Open Blockchains Are The Future w/ Mert

26:25Alex Thorn:Yeah, it's been tried a number of times. I mean, when I saw Canton launch, I couldn't help but think about the blockchain, not Bitcoin era of like 2016, where the banks all said, oh, you know, we love that blockchain, but Bitcoin, I don't know, that's like for criminals or something. It just, it seems like one of those ideas that refuses to die. And so, but I agree too. It also feels like an intranet, you know, and the internet is not built on private intranets. Exactly. Yeah. I mean, the main thing, and I have a few writings on this, but fundamentally, what makes a blockchain interesting in the long term is if you have shared state that's composable with all other actors, then you are basically compounding on that bandwidth.

27:15And then people can use that to build things that they just could not build in Web2, right? I mean, for example, if you're a developer and you want to just write a piece of code that moves money from Malaysia to, let's say, Turkey, how are you going to do that today? It's actually extremely difficult. You need to sign up for some KYC APIs and maybe get access. And then even then, there's so many fees and there's no unified interface for this but blockchain is really just api for money right or i call it the api for capitalism and you know you can put whatever adjective you or noun you want there uh internet capital markets all this stuff but that's the thing where you have people's local knowledge which is dispersed globally and time sensitive when you have that knowledge uh able to be expressed on some unified, let's say, canvas, then you have all these entrepreneurs building different things that are super useful without asking for permission from the banks.

28:12But otherwise, instead of using Chantin, why not just use PayPal's APIs? It's fundamentally not that different.

28:21Alex Thorn:Yeah, I agree. And you wrote eloquently about another noun for what you're talking about, which was cipher capitalism. And, you know, you said that, I've got a quote here, you said the blockchains in theory increase the bandwidth while reducing the noise of economic knowledge propagation at planetary scale. It sounds like you are a open permissionless cryptographic network maximalist when it comes to this technology. And, you know, I think you've already explained pretty well the, because it's not just use PayPal's API. To me, it's like, Just use a SQL database or something like you used to, which can be permissioned and shared.

29:00Alex Thorn:There are shared databases. I wanted to ask you a little bit on a more technical side about something happening in the Solana, well, I guess being proposed in the Solana ecosystem, which is called Multiple Concurrent Proposers, MCP, which at a high level, it changes how blocks would be produced and away from the sort of single leader model, right? where one validator produces and then another and then another and your likelihood is weighted by your stake. If you could, at a high level, what is the change? What is the proposed change? And then what is the state of that at the moment? It's first worth understanding that, yeah, again, like you said, it's a proposal.

29:46And so it's being proposed by Anza, which is the new name for Solana Labs, which is the main client entity. And they have good, I think it's called Constellation. They have a good website that's interactive that helps you visualize what happens when you send a transaction, where does it go, how does it get merged. But without getting too technical, the fundamental difference here is today when you send, so let's say you're on your phantom wallet and you send a transaction. That transaction gets sent through an RPC to either a set of validators that then forwards to the leader or it goes directly to the leader.

30:24So the way it works is every slot, which is, let's say you have a lot of time period that you're a leader and you receive information and then you pack it in and then you emit that information and then the network confirms and then you have a block. And by the way, in Solana, the blocks aren't like these discrete steps that you might be used to from like Bitcoin or Ethereum. They're continuous. They're just always being streamed at all time frames. But there's a few problems with doing that. right you're basically giving that unilateral uh leader full control or not full control but within you're giving them a lot of privilege to be able to do things like they can just not do anything uh they can censor things uh they can they can reorder things uh depending on how it's constructed and you know generally speaking there's some encryption so that doesn't happen but they can do a lot of things and um mcp is really just a way of saying or as the name suggests having multiple concurrent of these things such that you know if one of them uh censors you there is another one or there's n set of other ones depending on where you are in the geography that are likely to include you depending on the fees and the economics and incentives at play So it's really fundamentally a way of increasing censorship resistance.

