# Noah Jelich - Breaking Narratives: How Code Review Beats Marketing Decks feat. Hyperliquid - Channel: [ETHCluj Meetup](https://streameth.org/ethcluj-meetup) - Date: 2026-07-09 - Duration: 29:34 - Watch: https://streameth.org/watch/yt-L4dFn9YMERc - YouTube: https://www.youtube.com/watch?v=L4dFn9YMERc ## Description Everything today seems to be hype and empty promises, so where does one find the real edge? Learn how deep technical diligence and code analysis can cut through marketing spin, reveal structural flaws early, and find genuine innovation. ## Transcript Hello everyone. Are you ready? Do we have that end of day energy? Let's hear a who. >> Good owls. So, um yeah, this one this one should be fun. Um, everyone knows this, right? Yeah. Yeah. Yeah. That's the thing. So, Fugazi Fugazi, it's a wazy, it's a woozy, it's fairy dust. It doesn't exist. It never landed and it is no matter. It's not even on the elemental chart and it's not real. The words of Mark Hannah. >> Mark Hannah, right? >> I take corrections. I take corrections. Uh and uh and he's talking about the markets, right? He's talking about manipulating little investors, swindling them out of money and getting them to buy your uh shitty coin/ uh penny stock. And I do not agree with this. I I think there is something that is real somewhere in there. And I think there is a way to find it. But I think the way to find it is not by listening to marketers because they have one role and that is to sell sell sell. Uh so yeah uh a little bit about me. Uh TLDDR I've done a bunch of things uh >> oh of course uh ah okay the camera pans right. Um so ther I did a bunch of things. I'm currently at Cyber Capital which is the oldest crypto full exposure hedge fund in uh Europe maybe in the world if we outlive like a few more. and uh and I'm reading source like my life depends on it more or less. Outside of work, I have a toddler. I really love trains, bikes, and hikes. Uh so today uh I will tell you a few stories um of three different projects and let you make your own conclusions how to how these may end. Just a mandatory disclaimer I do have to give is that Cyber is an activist hedge fund and as such we affect projects that we invest in and we hopefully guide them to better outcomes. Uh that also means that this presentation is based on the state of projects in late 2025 may not reflect current reality and that none of this constitutes investment advice. Uh so disclaimers out the way let's get into it. We have three stories. We have Bitcoin scaling, an old one, double zero which is very niche Solana stuff and hyperlquid which is on everyone's mind. Um so about Bitcoin scaling, Bitcoin was made by not this guy uh circa what 2008 when a guy who's not this published a white paper under pseudonym of Satoshi Nakamoto. I think this guy's called Dorian actually, which is very funny. Dorian Nakamoto. Um, use of Bitcoin as a currency began in 2009 with the release of the open source implementation and by 2012 2013. Um, the limits of the throughput were being reached. One block every 10 minutes, one megabyte blocks, it was just uh, you know, too small, too small. And what started then was a block size debate. Uh Satoshi was gone, dead, whatever. And um and this blockside debate is framed as a technical disagreement over a single parameter. Uh but that framing I think is very much incomplete. Cyber was investing in Bitcoin at the time. Uh and as a conviction and fundamentals based fund needed to make a decision. So when you look at Bitcoin at the time, what you have is not a simple parameter. You have two economic models you're looking at. One is constrained block space which leads to high fees and thus layered scaling layer one layer two and the other one is expanded block space making it uh a highly utilizable base layer. Um what was exposed during this time of uh the block size debates from 2013 to 2017 was actually what happened to Bitcoin governance. How it broke down in practice uh through control of code repositories and narrative by malicious or not malicious in any case profitable actors. And if you were paying attention to GitHub activity permissions everything that's was going on you could tell what was going to happen. uh which if you are living in 2026 right now you know Bitcoin did not scale right uh and uh 2015 more or less uh the state of Bitcoin was this. So these were the key contri some of the key contributors to the Bitcoin core uh repos and these guys were all employed by a company called Blockstream. Um why does this matter? Because Blockstream is a profit for-profit company whose business model is centered around second layer solutions like a side chain for Bitcoin. These solutions are only valuable if the base layer is constrained. This leads to a incentive structure. You have small blocks which lead to congestion, congestion which leads to high fees, high fees which lead to a demand for L2s and L2s which are their their lunch, right? And uh this was all public, right? These guys had LinkedIn or whatever. And uh you could see which permissions they had in the repos. The first guy and the last guy had direct right to the main branch on the Bitcoin repo. And the road map was changing a little bit. Uh Greg, Matt, Jorge, Mark, Patrick, Warren, Adam, and Peter. Those are the guys the names of the key contributors. Uh this all was actually somewhat contradictory to Satosh's original idea. He was thinking the block size could more or less be increased that Bitcoin should scale and um and as Blockstream was working on not scaling Bitcoin uh other movements such as Bitcoin XT uh were working on increasing the block size. Uh they were unfortunately faced with literal oppression. So there were even large scale