48Yonatan SompolinskyIt's not only about development. I have no visibility into this. I had no visibility into previous listings. I never had. I did not know when previous listings happened and how. It's not that I couldn't know other people in Kaspa knew. I think many of them were or even all of them maybe were open efforts of the community. But I'm not the main person responsible for Kaspa success. I'm at all not in this position. So I guess there needs to be a very strong correction of expectations. I have my my contributions to make.

6:16:33And I'm I definitely like to work hard and enjoy to work hard, but it's not my my sole or my formal responsibility. And I'm not the face of it and I'm not the promoter of it. And in particular, I am not responsible to contact Binance or Coinbase and lead to the listing. It's even if it's a good idea and then someone else should should should exercise a good judgment and and implement this idea. It's it shouldn't if someone assumes this is my responsibility, then that someone has wrong expectations and wrong picture of my both my role and my my informal role and my actual activity.

6:17:22I'm never I never I never made even the slightest of efforts in this direction. So even if Coinbase were to list us tomorrow, I wouldn't know that.

6:17:40Vlad CosteaAltcoin Bonanza says, how old were you when coding and programming became a real thing for you? And something that I want to add to this question is, what was your first computer?

6:17:53Yonatan SompolinskyI'm not of that of that. I wasn't a programmer, a teenage programmer. I played video games, but not but I didn't have a formal informal coding training. Of course, I went to to early school for coding, but not not in the didn't have a strong experience in coding. Didn't have strong attraction to coding in DAGlabs, in the in the company that incepted Kaspa.

6:18:44I felt initially a gap between the mathematical approach of that the more abstract approach that I came with and the coding experience that hardcore engineers came brought to the table.

6:19:02But I should say that as time developed, I found that these that that the mathematical framework or mathematical way of thinking about the system, about the actual architecture of the system is not that detached from relative to what you would think. Meaning you can apply fundamental thinking and reasoning, even in context of discussing bugs, actual bugs, even if you don't know most of the details of the code base. This is this is a revelation that took me. That took some time for me to understand and comprehend.

6:19:51But today, I feel that the difference between the gap between thinking about the system from an architectural engineering perspective and from a mathematical perspective is or perspective is not that wide. I know some elite engineers would would think this is completely delusional, but for me, it's more of a actual realization. I don't find a gap to be as wide as I would have anticipated.

6:20:25Vlad CosteaSo what was your first computer? Do you remember that?

6:20:36Yonatan SompolinskyI remember vaguely the early Macintosh or Einstein editors.

6:20:48Vlad CosteaOh, so you're a Mac kid?

6:20:51Yonatan SompolinskyYeah, but not today. Today I'm fully outside Apple's product. I find it too organized for my style. It's like I don't like Geminis, too organized. I like more ADHD products.

6:21:09Vlad CosteaSo you mentioned that you were playing video games as a teenager. What were your favorites?

6:21:17Yonatan SompolinskyCar races. Many car races. Lotus was a game. Of course Space Command. I don't know if you know Space Command from 1989 or 1990. It's a space... You know what I'm talking about? Space Command?

6:21:39Vlad CosteaI guess you're a bit older than I am.

6:21:42Yonatan SompolinskyYeah, well that's another reason why I'm trying to refrain from these questions because then you'll infer how old I am, older I am than you, and that might depress me. But I think Space Command is like a spaceship that shoots at aliens that become increasingly lower if you don't get it. If you don't eliminate them on time.

6:22:16Vlad CosteaYou're describing Space Invaders, I think.

6:22:20Yonatan SompolinskyYeah, Space Invaders, but there's a reason I mentioned Space Command because there was a reason I remembered Space Command. This was the game. You can infer from this how old I am. I was playing this. I'm very easily addicted to games. That's why at a certain age I was compelled by my parents to stop with these and go study real things. But that was my early computer experience, early 90s.

6:23:21Vlad CosteaSo you said you were into racing games?

6:23:24Yonatan SompolinskyI remember Lotus. Lotus was a game that I loved.

6:23:29Vlad CosteaI remember on my first computer, which was a Pentium One with 66 MHz processor and 16 MB of RAM and no sound card. I had to buy that one. I was playing something MotoGP or something on MS-DOS.

6:23:50Just a second. The Cycles, that's the name of it. International Grand Prix Racing from 1989.

6:24:03Obviously I was not playing this in 1989, but it was one of the first couple of video games that I had on my first computer.

6:24:14Yonatan SompolinskySo did you play a Prince like that kind of games? Or more sophisticated ones?

6:24:22Vlad CosteaI think I like RPGs. I mean, I like the sense of progression, but I also like stuff that tests my reflexes, which is not very story driven. It doesn't require advanced graphics. It's just like Mario games or whatever. I'm a big sucker for that. But I do like a good story also. So in RPGs with good stories, I just love them.

6:24:58Yonatan SompolinskyOh, this is so nostalgic to Super Mario. The whole Nintendo thing. I wonder, I think a lot about Prince because it was such a simple game. And given that kids today get much more sophisticated. I know this is a very generic rant that anyone of the same generation can spell out, but I can't help it nonetheless. I was excited by Prince, but to the same extent that kids today are excited by much higher graphic and expressive games. I didn't enjoy it less. It was a slow, repetitive path in Space Command.

