Breaking Narratives: How Code Review Beats Marketing Decks - Noah Jelich

Transcript

So, um I hope you're ready cuz this one should be a little bit of fun. I also hope I do not inadvertently get banned off Ethereum Prague for life for being maybe a little bit critical of uh of Ethereum at some points of this talk. Um So, fugazi fugazi, it's a wazi it's a woozy, it's fairy dust. It doesn't exist. It's never landed. It is no matter. It's not on the elemental chart and it is not real.

>> [sighs] >> These were the words uttered by Mark Hanna in the famous movie of The Wolf of Wall Street as he was explaining to the young cub, so to say, how to sell BS and make a profit by scamming people. This has nothing to do with the crypto industry, of course. because I do not agree with this. I think we can find alpha, I think we can find good projects and I think we can actually do it well. And I'll be answering the question how. But first a little bit about me.

Uh TLDR, I did a bunch of things. Uh currently advising Cyber Capital, which is the oldest crypto full exposure um spot only hedge fund in Europe and maybe in the world if we had last like one more. and I'm reading source like my life depends on it. Outside of work, uh I have a toddler and I really love trains, bikes, and hiking. inspired by Dr. House, of course, uh today I will tell you the story of three projects and let you make your own uh conclusions to how these may end. Legally, I'm obliged to provide a disclaimer that

Cyber is an activist hedge fund and we affect projects we invest in, hopefully guiding them to better outcomes. And this presentation is based on the state of these projects in late 2025 or even earlier for Bitcoin and may not reflect current reality. And of course, none of this constitutes investment advice. So, disclaimers out of the way, let's get into it. Bitcoin. Have you heard of Bitcoin, guys? it's the first decentralized cryptocurrency. It's based on free market ideology and it was invented in 2008 by an unknown person who published a white paper under the pseudonym Satoshi Nakamoto. Not this guy.

Uh use of Bitcoin as a currency began in 2009 with the release of its open-source implementation. If you fast forward another, let's say, 3 4 years, um Bitcoin was reaching the limits of its throughput in 2012 2013. It had one block every 10 minutes with 1 megabyte blocks. And um the block size debate that many have heard of is often framed as a technical disagreement over the single parameter, the block size.

But I consider that framing uh quite incomplete, actually. Um you see, Cyber was investing in BTC at the time and they needed to make a decision. Um what was uh visible at that time was a clash between two economic models. Constrained block space that leads to high fees and layered scaling, so layer one, layer twos, versus expanded block space aiming for base layer utility, high capacity, highly scaled individual blockchain.

More importantly, this exposed something deeper. How Bitcoin governance works in practice through control of code, repositories, and narrative. And if you were paying attention to GitHub activity, mailing lists, and who controlled merges, you could see where this was going. Here's how. You see these guys? I hope none of them are here. Uh so, by the midpoint of the debate, a majority of key contributors to Bitcoin were employed by a company called Blockstream.

Um this is a ancient, but Blockstream was a pro for-profit company whose business model centered around second layer solutions, like Lightning Network. those solutions are only valuable when the base layer becomes constrained. This leads to a very clear incentive structure. Small blocks equals congestion. Congestion leads to high fees. High fees lead to a demand for L2s, and L2s are the cake. And this was observable at the time. These guys were all eventually employed by Blockstream. Um the first and the last guy here had repo control. They could merge directly. They could write directly to the main branch

in the Bitcoin main repos, and the roadmap direction was clearly changing. Um Greg, Matt, Jorge, Mark, Patrick, uh Warren, Adam, and Peter. Um now, this was quite a version from the original idea of Bitcoin. Um Satoshi Nakamoto in a forum post wrote this. I'll let you read it while I talk. Um And Blockstream was more or less working on not scaling Bitcoin, which was against Satoshi's original wishes.

While other movements were working on increasing block size, such as the Bitcoin XT client, but this is history. And what was happening at the time, again, a while ago, uh uh were apparently um, DDoS attacks on nodes that were not running the non-scaling clients. Uh, Coinbase, for example, recently in the news for layoffs, was targeted at that time because they supported the scaling versions of Bitcoin. And they were operationally disrupted for running alternative clients.

Meanwhile, the discussion channels such as Bitcoin Talk and Reddit were taken over and moderated across uh ideological lines. Um, and protections in the XT client were soon seen as censorship because they were rejecting uh spam transactions, more or less. This all somewhat culminated in 2017 with uh the introduction of SegWit, which was an efficiency increase in terms of block packing. It was sold as a scaling initiative, but it was actually a major blocker for Blockstream to execute their products, uh mostly Lightning Network, that provide alternative scale paths.