DDOS attacks targeting nodes. Uh there were reports of local like smaller IP ISPs being somewhat taken down by these. And Coinbase was targeted because they socially and operationally supported alternative Bitcoin clients that could let's say enable the upgrade to the uh to the block size. uh Reddit, Bitcoin talk were uh overtaken by moderators who did not support increasing the block size and they were moderated along ideological lines and even like technical decisions like Bitcoin XT started implementing spam protection because there were DDoS attacks constantly. Uh these were starting to be reframed as censorship. So it was quite insane. Uh this all culminated culminated in SegWit. Uh so this is an efficiency increase in terms of block packing separating the witnesses uh from the actual block data and uh yeah it's sold as a scaling initiative initially but actually it was a major blocker for blockstream to execute their products uh to provide alternative scale paths because they needed segregated witnesses and it was pushed with a promise of additional block size increases as well but those PRs that were supposed to be part of the original thing but got pushed as a separate PR um were held back in review forever until it just never happened and uh and what happened later in 2017 was uh an outcome that was quite expected. Fees rose dramatically uh in exceeding $50 per transaction on Bitcoin. uh and uh this was concerned the validation of the fee market but at the same time actually users migrated to alternatives. Businesses generally dropped Bitcoin support because it could not be used for any realistic transaction volumes and the chain split uh creating Bitcoin cash. Uh this rise in fees led to a permanent reduction in usage. Uh I really like public transport. So my chosen concept here is induced demand and it's opposite. So if you add lanes in America, you end up with 50 lanes full of traffic. If you remove lanes, you end up with two lanes full of traffic. How is that possible? So the same thing sort of uh if you add capacity to the blockchain it will be filled eventually and if you remove it it will be filled eventually like up to its limits. It will not be overflowing anymore because people will stop using it. And um today I actually do fear the same happening with Ethereum. And until recently, this was a personal point of contention, but now with the recent scaling initiatives, hopefully this this does get better. Um, Ethereum could reach 3K TPS with the improvements I'm hearing about, which is a massive increase. So, I consider that's a step in the right direction, another maybe 6x, and we're at uh the Visa scale, right? uh for cyber back in 2017. This meant actually stopping Bitcoin investment at picking higher convection, better scaling fundamentals based bets which did yield better results comparatively. Uh so that's the first story. Um the second story is about uh let's say maybe a little bit more obscure project but uh this and the following story are not historical research. They're much more current. Um and this was held uh researched in September. So um zero is some sort of fundamental upgrade to blockchain networking as they present themselves. Uh their pitch is that you have latency issues. you have uh problems getting uh if you want to make a truly high performance blockchain with a subsecond block time or sub 200 millisecond block time, you need very very low very stable latency uh to trade. Um and this can be achieved either by having everything very centralized geographically uh which is a spoiler for the next section or uh by improving access to backbone. Backbone being uh the actual cables the actual connections that you get between your uh locations when connecting. And their idea was to capture value uh for this latency improvement by charging a 5% fee on validators who use their network uh their physical network. Um and this is this is a strong claim but uh honestly I'm I'm a very pessimistic person. Uh so my first question was uh how are they lying to me? Um, is this a breakthrough or is this a repackage of existing infrastructure? And how how does this even work? Why would anyone pay this? Um, so the first thing I did was let's say a competitor analysis. There there's not really any competitors uh in the blockchain space directly, but uh let's let's do a market sanity check. Um on the left here we have Megaport. Megaport is a massive company uh that provides uh interconnected uh private backbone with low latency and uh their valuation is 1.5 billion funny dollars. um the the Australian ones. Um they have like 11,000 connection points across the globe and operate a gazillion data centers and like inter data center connections and private backbone connections and submarine cables that they are allowing providers to to give to them which they can rent out to small users. Um at the time of initial analysis um 0ero was very very highly valued like 2.5 billion despite having only about a 100 backbone connections. So that was the start. Um this seemed not quite right but okay blockchain overvalued companies whatever. Um, so this wasn't looking great, but then I was thinking, but maybe maybe just because it's blockchain, just because it's a circle, it's worth a lot more because it has this more scalable future or something like that. Uh, so I was thinking, okay, uh, so I was looking, but is anyone using this? And when you look at the validators for Salana, you see that like 40% of validators use double zero. So 40% of validators are paying a 5% fee to double zero. So it's like, okay, this is wild. Why why would they use this? So let's look at the technical improvements. They're stating high latency savings. Like um if you were to use the