6:26:00I don't know why I remember that Space Command rather than Space Invaders, but these games were so bad in today's terms. It was considered horrible. You wouldn't even consider it enjoyable in any remote sense, but I don't remember it as a kid. It was highly enjoyable. I can't think today if I would be able to play for more than a few seconds.

6:26:28Vlad CosteaWait, so by Prince, you mean Prince of Persia? Mhm. Oh, that one's amazing. Great animation. Even the story is great. Like, you have 60 seconds to save the princess.

6:26:44Yonatan SompolinskyWell, maybe the later versions. Graphic is amazing. There wasn't any amazing graphic back then.

6:26:57Vlad CosteaI mean, animation, not graphics, because the creator actually recorded his brother doing the moves, and basically he digitized the analog footage that he had on camera, on tape. Wow, interesting. It was unlike anything that was made at the time. It was before motion capture.

6:27:22Yonatan SompolinskyFor us, it was the most advanced game we knew, but I

6:27:27Vlad CosteaDidn't know the story. It was, so from a design point of view, I hate it nowadays when I think about it, because it has so many traps, including the bottles that you could drink, and some of them were killing you right away. Like, it was such a cheap way to take you to the starting level, and there was a lot of trial and error.

6:27:47Yonatan SompolinskyYou have to have the discipline not to give up, or you went back to the starting level.

6:27:54Now I sound very old when I'm ranting on this, but today's kids don't have this. Like, the games don't give you the same experience of grinding, right? You have to grind back, and cope with the failure, and start again and again. It's not the same experience at all.

6:28:18These sort of games were very productive to your development, even if you were addicted to them, they were still very productive. They did build some resilience in you.

6:28:36Vlad CosteaTo draw a parallel, I guess this is why kids these days don't like proof of work, and they just like the convenience of proof of stake. They don't understand the merits of something which requires a lot of work.

6:28:54Yonatan SompolinskyYeah, I see what you did there. I would say it's a more general phenomenon. There's a nice quote of Ronald Reagan where he says that liberty is always one generation away from extinction. It's always on the verge of extinction because the generation that received it doesn't know what it took to get there. I think this explains a lot of phenomena and generational divides.

6:29:29Vlad CosteaGoing back on topic, it seems like we have quite a few questions that stacked up. Ollie wants to know if you can imagine that there's going to be an ETF for Kaspa.

6:29:47Kaspa Knight says, how does DAGKnight's client-specified local upper bounds on adversarial latency circumvent the Pass and Shi impossibility results for parameterless, permissionless consensus while securing against up to 50% hashrate corruption? And what trade-offs arise in confirmation times and client synchronization for high throughput DAGs like Kaspa at 100 blocks per second?

6:30:24Yonatan SompolinskyThis is half trolling and half serious. This is what I mentioned previously, that proof-of-work enables DAGKnight to circumvent a known bound on protocol, so the known bound is both by Cynthia Dwork et al. And by Rafael Pass and Elaine Shi. The bound is in the traditional consensus, the bound states that if a consensus network is of partially synchronous nature, meaning it does not assume an a priori upper bound on latency, then you can't be more than 33% within the protocol token. This is a known result and it's easily demonstrated.

6:31:26If you don't know how long to wait, then an attacker can participate honestly in one faction and honestly in the other, and then to trick clients to approve transactions. It's a classic consensus attack for manipulation and it's well known that a network without a priori latency bound cannot be more than 33% tolerant. Rafael and Elaine generalize this into proof-of-work. I actually conversed about it with Elaine in Cornell in the Ethereum boot camp when I visited there and the main discussion was around this claim whether you can accelerate, whether you accept transactions without a latency bound, and then Elaine and Rafael published the paper.

6:32:39And in the paper, they show that a protocol cannot be responsive to the actual latency that it is observed in the network. And in DAGKnight, we do not assume, I mentioned this briefly before, in DAGKnight, we don't assume that you can't set the latency in your local client based on what you see in the network. If your assumption, let's say the network is smooth, if your assumption is that the network is, let's say the network has two seconds or maybe let's say that the network has 200 milliseconds latency.

6:33:22You can't set the latency according to what you see in Kaspa network. That would be insecure and would break the impossibility bound because an attacker can always help the network be secure until it wants to carry out that attack. So when we say that DAGKnight is responsive to the network's latency, we do not mean the actual observed network latency, but the actual worst case network latency, meaning you have to have some assumption on how bad the latency can become if there is an attacker lurking in the network.

6:34:08So meaning if you think the latency is currently smooth, but your own threat model, your own analysis has led you to believe that the latency in Kaspa is really 10 minutes, but the attacker is just waiting in the dark and waiting for the right opportunity to split the network, then your local client should set it for 10 minutes. It can't set it for two seconds. But if you have a way and you should have a way with simply pinging different locations, even outside Kaspa network, to reason about the latency being two seconds, even in the worst case maybe three seconds in the worst case, then your local client should set it to three seconds.