31:48And now it is certainly quite contested, I would say. So one person might say, for example, that current slot times in Solana are 400 milliseconds. And, you know, one thing that might be useful to reduce the censorship possibility is, well, just reduce that in half. such that they have less privilege than they used to. And then perhaps teams like Gito and Harmonic are working on different types of blockpacking systems. And perhaps we experiment with those. And once we do that, we take the learnings and then finally make the big change. So, and to be frank, it is a complex change. I think it introduces some moving parts that are obviously higher risk than parts that aren't moving or parts that aren't there to begin with.

32:44So I think we're still in the phase of debate, right? And Solana is very good at this where there will be a proposal and then everybody will just fight the whole time. But it's not like a tinfoil conspiracy theorist on Twitter fight. It's more like nerds that are like, no, no, no, here's the algorithm or here's the method you should use. And so I would give it a few more months of bickering back and forth until there's some optimal refinements made.

33:12Alex Thorn:Yeah. How do you think if, well, I guess, well, let's table that. Maybe I'll follow up with you in the future once there's more concreteness to the proposal. But one bickering episode I recall vividly recently in the Solana ecosystem was in regards to reducing network inflation. That was last year. There were some various proposals. I know Galaxy Research put forth a pretty small proposal, but a sort of a high level idea of using Futarki to decide what the correct inflation should be. What are your views on inflation in Solana? It does seem like staking rewards generally are going down across all of crypto.

33:52Alex Thorn:Should they? What benefit or cost might there be in the case that they do come down? So this is a very debated topic. And I think nobody has or not many people have the full picture. And so I'm actually one of the people who proposed or helped define the first proposal. But then I also made the second proposal, which then I basically stopped. And so the way it goes is if you look at the chart of Solana, and generally what you'll find is somebody will show you the market cap chart and the price chart. and they're like, oh, look, the market cap hit an all-time high, but the price is low, right? Which implies, obviously, that the tokens in circulation have increased and it's not necessarily a great monitor system.

34:40Now, most people associate this with inflation, like the printing of new coins during later schedules. However, if you look at the data, it's much more... The dominating factor is really the unlocks, where obviously Solana started not necessarily like Ethereum or Bitcoin, but more it needed venture capital funding. Kobe actually has a good piece on this where he compares like how the token models have evolved over time. And Solana was actually quite good overall. It wasn't as good as Ethereum, but it was still pretty good. And then the projects that came after it, they were just incredibly bad tokenomics in terms of the private to public kind of flows.

35:24or the returns you can expect. Now, those are most, for the most part, they are largely over. There's obviously some still remaining, but their effects will not be nearly as much as it was before. Okay, so then you look at, all right, what about the money printing itself, right? What is the set of properties that are concerning about how the network issues new tokens? And if you look at it, it's basically a curve that was copied by Cosmos and it's just decreasing over time, 15 % year over year. And it is, I think, and people will look at and think that the inflation is something like 8 % or 14%, but it's actually not.

36:10It's really much lower than that. And then you look at, okay, if I were to push forth this proposal of actually reducing the inflation, what's the best way to do it? And the proposal at that time was a relatively complex one. It was based on the staking rate, right? Like the framing is, okay, how much do you want to pay for security as a network? If such and such percent of the sole in the network is staked, then if the target is higher than the current state, then perhaps you increase the staking rate so that people are incentivized to stake. And if it's lower, then it gets lower so that you're not incentivized to stake.

36:49But that's a pretty complex mechanism, all in all. and you're not super sure how the second order effects of that might look like. I think one thing that threw institutions off was that they felt that it came as a quick proposal and that it wasn't necessarily in line with a mature monetary system. And so, you know, the proposal didn't pass. So like, this was a proposal that was backed by me, Anatoly, Multicoin, and a bunch of others. And it actually didn't pass, which, you know, ironically shows that the network is much more decentralized than even we thought, because we couldn't even get it passed um so now we're in a state where you know unlocks there's not much of them and the inflation is going down year over year and so it doesn't seem to be the biggest fire to tackle um i think there's much more interesting things you know things like quantum things like increasing bandwidth uh and all sorts of other uh changes to make yeah that

37:49Alex Thorn:That makes sense. I think, and it has the sort of urgency and the debate of it has sort of died down. I think that's right. Let's shift gears. You mentioned quantum. I know that there is plenty of interesting things to talk here in this space. There were those big papers from Oratomic and the Google Stanford team just, I think, last week or maybe the week before that suggested that the math is simpler. The complexity of a eventual quantum computer wouldn't need to be as high as previously thought to run Shor's algorithm, which would let, you know, a quantum computer brute force, ECDSA and other elliptic curve cryptographies.