public internet between um between Europe and Latin America, on the public internet, your route would be let's Does this work? No. Maybe maybe I don't know how to use it. uh your route would be from Europe through uh North America to South America and this takes around 200 milliseconds. Um there is a special little cable called Ella link which was built with the support of EU funds in 2022ish and it cuts this time down to 120 milliseconds which is a massive improvement. uh the actual latency of the cable is like maybe 50 milliseconds but you know there's also local networking at all these hubs um so okay this this is a cut this a cut in the networking so I was wondering okay but is this something very unique that they provided or could I get access to link you know why why is it still $2.5 billion because llink didn't cost that much to build and that's one of their two major savings hubs Um so I reached out to major validator hosts in Sao Paulo uh which is our main hub uh for for the validators. Uh those are Latitude and Vulture. And I asked for trace routes to known servers in Europe uh to check um how does this access to this magical link cable work and uh their integration engineers were very very helpful. They provide me with backbone provider spec sheet. They provide me with trace routes and access to their consoles. And what I found was that these providers already had access to link. Um looking at the trace routes, you can see direct jumps from Sao Paulo to Marseilles uh which match the latency of the LLink cable. Uh so in practice double zero only adds a value in very narrow edge cases like where maybe these backbone providers fail because of an extreme outage event or similar though latitude has three backbone providers. Um the reality of this was came to to light only when I started doing the social engineering research. Um I started contacting validators. Uh oh yeah just uh a cool detail. So when contacting lat I'm a big node runner right I have all the capital in the world to deploy to run validators in south Paulo for some reason. Uh so social engineering is like your best tool for extracting this information take on a role. Uh and in this case uh I was asking uh some validators friendly tips on running uh a validator or joining them or figuring out their fee structure because I heard that zero is taking a cut. So I was like okay but why are you using it? Uh and uh and let me just read this uh yeah so basically they st more than 5% of the validator stake which offsets the fee we have to pay them. The effect on the validator so far is not that expense expressive because the network effect isn't achieved yet. So what's actually happening is that zero is bribing the validators by giving them stake that offsets all the fees. So funky um uh this was not documented at the time. Right now there is a public document stating that they have like an initiative u delegation initiative to scale or whatever with like $300 million to deploy and uh it it seemed that what looked like maybe a product market fit because you look at 40% of validators using it is actually just financial incentives plus some weird technical differentiation. Uh however this does not mean that Double0ero can't succeed because with enough capital they could subsidize adoption, capture the market share and actually build out something. They do have a few technological let's say advantages such as multiccast which could work but the majority of what I see is actually not a technical mode. Um, and to compare uh to some very recent events like a few months ago or maybe a bit more, Megaport actually acquired Latitude which is one of the major validator runners for a lot of crypto stuff. So Megaport could just is surely providing Latitude access to their backbone. So a lot of the market is actually already gone and it's just a matter of a few let's say targeted deals between data center providers or maybe a little bit more infrastructure to build built out because again these submarine cables tend to become public goods eventually as will this EU supported one LLink it's just private for a few years. So yeah, this was a very interesting research and thanks to uh not investing in it when it seemed so hot and taking some time to do some serious diligence, we missed uh this joyous loss that we could have had. Um so yeah, that was the the good stuff. The last story is about the the new kid on the block, next big thing, the the everything thing. Uh Hyperlid, it's hype. it has to win, right? Uh so if we look at this landing page that they have, it says here it's DC centr whatever something uh I don't know what that word means anyway. uh decentralized. It says it's transparent fully onchain order book and uh it sort of implied that it's very fast due to let's say natural means. Um and of course my my usual question is is it all a lie? And uh but first let's check what what do oh no I already I need to put this up front. I need to shift the order cuz I make people depressed and then when I ask stuff like do you think Hyperlid has globally distributed validators? Let's do a show of hands. Everyone is just looking at me like buddy I know it's all a lie already. So h but yeah globally distributed validators. Huh? Anyone hopeful? No, permissionless participation >> maybe transparent execution. >> Yeah, >> says right here. Well, we'll see about that. We'll see about that. Uh, so the first step to analyze anything is to go to the GitHub repo. Um, what is this? Only Ubuntu 2404. Why? Oh, it's just pre-ompiled binary. There's no code. For lowest latency, run the node in Tokyo, Japan. Huh. To avoid jailing, it is recommended to achieve 200 millisecond two-way latency to at least one-third of validators by