6:34:56So the short answer is you have to use some internet pinging or Kaspa pinging external methods to reason about them or convince yourself about the real worst case latency in DAGKnight, and that's what you set in your client. How exactly to implement this, what exact method we will take? I don't know. Even if we will decide on something, you can decide on something else for any client to decide on their own to set in their local client. What are the trade-offs? There was the word sync there, but I don't think that was meaningful.

6:35:50The trade-off in DAGKnight confirmation times is not different than Kaspa, than Kaspa's GHOSTDAG or Bitcoin. It's just the nature of all of the finality.

6:36:05Vlad CosteaThere's a much simpler question from Aurelien Panitza, and he says, is Aviv Zohar still involved in researches related to Kaspa? Thanks to Yonatan for sharing that much of his time, pretty awesome.

6:36:24Yonatan SompolinskyAviv Zohar wasn't involved in Kaspa. I offered him to be, but he chose to do other things.

6:36:38So yeah, I do think I could have benefited from his maybe extra research and external advice unrelated to research. I think he could have added many things that are missing, but again, history is what it is. It's the route that was chosen.

6:37:25He was involved with, of course, with GHOSTDAG, even after DAGlabs was launched, but not after Kaspa was launched.

6:37:37Vlad CosteaOkay, there's Kurchak who says, tough question. Do you see a world where things are priced in KAS in the distant future, perhaps after a phase of stablecoins and a phase of hyperinflation? Is Kaspa built for this?

6:37:53Yonatan SompolinskyThis sort of question is, to me, belongs to the set of questions that I ask myself first or I push back to the person asking the question about crypto in general because my answer would highly depend on whether any other asset in crypto achieves this or not. I don't think placing the weight on Kaspa is a good way of thinking of this. There's lots of questions.

6:38:38There's lots of questions that my immediate response, depends what happens with crypto. Kaspa doesn't have the power in and of itself to push crypto by its own merit.

6:38:52You can consistently be a Kaspa maxi and still be and still acknowledge that Kaspa's fate is intertwined with crypto's fate and it can't be too much of an outlier. So, for me, this is more a question of where do I think crypto can get rather than Kaspa.

6:39:28Thank you. I would say I don't see a world where Bitcoin is highly used other than store of value and Kaspa is not. I can see a world where Bitcoin is not used, you're not the king of collateral, just store of value. But if Bitcoin is used, I don't see a world where Kaspa is not adopted as well.

6:40:00Vlad CosteaMark Kruk says Hi Yonatan, you compared Kaspa contracts to Solana. Explain why Kaspa will be better and why banks will ultimately adopt the better tech.

6:40:15Yonatan SompolinskyI don't know that banks will adopt the better tech. Maybe they will be replaced by the better tech. But the reason I'm referencing Solana is

6:40:30It's more because it achieved due to its monolithic design, a very cohesive society of developers so to speak, and Ethereum suffers from so much fragmentation and the developer experience in Ethereum has become too fragmented.

6:41:15The Ethereum ecosystem became too fragmented and it's no longer clear what does it mean to be in Ethereum. Like let's say someone takes Kaspa and implements a bridge to make Kaspa a sidechain to Ethereum with Polygon was able to pull off under the name of L2. Like is it still Ethereum, It's like a spectrum of gray. It's unclear what does it mean to become Ethereum, whereas it's very clear what does it mean to build in Solana. And this is what I'm referring to.

6:41:52There's also, of course, the rust nativeness, which is nice. And again, these things will not be coded by me. Hans is working on this. Of course, Michael is working on this, and others. I'm not implying that I'm coding this design, but from a vision and direction point of view, it's clear, to me it has been clear for several years now that the way that users and Kaspa builders should feel in the ecosystem is much more similar to Solana than to Ethereum.

6:42:51Of course, Ethereum is a much more interesting community in the sense of research and ideas. And I guess it's the most interesting community and most contributing most to understanding of DeFi and other ideas in crypto. So of course, it has lots of merit.

6:43:20So, this is not a jab at Ethereum so much as an aspiration to learn from its mistakes. And I think a lot of what's happening at Kaspa is a timing issue, it's a late runner advantage because we are developing smart contracts in the era where ZK is in real-time proving is in the horizon and ZK is already production ready, then we get the benefit to be the first or one of the first layer ones to incorporate ZK design in advance. Like It is very clear, as Vlad mentioned sometime along this conversation, at some point, that if ZK software was feasible at the time, Bitcoin would have looked different, and Bitcoin Core Devs would have chosen different, or would have stuck to different views.

6:44:23So, Kaspa has this advantage that we have the right timing, we're developing smart contracts exactly when ZK tech is ready. You can consider this smart or lucky for your choice, but the bottom line is we want the same developer experience as Solana, and the same social cohesiveness of Devs as Solana, and at the same time maintaining Bitcoinite security in the sense of separating the entire financial logic from L1. This is an endeavor that is, not a moonshot, it's something that once we publish the formal model and paper, both a yellow paper and a more academic-like paper, it will become clear what the formal claim is and what the formal guarantees are.