38:29Alex Thorn:still not one of these things is really built that's anywhere close to doing that. So it's like the sort of software and math side got simpler, but it seems like the hardware side is still elusive. But what is your view generally on the threat that quantum poses to cryptography and thus most major blockchains? Well, I think there's two things to look at here. One is that just risk management. The risk that quantum poses is, non-linear fat tail, black swan type risk, or it's convex risk in that the benefit to preparing or the benefit to not preparing is basically zero, but the cost is basically infinite.

39:15Or I mean, it's not infinite, but it's much higher than the counterfactual. So from a basic risk management perspective, I think it obviously needs to be looked at and considered seriously. However, obviously, it is quite early. I don't think there's any reason for panic today. I think what the discussions or the new improvements should do, in theory, is communicate that, look, there must at least be a plan for this. It doesn't need to be implemented. It doesn't need to be fully ironed out. But you need to have some concrete answer in the potential case of new information entering the economy, which is that, okay, well, actually, it moved up again by another three years so that you're not in panic mode.

40:05So I think Bitcoin will be fine. I think one thing that has frustrated people, me included is the complete dismissal of it from even the realm of possibilities, right? Which kind of communicates to me that certain members have either very interesting risk management or they're not necessarily interested in actually improving these systems, right? Like when you are a big proponent of Bitcoin, you're like, well, it's a huge threat to prepare for quantum because like you know such and such people have bags in the quantum companies it's like okay now we're starting to get into a territory where perhaps the actual concept of quantum breaking this cryptography is not primary but it's more so like a game of social clout um so i think it needs to be taken quite seriously especially for things like bitcoin and zcash where basically the entire premise is being um bunker coin in some sense like it's a it's a thing that needs to hold up in the failure of other systems.

41:11Things like Bitcoin or Ethereum and Solana will have generally an easier time because they're much more nimble versus things like Bitcoin. It is just extremely decentralized, right? You have so many different actors and diverse points of view that having some rough alignment beforehand is going to prove beneficial. And I think, you know, I'm a tech accelerationist, So my thinking on this is that the timelines will keep moving up unexpectedly, but I don't think it'll ever really be too late. So it's, you know, the short answer is I think the risk is real, but the panic is overblown.

41:55Alex Thorn:Yeah, I think that's a fair approach and thought process. One of the things I saw as it relates to Solana, although I think this would be true for all of them, is Solana did a pilot with Project 11 in December to, I guess probably on a testnet, maybe you know more, but to test out some post-quantum signature schemes. and it worked. But one of the issues with the actual technology and mathematics behind solving quantum and implementing post-quantum secure cryptography is that the signatures are very large. And crypto, you know, blockchains in particular, you know, a large signature doesn't matter too much for like TLS or something you use in your web browser, but basically all of blockchains are built around signatures and them being very small.

42:44Alex Thorn:and so it would be true for Bitcoin. I know there was a 2.5 kilobyte per signature proposed by a block stream researcher for Bitcoin just last week. That's still enormously larger than the current signature scheme. In general, I think they're 20 to 40 times larger than what we're all basically using, which is some form of classical elliptic curve cryptography. And the Solana one found that, again, with the current post-quantum algorithm that they used, it would reduce the transactions per second, you know, the throughput on Solana by 90 % to implement it because they're just so large, meaning you can fit fewer transactions per slot or block.