Um, it was pushed with the promise of additional block size increases, but the bump never really happened. Uh, PRs submitted for the block size increase, which is again just a very simple parameter change at that time, no real constraints technically, uh were held up for review by the prior people and by Blockstream employees. Uh, what happened in 17 2017 was uh a locked-in outcome. Fees rose dramatically to over $50 per transaction, and um well, some celebrated this as a fee market that replaces inflation for a short time, uh meant businesses dropping support. Steam was actually considering building on Bitcoin at that time, uh user make moving to alternatives, uh

like Ethereum, and a split of the chain, uh creating Bitcoin Cash. This rise of fees was a temporary success for profitability, but over time this meant uh the opposite of induced demand, which I love as a concept in public transport, because there was no infrastructure to meet the demand for transaction, people stopped using Bitcoin for anything of a higher volume. Uh for Cyber, this meant stopping Bitcoin investment and picking higher conviction, better scaling fundamentals-based bets.

I fear the same happening for Ethereum. Personally, if it refuses to scale meaningfully. I'm hoping the new mandate will include serious scaling efforts. So, this is the story of Bitcoin scaling. And this is how Bitcoin was manipulated. The second story is about marketing presentation. So, Double Zero, not very known to many, but it is like uh Solana-based uh fundamental upgrade to blockchain networking. Uh if you are familiar with networking principles, it provides access to private backbone, which means faster, lower latency, uh fiber optic links between data

centers, between continents, etc. that are not part of the public internet that you would normally use. More or less a VPN service, but very very advanced and very fast. Um and they they have a nice pitch. Uh we reduce latency, and that's why people love us. We improve global distribution, making Solana less centralized, and we capture that value via 5% fee on validators. and I I would look at that, and I was like, "Okay, this sounds very somethingy. I wonder what's actually in the background." what was in the background?

well, the first step was an let's say market sanity check. Um Megaport is a huge company that does more or less that thing with 11,000 access points across the world Uh uh and it has a 1.5 billion weird dollars market cap. and they are offering more or less everything that that double zero could offer but they're doing it at about 100 times the scale instead of about 100 access points they have 11,000, right?

Um and at the time what they were considering this so back in September uh double zero did climb to more or less two or three billion dollars in market cap and not the Australian ones. Um which made it surprisingly overvalued. Uh and that that raised a massive red flag. Like what is the new value being created here? So I took to the technological research and I was contacting validators via email, contacting server providers and provisioners and trying to get a hold of the biggest link between Europe and Latin America called EllaLink which is a private backbone that was funded by the

EU in like 2022. and that is currently not part of the public internet because this was let's say their biggest gainer route. They were making I don't know 100 milliseconds of latency improvement compared to the old route that goes through the US and then from US to Europe. like yeah from 124 milliseconds instead of 201. so I reached out to Latitude in Wolters in Sao Paulo and asked for trace routes. Um and I cross reference this with the backbone provider spec sheet that they gave me because I was a big validator host about to come to their servers and make them some money, right? Um

and what I found was that those providers already had access to EllaLink and if I looked at my trace routes were going straight from Brazil to to Marseille, more or less, across the private link. Uh which meant that Double 0 only maybe adds value in very narrow edge cases, such as extreme networking congestion events. More recently, Megaport acquired Latitude, which is a major validator hosting provider, which means that this access to these private routes should become even more standardized and part of the package. Uh just to clarify, uh none of these companies the uh hosting providers charge anything extra for access to this private backbone.

They have it through their existing networking providers. So, the final step was uh why is anyone using this if there is no real incentive? So, I reached to out to Solana uh validators and I was asking, "Are you using Double 0? Why or why not?" So, yeah, so basically they stake plus 5% of the validator stake, which offsets the fee we have to pay them. They were being bribed. They were being bribed to use uh Double 0. This was since documented as a delegation improvement, blah blah blah program, but at the time this was not in any public documentation.

So, what looked like product market fit, because they managed to get like 40% of stake coverage with Double 0, 40% of validators running them, was actually a bunch of manipulation. They had about 300 million dollars to deploy a stake to get people to switch. Now, that doesn't mean that Double 0 can succeed. With enough capital, they can subsidize adoption, capture market share, and then become some sort of coordination layer. They do have some technical moat with uh multicast and similar stuff, but the practical reality was that they could be replicated by incumbents like Megaport overnight.

And they didn't have that much real technical edge nor voluntary adoption. The third one, hype. Hype. Hyperliquid. So, um, what is Hyperliquid? Who here thinks that Hyperliquid is, um, a fully on-chain order book?

Show hands. Have I made you too pessimistic already? I cannot sell you a story? No? Who thinks it's at least decentralized? Come on. d e c e n t r a l i I'm not going to spell that far. >> [snorts] >> You're becoming security engineers. You'll need to take antidepressants just to put money into index funds. Well, at least it's extremely fast. That that one is true. And the leverage is true.