6:45:31So it won't remain opaque for too long, the claims will be made concrete in writing, but I just want to make sure that it's not falsely understood that you'll be able to code actual Solana contracts in Kaspa, because Solana itself is not a ZKVM. You need to create a ZKVM. So the developer experience would be very similar to Solana, but formally you will write in a ZKVM, a zkEVM, SP1, maybe RISC Zero. So it depends not only on us, it depends on what demand for a prover system or prover network there will be, but I guess we will whitelist anything that has been, is that just reasonable in the eyes of, that has consensus on its soundness.

6:46:38So it's not importing Solana as it is, it's more aspiring to achieve the same level of cohesiveness and experience and Rust nativeness, etc.

6:46:52Vlad CosteaKaskat, and I guess his Kaspa domain is noderunner.kas, says, who are your heroes, past and present?

6:47:05Yonatan SompolinskyGarry Kasparov, I'm a child of the hero. The chess player. Yeah, I had the fortune to, the honor to meet him a few years ago in person. But I strongly remember him as one of my main childhood heroes.

6:47:31I think my current hero, I will keep to myself, but that's in my childhood.

6:47:45Vlad CosteaDark Lord says, how do you view the shift in attitude from academic thought leaders such as Charles Hoskinson, Vitalik Buterin, and Adam Back, who once publicly commented on your work, commended your work on papers like GHOST and SPECTRE, but now either avoid acknowledging your contributions or in Adam Back's case, openly criticize Kaspa as a scam?

6:48:24Yonatan SompolinskyI actually don't have a strong opinion on this. I'm not thinking about these matters too much. I think Kaspa in general is somewhat segregated from crypto in the sense that it's just, and there's fundamental reasons for that. I don't know if I want to get into it. It's definitely not in the context of answering this specific question, but Kaspa is segregated from crypto. There's the assumption that everyone knows everything about Kaspa and is just intentionally avoiding it is, I don't think it's rooted in reality.

6:49:11There's also the choice to be, you know, the choice of the product where individuals in the project to be less visible and, you know, it's a project with a deliberate effort to avoid figureheads, and I think this will be the case as long as Core is not scaled up. So the more people join the core of Core, the more visibility Kaspa will have.

6:49:47I originally made the choice not to become the project's figurehead. And I think that that's a price that Kaspa is admittedly. Paying in that people find it easier to associate a project with someone that continuously, with people that are synonyms to the project, for instance, Ethereum is synonymous with Vitalik and Avalanche is synonymous with Emin Gün, And Kaspa is not synonymous with its founder or originator. And this is a choice, and I think it's a good choice for Kaspa. But I think it's a price, this is one of the implications.

6:50:36This is about others ignoring it. I do think that the new innovation, the vProg design, is, I do think that as we implement this layer, we will naturally both push for and attract the interest of broader communities. So there's the DAGKnight will interest the Proof of Work community, and the vProg design will generally interest the DeFi community and researchers and thinkers. That I see happening.

6:51:30Regarding Adam, so first of all in general, I'm order of magnitude less aware of things that are happening on Twitter than many people assume. It's not that I'm not aware of it's just I'm order of magnitude less aware than others assume. So this is like a general comment about people's assumptions on my online presence, or online awareness. Even when I'm on Twitter, sometimes, or frequently, my feed is not only Kaspa, not necessarily mainly Kaspa.

6:52:15I'm definitely interested in LLM, and my feed is also flooded with that bleeding edge technology as well.

6:52:26So I'm not necessarily, I don't know necessarily most of these arguments or clashes on Twitter. I'm emphasizing this because when I do see people mentioning this, people assume I'm aware of the context and I have no idea what they're talking about. It's kind of weird to have people assume too much. Anyways, regarding Adam, I just don't have a very strong opinion, at least not at this hour. I actually don't think about it too much. I don't think Bitcoiners would ever, again, until the next electromagnetic pulse gets to us, I don't think Bitcoiners will, I don't have this expectation that OG Bitcoiners will appreciate Kaspa.

6:53:26I think maybe to give an analogy will explain this. There's the Michael Jordan camp and the LeBron camp, and there's the whole, the everlasting debate on the GOAT.

6:53:47Vlad CosteaActually, I'm on the Kobe camp, but that's another conversation. Oh, that's very you.

6:53:52Yonatan SompolinskyYeah, that's very you. Yeah, the founders no longer around. So, even if you are in the LeBron camp, you should have zero expectation from people that saw MJ playing to have any flexibility to think that the discussions started. People that witnessed MJ would laugh at the very thought that you would consider LeBron. Even if you showed them, you know, you spend as much time as Nick Wright spends to explain the whole, the whole justification for this. It's still delusional to think that people that saw MJ would even fathom this idea. There's many factors to this.

6:54:52There's a generational gap. There's the fact that there's very little social media. There was very little social media, social media at the time. There are very little videos of MJ missing a shot. Not to say missing a buzzer beater or a game winner. So, Michael Jordan is Godified and there's no way to, there's no, it's not reasonable to expect people that witnessed him in his prime to consider anyone else. So, this is not me turning the other cheek, rather being very practical on people's mindsets and fixations.