43:25Alex Thorn:You know, does your tech accelerationism, it sounds like it makes you think, you know, you probably will get quantum computing at some point. Does it, do you have the same confidence perhaps that we'll also get, you know, a more durable and compact post-quantum cryptography that we can all use that doesn't lead to major degradation and throughput? Well, I think if such a thing occurs, then what you must do is rethink the system designed from first principles because now the assumptions have changed. If the assumptions were that signatures are cheap and small things and as a result you can pipeline the lifecycle of transactions in a certain way, perhaps now that looks different.

44:07And perhaps now there's some off-chain components and maybe there's a ZK proof that, hey, this is the signature and here's the part that's stored on-chain, right? So you would just rework with those assumptions in mind. But like, you know, let's say Quantum is 10 years away. I think by that point, Solana will have scaled to such a point where I don't see it making a huge difference in performance because there is so many low-hanging fruit. I mean, we just released, or we're in the process of releasing a new token program that'll make tokens, transfers 20x more efficient. And that's just one change, right?

44:47So I think there's a bunch of those.

44:49Alex Thorn:Yeah, I think that makes sense. I also am an optimist about this problem. I guess some Bitcoiners would call me a pessimist because I believe it exists, the potential problem. but that even something as disparate and contentious as Bitcoin development will eventually decide that the risk is high enough to justify some work here. And that also humans are smart and we eventually figure stuff out. And it used to be 20 kilobytes, now there's a 2.5 kilobyte one, et cetera, and that will probably come down over time. So I agree. I don't think people should panic. I like the way you said that before.

45:28Alex Thorn:Before we wrap, Just a couple more questions. One on AI. There was just a new story. I know you're a big user of AI, and I think everybody should be. There was a story that came out that Anthropics' new model, Mythos, is so incredibly powerful that they are guarding it and only giving early access to a select number of elite institutions, banks, mag-7 companies, specifically because it is like an order of magnitude better at penetration testing and finding vulnerabilities in code. And they said that they'd found thousands of zero-day vulnerabilities, right, which are those that are previously unknown to the developers of the software.

46:09Alex Thorn:I don't know, maybe your reaction to that, but also then one of the thoughts I had was, gosh, I wonder if they've pointed it at Bitcoin or at Zcash or at Ethereum or Solana. It sounds like they've been very responsible about this and maybe to the extent they found zero-day vulnerabilities, they've been doing responsible disclosure rather than release the model to the public. But, you know, generally, what does that make you think about the fears and power of AI, but then also, you know, what do you think the odds are that there are, you know, zero-day vulnerabilities in crypto that we're not aware of that maybe this thing is finding?

46:43Well, the odds of that are almost certainly 100%. As long as humans are writing code, there's just going to be issues in information transfer between the mine and the, let's say, the transistors. And part of me with this mythos stuff thinks there's certainly a large element of marketing to it because this is not necessarily the first time we've seen an approach like this. It seems to be a very common thing in AI to message it like this. And this was before even we got LLMs. It was all about AI safety and doomerism and this thing will destroy everything. And certainly it's extremely good at coding.

47:24and I can definitely see it discovering security issues that were previously undiscovered. Now, I don't know about thousands or, you know, things that are... I'm sure they found a few. If it were thousands, though, that seems unlikely to me that no human had found them before. Given that these systems... I mean, I worked at the big tech companies. their security teams are insane. I mean, like, it's hundreds of thousands of people, like, many people working with different contractors and all this stuff, you know, Palo Alto technologies and whatnot. So, I think, you know, the question I go to here is like, okay, well, what happens when DeepSeek gets access to this or other regions of the world, right?

48:13Like, what is going to happen then? And I have been pretty paranoid recently since, like, the drift hack and how North Koreans were involved in very elaborate social engineering. And so I have been thinking a lot about this in terms of hardening wallets. This morning I was like, it's really weird that we are storing wallet seed phrases on browsers. They're encrypted, sure, but they're Chrome extensions. That's kind of wild to me. And so I think there still needs to be a lot of growing up. And I think AI will accelerate that. The trick here is, of course, how do you survive this without breaking, right?