stake. At least they're not slashing. And validary may may connect to up to two nonvalidator peers by specifying their IPs as sentry IPs. So they have an offiscated binary and uh minimal documentation to how it works and instructions to how to run the node. Um what this actually tells us is that Tokyo is for some reason very important. Uh and it tells us the network structure sort of. So I know there's like 30-ish validators and each validator may connect to two peers. So that tells us that this is a network with a ring structure and not another thicker relayer ring. So the only ones that actually know what's going on chain uh are the validators and the relayers. Nobody else has any gossip that they could potentially get because the transactions go straight to the relayers. Uh there's no other way to get them included. You're not going to gossip your way into a transaction hyperlid. Um the other one is this whole jailing plus recommendation to do Tokyo Japan. And uh I was very curious to how does this actually work? Then unfortunately no requirements are insane um and the nodes will refuse to connect to you. So uh I needed to get the data. Uh here I must thank Maria uh who are one of the validator runners for Hyperlid who were so nice to provide me their node logs uh and allow me to plug into the actual data pipelines and get everything I needed and everything I needed is sad. Uh so so I I was lucky enough to also go to Japan for uh some of this research and participate in ETH Tokyo and get contacts with the Japanese runners um such as uh Frack and Hyperquant. And what we're getting here is that more or less all of the validators are what you would sort of expect in a couple ASN. These couple ASN are um all in Tokyo. It's just a data center cluster within like 50 meters of each other. You could drop a questionably small bomb. Oh god. You could drop a small bomb on the data centers in Tokyo, Japan, and take out Hyperlid. God, last it was more of a joke last time I said it. I swear. Uh and uh and what does this mean? Uh well the reason why it's like this is because Hyperlid is running insanely fast and it's running like this exclusively due to the low latency. Um it could not do this at any real decentralization. Um and in practice you can see the validators uh here but the stake and recent blocks actually tell you more of a story because of the last 500 blocks generally 499 or 98 will be produced in the data center cluster in Tokyo, Japan. Um Hyperlid due to its um rules cannot decentralize actually um while scaling for sure. Um, so let's add another double another double E here to the top and go into the final piece um of the problems with Hyperlid that I found while researching it and that's uh nons behavior. Um so I was okay hyperlquid achieves its speed via uh v via this mechanism um of collocation mainly but that's not enough. There's there's centralized chains that are not as fast as hyperlquid. So there is surely something else achieving allowing them to go 100x the speed of base or something like that. And uh that's something is another key detail in how nonsense operate. So on ethereum to prevent double spends we have nonses and uh these nons are uh sequential to get your number seven executing you got to get the number six done first. Uh and you can replace number six if it's stuck but that's more or less how it works. You get a few transactions per block. Sure, you can, but they all need to be valid. Um, and all submitted transactions are thus visible and ordered. Maybe if they all execute in a block, you could have it done a little bit differently, a little bit more suddenly, but generally you can order your transactions. Um, and you cannot skip a transaction really, not quite not very easily. On hyperlquid nons are only strictly increasing and the highest 100 nons are stored per address by the validators. Um what does this mean? Um users submit a large batch of transactions and to cancel a transaction they replace it with a noop and to and when it comes to execution these transaction actually sit in the mele a bit longer. So when we're looking at that last claim that it's transparent, it's not actually transparent because if these transactions are not executed and are sitting in the mele and these are orders that we're making, uh it's an order book and there's so many of them. Uh that means that and we can skip some of them very simply. That means that there is a problem here actually. Um because who knows the real state of the order book. When you go to an explorer, you see an order book with the transactions that were actually executed on chain. Create order, delete order. But the way Hyperlid suggests you use it is not by creating and deleting orders, but by creating orders and replacing them with no ops. But if you're deleting the order by replacing with a noop, that means you're not deleting the order that was visible on chain. You're deleting a transaction that was only ever in the meool. And that is how 95 to 99% of the liquidity on hyperlquid operates visible only to the validators running the mempool. Um so this this is a major issue um of hyperlquid. makes me seem like uh like utopia uh because it's it's a more or less a memple focused system with hidden order flow and no visibility, no transparency. Uh so saying it's fully on chain is uh is actually insane. So uh those were my insights. Uh thank you for coming to my TED talk. Bitcoin perverse incentives double zero well uh mafia behavior if I can say that not as in any serious way and uh hyperlquid which uh which is a lot of stuff hidden behind uh a very very thick curtain. Thank you very much for coming and if you'd like to know more, feel free to ping me on Telegram, LinkedIn, wherever.