6:55:45So, now today Bitcoin, so that's real Bitcoin maximalist, or what I call crypto-minimalists. Of course, today the word Bitcoin maxi on Twitter refers to a more, you can also say Solana maxi, which is weird, but okay. Like you can say, you can say Fartcoin maxi. I don't know what that was, these things mean anymore, but it's not, it's, it's, it's not referring to that, to that initial group of people and an initial protocol that propped Bitcoin at the beginning. So, I think this is where, this is where I come from.

6:56:31I don't have any expectation. I'm neither insulted nor surprised. I'm kind of indifferent to that. It's not, I'm not saying this is the correct stance so much as it is just my, it's my emotional stance, I guess, towards this. I'm not promoting, I'm not advocating for anyone else taking, using this as guidance or guidance or recommendation of what Kaspa advocates should or shouldn't do and not at all. I mean, there is merit to these pushbacks and debates. I'm just reflecting why I am not naturally part of it or can't, don't feel comfortable in part of it.

6:57:25I just see them as the MJ generation, which will never fathom anyone else, any other opportunity, any other possibility.

6:57:36Vlad CosteaThat's an interesting metaphor because the follow-up question is from Warisantv who says, can Kaspa flip Bitcoin in the future and flip means exceed in market cap, I think.

6:58:01Yonatan SompolinskyI think that the market cap of money and the market cap of silver are of the same order of magnitude. I mean of USD. What's the current market cap?

6:58:41Vlad CosteaFor Bitcoin it's 2.2 trillion, for Kaspa it's, just a second, 2.2 billion. So basically a thousand times.

6:58:55Yonatan SompolinskyNo, I was, I was not looking at this. I was looking at, as I thought, it's the market cap, not silver, gold is the same, is roughly the same as the market cap of USD.

6:59:13I don't think Kaspa will replace or any asset will replace Bitcoin's gold status. But for the money, for the money role, whatever money entails in a, in a broader sense, again, then, and what we envision, it's not necessarily in the direct, indirect payments. It can also be the parking, the parking spot for our investments or the collateral for our loans, whatever property, traditional money functions in a, in a financially enabled ecosystem.

7:00:04Why not? I mean, it's, it is surprising that USD and gold are the same order of magnitude. Of course you can measure USD in various, in various, you know, metrics. It's a too well defined concept. But I remember that, I remember seeing that and asking, I remember seeing this and being quite surprised by this comparability of money and gold.

7:00:42Vlad CosteaI did not know this, but now that I think about it, I think that the market cap of cash was larger than the one of gold eight, nine years ago, when Roger Ver was so adamant about making Bitcoin compete with USD. He was saying, well, there's a lot of money in this. It's not just about store of value.

7:01:10Yonatan SompolinskyIt just changed since like what a magnitude or?

7:01:13Vlad CosteaNo, I guess the price of gold went up. I remember people were making fun of Peter Schiff in the previous years that he criticized Bitcoin in 2010 or 2011 when the price of gold was the exact same in USD as last year. But this year has been incredible for gold so far.

7:01:39Yonatan SompolinskyYeah, of course, this is, this is a moonshot and this is not something that I am actively pushing or know how to actually push for. I have no idea. I am very focused on what I do know how to do and not very, okay, I don't think, I think for this to happen, Kaspa must be more in the complete stage of that, the latter alluded to, and in particular shouldn't have even candidate figureheads in the, in the area.

7:02:38I think it's not only or primarily like values of decentralization, more of a practical assessment of what property of an asset needs to, what property does an asset need to manifest, to truly manifest in order to even consider such market caps.

7:03:08And I think other than Bitcoin, you have to have very strong Bitcoin that's why I'm pushing. When I say Kaspa must Bitcoinize, people think that people have read me and read my statement as saying Kaspa needs to replace Bitcoin you know, or compete on Bitcoin or to be the second Bitcoin, second to Bitcoin, all sort of interpretations. But I think I made it clear, where the context was clear, that it's not so much a technology, a technological positioning, more of the culture and the actual, so when There are several properties that heavy holders or heavy investors, people that would know or would be interested in money, they would need to see in a system in order to believe it can be money.

7:04:22And as long as Kaspa is not complete and that your humble servant is in the area, I think this is, like, this is still a blocker, but after that, let's go for it.

7:04:50Vlad CosteaThere's also a Tina on the chat and says, oh my God, oh my God, you guys are still here, have you two had dinner yet, lol. Did you have dinner, Yonatan?

7:05:05Yonatan SompolinskyI thought you were all for privacy.

7:05:09Vlad CosteaOh, I can reveal that I haven't had dinner. I just had a big lunch.

7:05:19Yonatan SompolinskyThere's no complaint on my side. I like to hear like the idiot interviewer as you were honestly referred to.

7:05:31Vlad CosteaG.F. Burke says, does Yonatan talk or collab with the KYI team?

7:05:41Yonatan SompolinskyI don't know who they are.

7:05:43Vlad CosteaKey, Kai, I don't know. AFL, AFL. I guess he previously asked a very good question, says, what do you think of the DeepWiki for Rusty Kaspa? Michael says it's mostly accurate.