48:52There's a potential where you come out much stronger and anti-fragile, let's say, but then there's a place where you die. And it's like, how do you make sure you don't die? And I think everybody in the industry needs to do a very thorough job and it needs to be a very conscious effort. So Solana Foundation just released, you know, a very big program for helping teams harden their security. But I think it needs to keep happening. I think, especially like even with things like the Iran war, which is kind of a tangent, but like a lot of that was due to like the Israelis hacking the infrastructure in Iran, which is like people don't even know, like in real life, it's very easy to understand security threats because like your heart rate goes up and you're like, okay, shit, I'm going to die.

49:42But on the internet, you don't even know that you've been exposed. So So, you know, we'll see how it goes. I certainly think there's issues in blockchains, though.

49:53Alex Thorn:Yeah, it would make sense that there are. I mean, part of me hopes that, you know, with something like Bitcoin or Solana, Ethereum, like they've been around for so long with so much code available that, like, it's the biggest bug bounty in the world. And surely they've already been looking. But I do. It seems like the pace of AI development, particularly on the frontier models, it's astonishing, right? And it may even be accelerating, not even plateauing. I think your point about DeepSeek is really interesting as well. Maybe one last AI question, because that's an anthropic model. It's private.

50:29Alex Thorn:It's a paid proprietary model. But you've got Gwen and DeepSeek, and I saw Google release Gemma 4 as open source. Do you think there's a time maybe when the frontier models kind of run out of data to train on, and so they sort of plateau and then the open source models take over or catch up?

50:54Well, it's an interesting question. It's sort of, it's similar to like the operating system wars of like Linux and Windows and Mac, except now like the currency seems to be compute and how fast you can do these things. So I studied AI in school. you know, measured math. And, you know, I actually built my version of Jarvis back in the day with some deep learning algorithms. And as far as I can tell, I think there is... And I will just copy David Deutsch here, who's like the Oxford physicist who invented quantum computing. I think there's a fundamental limitation to scaling this way that like once you get to a certain level, like the improvements are just can't be exponential.

51:45when you compare it to something like the human mind and how it computes things, it uses much less resources, right? So there's something clearly different with different types of intelligence. And the way NALM works is really extremely sophisticated autocomplete. And so I think there's a lot of art and science in the contextual engineering of it and these harnesses and orchestration and stuff. And I think that's where open source will be very interesting, right? Like OpenClaw seems relatively simple, but like it was an open source, it is an open source project that the AI labs with all their funding just did not think of for some reason.

52:24Right. So I think there'll be a lot of those. But like in terms of like the pure underlying, let's say the engine, the physics of it, it's going to be very challenging for open source to compete on that front.

52:37Alex Thorn:Yeah, one of the sort of optimistic things as a big supporter of open source technology that I am, that I think about, it's almost like another analogy is like proof of work. And you have the, everyone can mine on their CPU, and then you need a PGA, and then eventually ASICs, and obviously GPUs in there too. And it results in a much more centralized big data center, you know, Bitcoin mining operation, right? It centralizes over time around big compute. And one of my hopes is that the open source models sort of catch up to the frontier model. So you actually see the reverse happen rather than, you know, giant data centers feeding clawed, you know, prompts and inputs back, outputs back to users from huge data centers is a decentralization that may occur where everybody's Mac minis.

53:23Alex Thorn:and laptops like get so good that perhaps you get like a dispersion of the compute out to the people as opposed to a forever centralization. I don't know that that would actually happen, but it does seem like it might be a trend that goes the opposite way that proof of work did one day as those model. If the, I mean, it would require the frontier models to kind of plateau in terms of their level of, of intelligence for that to happen. But I guess a man can hope, you know? Yeah, I think one idea I had that I was looking forward to back in the day was like the idea of people having their own data centers like in their garages in some sense, where you just have these stacks of hardware.