They don't say much about the speed here, though. Uh, so, uh, the question is, but how is it fast actually? You know, the five whys? Go deeper. Uh, so the question was, is it all a lie? Um, and, uh, yeah, Hyperliquid is speedy. >> I I opened their GitHub repo. there's no implementation. What you find is, um, well, nothing. Dockerfile, license, readme, other readme, docker compose,

pub key. a link to download a binary, minimal documentation, and uh, some instructions how to run the node, but not how it works. Um that already is a major signal. You it's literally a closed source project, right? Uh you cannot verify consensus rules, execution logic, data flow. Somebody with a little bit more time than me did like a proper reverse engineering work and found some good trash inside there, but uh yeah. I just looked at what I could find with uh behavior analysis.

Um so this was definitely not a gossip-based simple infrastructure. Uh I read through this very you know, there's a lot of detail. It's a very long read me, and I found a few things. Okay, so precompiled binaries only Ubuntu 2404, sure. Um for lowest latency run the node in Tokyo, Japan. Huh? Why? And uh to avoid jailing, it is recommended to achieve 200 ms two-way latency to at least 1/3 of validators by stake.

And also validator may connect up to two non-validator peers by specifying their IPs as sentry IPs. it's a circle of validators, like 28 of them, that are all very close to each other because it's 200 ms two-way latency, and all of them can only talk to like a few other relayers. No direct contact with validator, sure. Um that's interesting note topology. this is not gossip-based architecture.

This is not very open. Um But then came the Japan part. So, um I guess everyone must be in Tokyo. How centralized could this be? Uh especially combined with the protocol level penalties for not being in Tokyo. And um it was very hard. The node requirements are insane. And I didn't want to um try too hard. So, I managed to contact Meria, who is a validator runner on Hyperledger, and get them to give me a dump of their node state and node analytics and more or less plug into their existing reputation on the network. Cuz again, you cannot ping the validators and get this data directly.

And what I found was uh when looking at ASNs, that all the validators, all the stake, and all the recent blocks are in these few data centers, mostly in this Amazon one. all of these are actually just in Japan, just in one data center cluster in Japan. Except for this one running in the UK as a penance, I suppose.

With Well, out of 500 blocks, they got two blocks, despite their stake being a little bit better than that. And they've been jailed for half the time. So, yeah, penance. >> [laughter] >> And why why is this like this? Because the consensus rules require it. It is very clear. If you want to participate, you colocate. You could drop a single 50-kg bomb on Hyperledger. So, okay, maybe it achieves most of its speed by being completely centralized within about 3 to 5 milliseconds of intra data server latency.

But what about the rest of the claims? How is it so speedy? Uh how how what about the insane TPS? It's just pumping blocks like 50 a second or something. Like the block time is nothing, and they're running at sub-millisecond processing times for everything. And this is where the real trash was found. Um understanding execution of Hyperliquid. On Ethereum, nonces are strictly sequential. If I submit a transaction with nonces six and seven, God, I did this to myself. Um the one with the nonce seven cannot be executed before the one with the nonce six is. You cannot skip nonces, and all submitted

transactions are visible and ordered. Hyperliquid has a very different uh strictly increasing uh not incrementing a nonce structure, which means that if you submit 50 transactions, you can invalidate like half of them by just executing one with a higher nonce. And there's also some time-based mechanics. This is This enables speed. Uh how? Uh they more or less suggest uh they store the 100 highest nonces per address, and you can just submit more nonces to to delete the old ones. So, you can cancel all transactions by uh by just putting more transactions in or by putting transactions with a

no-op or similar. Uh but this is also telling you something. If you can cancel them by changing how the transaction works, that means that when you're putting in orders, because here we're talking about canceling orders very fast. Hyperliquid is uh perp decks sort of situation. It is a trading terminal. Uh that means that when you put in an order, who sees that order? It's If you're canceling it by canceling the transaction, that means that create order transaction is never executed, right?

And uh that means insane MEV, probably. Uh at least for the inner circle circle of 28 validators who are actually running the mempool. Because what I found that all the transactions are mempool matched, more or less. You have this mempool, you have the orders sitting there, you can cancel them at any time, and they sort of sit as long as they can before they have to be executed, and try to be matched within the mempool. Which means that validators can MEV.

This is how the speed is is achieved, and this is how the execution of hyperliquid is achieved. So, it's all really an illusion. They They did achieve some good stuff, though, just just to say something positive. They did a very good infrasplate between the hyper core and hyper EVM. I think that's good. Because they put stuff that needs to be fast on its own execution engine, and they made an EVM one for whatever doesn't need to be fast. But a lot of what they're selling is actually very bad. And until about half a year ago, or a year ago, it was actually also permissioned, to make it even worse.

So, yeah. Bitcoin with many promises, twisted by perverse incentives. Double zero promises infra, but is really not doing that so much. And hyperliquid, the greatest hype ever, has a very very very ugly back end. That was it for me. That was the sort of research that I do. If you'd like to know more know more, I'm open to a chat. Here's my Telegram, here's my LinkedIn. Do we have any questions? All right.

Thank you very much. And we'll be moving on to next speaker quite soon. >> [music] >> Woo!