7:05:59Yonatan SompolinskyIt's amazing. So there's a, I think, Cognition team. It's on Twitter. I think they are called Cognition. They are, they're, they developed an agent that's called, they call Devin, that answers on the Rusty Kaspa, the Rust answers, or on many GitHub repositories, not only in Kaspa, but it's amazing. I'm actually trying to create now an agent that will integrate also with Devin and be able to answer to answer questions on Kaspa. I think the LLM, one of the, one of the promising one overlooked admittedly niche values that LLMs can bring is to, is to solve the paradox that we were touching on from different angles that there needs to be some central points of information for a project.

7:07:09And at the same time, they shouldn't be individuals. And this can be an individual humans, and this can be in several contexts, not necessarily only in crypto where you would want an intelligent centralized and intelligent central governance, but not the human being. And in this context, I think even outside the content, even outside the context of Kaspa, even outside the context, initially crypto, there will be interesting use cases where you would see organizations that are governed by LLMs and a more humble or micro implementation of this idea is.

7:08:00It's something I'm trying to create, and I think Devin is very helpful there. From the Devin, the DeepWiki agent. But i'm, I'm excited by the by the possibility that this will, that this idea of LLMs, autonomous LLMs, would just be a solution to think. You can you can imagine it being a solution to many problems that we didn't think are solvable because you need intelligence, centrality, a central intelligence agent, but You can think of it, think in terms of let's say governance, the governance of a, of a software, open source software, where you would want to make decisions not based on the vote, based on, not based on staking, but based on actual evidence or reason, then you could, if you train the model in a transparent way, then it could be the case that these things will have many, many many, many shapes. In this.

7:09:11They will take many shapes, these LLMs, autonomous LLMs. I'm not sure how I got into this. What was the question about the, yeah, it's mostly accurate and I commend the team that they've developed it. Thanks a lot.

7:09:37Vlad CosteaLavendi Pro says, will we see Yonatan in London this year?

7:09:43Yonatan SompolinskyYeah, I was, I was, I was invited to, by KEF, to London, some London conference, I don't remember what it is. I'll probably go.

7:09:58Vlad CosteaAwesGhosty, I think, says, how do you think the community could improve? This includes community developers and community on X.

7:10:13Yonatan SompolinskyThanks for this question. The community can, I think, become more informed on crypto in general. That will help everyone understand what Kaspa contributed and what it borrowed from other projects.

7:10:28I think that's, if someone, if someone wants to be a very good, promoter of Kaspa or advocate for Kaspa. I really suggest knowing a lot about Ethereum, Solana, of course, but Ethereum mainly, Bitcoin mainly. Like many of the things that we are innovating are deeply rooted. Their V1 is what was innovated by others and appreciating our innovation and contributions must be in the context of those previous works.

7:11:16It just doesn't make, just consider, I think if you take the power of the Kaspa community today and combine it with more, with some more context on crypto, I think that's an unstoppable force.

7:11:34But it requires this effort of curating or securing in context on crypto. Like for instance, the ZK rollups, understanding that this in and of itself is not an innovation that stemmed from our community. Understanding that most of Kaspa's or the main, the base layer, the base ideas are of course rooted in Bitcoin. Not to downplay our contribution rather to position them correctly. And I think, of course, the lack of context was also very facilitative and have the true benefit of being loyally enthusiastic about Kaspa. And it's a very loyal and loyal to the project, a very self-loyal community.

7:12:44And I think combining this with slightly more context on crypto would become a game changer in terms of community outreach, the project's outreach. I think other founders would find it more difficult to ignore Kaspa if this change or development takes place.

7:13:16Vlad CosteaAustin wants to know what is the theoretical maximum transactions per second for Kaspa if it can become as fast as the internet?

7:13:32Yonatan SompolinskyWhen I consider fast, I'm interested in the first confirmation for me. So there are three, there are three latency, right? There's the inclusion latency, and then there's the probabilistic finality, and then there's the final finality. So when you, when I say this is proof of work, this is internet speed proof of work, I'm referring to the first inclusion. And for this, there's no limit. You can do 10 to the power of 100 transactions. The trade-off is that the more you increase the throughput, the latency, the block propagation latency grows, and then the second latency, the confirmation latency grows.

7:14:24But if you're thinking like me in terms of UX, of the first inclusion, then there's no bound. I'm glad I don't want you to feel forced to continue this. I suppose we should be respectful of your time and platform. We don't need to exploit all of your generous efforts for this. So feel free to pick maybe one or two more questions and then cut it there.

7:14:56Vlad CosteaWell, that's a lot of pressure because there are a lot of them. There is also Tyler who says that during this interview, over 180,000 Kaspa blocks have been generated.

7:15:11Let me pick something interesting from here. Luca Pichinotti says after 100 blocks per second for KAS and independent implementation of L1.5 and interoperability of L2, where do you see KAS? On which side the L1 could be improved? Will there be full zk?

7:15:37Yonatan SompolinskyFull zk, I guess, in the sense of proving the L1. Otherwise, I don't know what this is referring to. I would pick another question or I would say, again, for me, as I mentioned previously, there's one additional, perhaps one additional to me, it's a major upgrade, but technically it could be a minor one that can change the way a transaction flows in smart contracts. That would be the other upgrade that I would propose.