54:05Like, I mean, people already do this with gaming, right? Like with PC gaming, they do some extremely custom shit. And now that the other side isn't necessarily like, you know, elders, trolls, but it's actual intelligence, that gives them a better incentive. And perhaps there's something, you know, Deepin projects are kind of trying this, but I haven't seen a good approach yet. I think there is a world in which you have local data centers that are much more decentralized. But again, you always know the same problem that the giants have this weird capital edge over you where they can raise almost infinite amounts of capital.

54:46and partly through this marketing that they have where they're like, if you don't give us money, you're literally going to die. That's basically their strategy. And so it is hard to compete with that.

54:56Alex Thorn:Yeah. Yeah. All right. Last question here, Mert, before we wrap. You've talked about Zcash and privacy a lot. I know you've been building in the Zcash ecosystem as well. You care a lot. You've written about cipher capitalism. You're undoubtedly on the open and permissionless side of this industry, what gives you hope and excitement about the growth and adoption of permissionless blockchains going forward and with your work at Helios? I would say that, and the reason I've been talking about the cyber capitalism thing, which is just a word I made up, is that if you just look at the development of other, just of progress in general, the concept of progress, in science, in biology, and in the economy, it always happens through the same mechanism, which is the P2P voluntary exchange of pieces of knowledge.

55:51And then they're basically fighting with their environment or their peers until the best ones propagate up, right? This is how biological evolution works. This is how science works, where you publish a paper, you criticize the ideas until you have things that basically error correction reduces and you're like, okay, this is pretty approximate to reality. And capitalism is the same thing because it's the economic model of that concept of knowledge propagation. However, it has... So that is to say capitalism is an information system. However, in real information systems, you have the concept of a stable carrier.

56:38When we're talking on this call, our messages are encoded in zeros and ones and some redundancy. And that's a very high entropy piece of information. It's unstable. It contains a lot of surprises. And the way that works is that the carrying channel, the wires, the antennas themselves, need to have low entropy. They need to be very stable in how they propagate that information so the information itself doesn't get corrupted. And so that's actually a very perfect parallel to how the economy works, where like, you know, an entrepreneur built something, but then perhaps somebody prints money or perhaps there's some rent seeking happening or property rights aren't respected or like your money is surveillable.

57:23And as a result, you are coerced into not doing something. And so it's a very lossy information system. But with blockchains, you have a system that is, in premise, a giant financial computer that can be used by everybody but controlled by nobody. And so it is like the perfect evolution of that information system that you can use to evolve capitalism. And capitalism is basically the metamachinery, which is to say it's the machine that makes the other machines possible. And so if you have a world where you have private encrypted money, you have contracts being enforced by cryptography as opposed to humans.

58:02And humans will never get out of the loop fully, and they shouldn't either. But we have already run the experiment what happens when you apply cryptography to these information systems to increase the bandwidth and then reduce the noise. And crypto is really, and I haven't seen it framed this way, but I think it's the correct way. Like with the system, you can actually scale capitalism from not just America, but to the entire world. Right. When people have access to dollars or other currencies and there's free flow of capital, then the best ideas kind of propagate up to the top and more people have access.

58:33And then I think that's like the thing that's missing that. But anyways, that's what excites me. I'm not super excited about any particular vertical in crypto, but really just this meta concept of it being the rails for improving economic knowledge and its propagation.

58:52Alex Thorn:All right. Well, well said. Mert, co-founder and CEO of Helios at Mert on X. Mert, thank you so much for coming on Galaxy Brains. Of course. Thank you for having me. That's it for this week's episode of Galaxy Brains. Thank you to our guest, Mert from Helios. And our good friend, Bim Netta Bibi from Galaxy Trading. Everyone have a safe and happy weekend, and we will see you next week.

59:18Alex Thorn:Thank you for listening to Galaxy Brains, the weekly podcast from Galaxy Research. I'm Alex Thorne, head of firmwide research at Galaxy. Follow me on X at Intangible Coins. Follow Galaxy Research on X at GLXY Research. Read our written reports at galaxy.com slash research. And don't forget, if you like Galaxy Brains, to like and subscribe on your favorite podcast platforms like YouTube, Spotify, Apple Podcasts, and more. We'll see you next time.