7:16:27And whether there'll be full zk, while I don't understand the question, I would say there's lots of promising progress on the hardware side and real time proving. And this will have an effect even on our design, how much, how much trust, how cheap communication between vProgs is. But other than that, yeah, let's pass it to the next question.

7:17:00Vlad CosteaSo Angel Puente Reyes, I guess that's Angel if it's Spanish. Yonatan, could you mention atomic composability and logic zones in the context of Kaspa?

7:17:16Yonatan SompolinskyYes, so the main thing that, the main feature that prevents this fragmentation that Ethereum, maybe let me backtrack and explain what happens in Ethereum. What happened in Ethereum with the fragmentation of L2 to different rollups is that people started thinking or advocating for the interoperability of rollups. But even this interoperability is more than just using Ethereum L1 for standard bridge or standard message communication layer between the rollups in an asynchronous manner. To interact with layer one, then your rollups sent to L1, then L1 sends to the other rollup.

7:18:05It's still, it's called async composability. And many, many people in the Ethereum community, including Vitalik himself, argue that asynchronous composability is more than enough for most of the needs. And what I argue is that this could be the case only if you already succumbed to the existence of rollups to begin with. But if you want to prevent this phenomenon in advance and you want people to build contracts with, not contracts that try to, if you want people to create a contract that with incentives to interop with one another and not with incentive to attract as much activity to this island without interoperability with others, like rollups natural incentive is, if you want that, then you have to have synchronous composability.

7:19:22Let me state it clearer. Let's assume this is not the case and let's understand the problem. Let's assume we're giving up synchronous composability in Kaspa, which means we will implement Solana-like CPI again with zkvms, but the messaging between programs would be through L1. So you will suffer a proof latency to L1 and a proof latency to the other program. Then other than the bad UX for composite transactions, for transactions composing different contracts, because the users will have a risk of their transaction not implementing the logic they wanted.

7:20:07So due to this financial inconvenience and risk for users would highly prefer smart contracts that interact in a synchronous manner, and this will incentivize users to work and cooperate within rollups. Rollups, in this sense, are an entity that, if you don't implement synchronous composability natively in the program, you're giving rollups an advantage over regular smart contracts, meaning you're giving entities that want to siphon the network effect of Layer 1 an advantage over regular smart contracts.

7:21:02This is why it's very important to commit to synchronous composability from the outset, even though it requires more heavy efforts on the research and development side, in order to ensure that the development of DeFi products on Kaspa is cohesive.

7:21:33There's one society that's developing and expanding, and not fragments and islands of societies. This is the context for the motivation to commit to synchronous composability, and why we are taking a very different route than what Ethereum has took. And if you look today at what Ethereum is talking about, then there are voices in Ethereum that are trying to reverse the wheel and go back to a more L1 enshrined native. These things mean slightly different things, but it's in the same direction of trying to make everything cohesive through L1, rather than delegating the activity and standards to L2.

7:22:37Vlad CosteaThank you for that. There's someone in the chat who has been trying to ask a question for about two hours, and for some reason I didn't pick up this question, so I'm going to read it now. His name is 33Ethereum, and he says, when Sparkle is developed, even though it isn't the same as running SC on L1, I guess SC is synchronized composability, will the UX still feel as though it is monolithic? Will there be bridging necessary?

7:23:15Yonatan SompolinskySparkle is Anton's project, I don't know what's the status of Anton's efforts. Here Anton is definitely to be credited for throwing me personally in the direction of Solana years ago. He's consistently told me about the advantage of running or developing Solana versus Ethereum, but other than that, Anton is the address, is a person to discuss Sparkle with. But I do want to emphasize that, yes, in the monolithic versus modular debate, I view Kaspa as appearing monolithic.

7:24:17In practice, you will be able to integrate into Kaspa several proof systems, you would be able to run a zkEVM, SP1 VM, maybe RISC Zero, maybe Cairo, you would be able to integrate your own ZKVM, and you would definitely be able to dictate your own programs, state size, state growth regulation, throughput regulation through gas, so definitely there will be clear autonomy and sovereignty for builders to dictate their condition and their terms.

7:25:05But for users and developers, it will still feel much more monolithic and much more similar to the experience of Solana. Of course, not the same, there will still be the more diverse VM, diverse state sizes, but it will feel more like native, not in the sense of using L1 computation or resources, but in the sense of having the same cohesive standard, the cohesive way of talking with one another. So I think it will feel like native, it will feel L1-ish, even if it's not an L1.

7:26:01Vlad CosteaI just want to announce that seven or eight months ago, I've had Amir Taaki, and he spoke for seven hours and nine minutes, 26 seconds. Right now, we are at seven hours, 26 minutes, so even if we do include the break, which took like six, seven minutes, you beat Amir's performance. So congratulations on that one. It's not the absolute record, which is eight hours, 21 minutes. I'm not sure if you want to go for another hour just to break this otherwise kind of futile milestone.

7:26:39Yonatan SompolinskyI wouldn't want to keep it only for the sake of breaking the record. It will be disrespectful for your podcast, I think it was very valuable.

7:26:50Vlad CosteaBut I'm happy I was able to serve as sort of a bridge between the community and yourself, and also I got to learn more. I got to fulfill my curiosity. I think the only question that remains, and I don't know how to answer this, is like how can all of these people still call you a scammer, even though you spent the past seven hours not trying to sell anything and just explaining how stuff works, sort of beats me. Maybe that everyone's a scammer if they launch an altcoin that competes with Bitcoin or tries to complement it. I don't know.

7:27:34Yonatan SompolinskyLet's bring them on your podcast.

7:27:36Vlad CosteaI should bring Luke on my podcast, but he's gone mad. I don't know. It's very hard. The issue with Luke Dash Jr. Is that what he says and what you understand, because he has a very different understanding of language, and there's going to be this disconnect between what you understand from what he's saying, and actually what most people understand from these words, and what he thinks that he's saying. So you have to constantly play this game of getting him to define his terms.

7:28:12Yonatan SompolinskyI seem to be in your realm of expertise.

7:28:18Vlad CosteaSure. I mean, I'm going to try to get him on, even though I think he's sort of compromised. He was so much better a few years ago. Now he's gone mad. There are running jokes that he's going to cause an accidental, more or less hard fork in the future, because he's running his own node implementation, and he's the only guy who's actually competent in that whole team, and he's surrounded by yes-men.

7:28:46Yonatan SompolinskyI was actually hoping that the hack that he suffered was just a way to evade tax payment in the US. I thought that many competent people, if they have very large amounts of Bitcoin, and they want to evade tax, if they can declare it to have been hacked, this could help them avoid tax payments, and so they get the hit of their reputation, but they get to save the money. That's what I kind of thought that would have happened, or that was the backstory.

7:29:29Vlad CosteaI don't know, because what he did in the aftermath of this incident was to call the FBI to raid an entire event in Atlanta, and get the list of attendees, and basically everyone got doxxed, and it was horrible.

7:29:52I have respect for his accomplishments, I just don't respect his views anymore, and the moment when he rubbed me the wrong way was when I was trying to talk about sidechains, and he sort of said, why would you want sidechains, because at least with more centralized systems like federations, you have someone to sue in the event that funds get stolen. The security of these scaling layers ultimately lies on the judiciary branch of nation states. We're building technology that functions without state permission, but once we get to layer 2s or sidechains or whatever, we're gonna go settle in court, I think that's dumb.

7:30:45Yonatan SompolinskyThis sort of escalation of compromises is not unexpected when you're so adamant and fundamentalist about, or when you're too puritan about taking the extreme fundamentalist approach towards the system, then when you try to, then when you do succumb to reality and say, okay, let's, let's, let's, let's integrate something that's useful, then suddenly you just go to the other extreme and use federations or some forms of committees.

7:31:37I hope Bitcoin stays, I hope it stays pure and not, if it's fundamentalist, let's keep it pure fundamentalist.

7:31:53Vlad CosteaWell, I hope Bitcoin actually scales and does more interesting stuff. And I hope we're gonna see more of that this year with the ZK rollups, even if they're optimistically verified. And I don't know, I signed up for a kind of revolution that never happened. And now I'm in a situation where I'm supposed to watch what some government employee is saying about interest rates in the U.S. To know whether or not the price goes up. The price of Bitcoin goes up or down, or I'm supposed to follow some ETF fund manager to understand what's gonna happen.

7:32:32It's, I don't know, I didn't sign up for this. I just hope that there is still sort of a subversive element in Bitcoin. And it's not completely drained. And it's not just something that's gonna be used only by banks and custodians. And on-chain transactions are going to become this sort of rarity.

7:33:02Yonatan SompolinskyAll the power to you, Vlad. I think this, I think Bitcoin can only benefit from your voice attracting order of magnitude more prominence.

7:33:18Vlad CosteaI don't know, I'm just trying to remind people of the early days. And while I'm at it, I just want to thank everyone who sent Kaspa to this wallet. I was able to successfully redistribute the coins that I received around the first hour of the interview to people who stuck around past the four hour mark. This is like a reward. I didn't want anyone to think that just because I put a QR code and I set up a wallet, I'm sort of begging for funds or whatever.

7:33:52Look, someone just sent me two Kaspa coins. So I'm not trying to profit off of anyone. I just wanted to try the technology and give it all back to the community. I don't try to do anything here. The show is already sponsored. I'm fine.

7:34:10Yonatan SompolinskyBut do you have a Kaspa address?

7:34:12Vlad CosteaI do have it on the screen. So someone watching will have to scan. Maybe I should generate another one that's personal because this one is too exposed. Yeah, but it's fine. For this address, I'll just give everything to the community. And thank you guys for watching and listening and asking some very good questions that I couldn't think about over the last seven hours. Is there anything that you would like to add, Yonatan, before we wrap this up?

7:34:49Yonatan SompolinskyNo, I appreciate very much your moderation, your questions. If it was a privilege, I'm very grateful for that. Thank you very much.

7:35:02Vlad CosteaThank you for your patience and responding to all the questions. Thanks to the community. And I guess I'll see you soon. Good night, everyone. Ciao.