Roman Melnikov

161 posts

Roman Melnikov

Roman Melnikov

@dimdumon

ex DagLabs / Kaspa L1 C-Arch & Vp RnD, ex Panther Protocol ZKP-Arch, Privacy, ZKPs, L1s, Blockchains

Katılım Mayıs 2021
130 Takip Edilen895 Takipçiler
Igra Labs
Igra Labs@Igra_Labs·
Loading...
Igra Labs tweet media
English
20
45
243
8K
Roman Melnikov retweetledi
Igra Labs
Igra Labs@Igra_Labs·
Galleon closed mainnet on Node V2 is ready for community node operators! Setup guide: github.com/IgraLabs/igra-… Note: configured for Galleon testnet. Mainnet configuration coming soon. What this means: decentralization and censorship resistance from day one. Igra state can be read, written, and verified on community hardware. For questions, comments and support please join our Discord: discord.gg/igralabs For the technically curious, here's what's under the hood: 1. V2 Architecture New Viaduct and Adapter components for L1-L2 bridging. Viaduct manages ATAN-to-Consensus data handover with async streaming. Adapter connects Kaspa (L1) to EVM execution layer (L2) via Engine API. Includes warm/cold start positioning and fallback mechanisms. 2. Engine Stability & Reorg Handling Critical fixes for Fork Choice Updated (FCU) unwind operations, deep reorg safety, and disk consistency. Plain state rollback, synchronous block removal, crash prevention. 3. ATAN Synchronization Complete ATAN system for blockchain state sync. Auto-import rewrite, importer service, finality period validation, consensus sync checks. 4. Transaction Pool Improvements FIFO ordering, synchronous pool clearing after FCU success, discard reorged transactions. Ensures reliability during chain reorganizations. 5. gRPC Protocol MigrationComplete migration from WebSocket/borsh to gRPC/protobuf for wallet communication. --- Test it, break it, let us know.
English
28
94
325
18K
Roman Melnikov retweetledi
Dr Martin Hiesboeck
Dr Martin Hiesboeck@MHiesboeck·
Fascinating
Bull Theory@BullTheoryio

JP MORGAN MANIPULATION of MSTR IS WAY BEYOND YOU THINK. Here's how everything was planned. Let’s start with an important name: Jim Chanos. He is one of the most well-known short sellers on Wall Street. He has built his career by shorting companies he thinks are overvalued. ➱ In May 2025, he openly said he is Long Bitcoin and Short MSTR. This helped create the idea that you can support Bitcoin but attack MicroStrategy separately. ➱ In July 2025, JPMorgan increased margin requirements for trading MSTR shares from 50% to 95%. This resulted in lower trading volume for MSTR and even triggered margin calls as investors weren't able to add more funds to meet the new requirement. ➱ In August 2025, JPMorgan released documents for a new structured product linked to BlackRock’s IBIT. So they were already preparing their own Bitcoin related product months before the MSCI issue came into the picture. ➱ On October 10, MSCI released the consultation note. It says that companies holding 50% or more of their assets in Bitcoin or digital assets may be removed from MSCI indexes if their activities resemble a digital-asset treasury. This directly affects companies like MicroStrategy. It’s also important to remember: MSCI = Morgan Stanley Capital International. It was originally created by Morgan Stanley. So the index changes have a link to a major bank now offering its own Bitcoin-related products. ➱ Just four days after the MSCI announcement, Morgan Stanley filed with the SEC for a structured product tied to IBIT. So one arm raises concerns about BTC heavy companies, and another arm releases a product that gives exposure through IBIT instead. ➱ 2 weeks ago, JPMorgan filed for its own IBIT linked structured note. ➱ Then on November 20, JPMorgan did two things on the same day: • Published the documents to sell their IBIT note • Brought back the MSCI issue (which was already 42 days old) and highlighted the index removal risk around MSTR This timing is what raised questions. This is why many traders now believe the pattern looks like this: 1. Raise doubts about BTC-heavy companies like MSTR 2. Create fear around potential index removal 3. Launch new Bitcoin-exposure products 4. Let capital shift from MSTR into IBIT-linked bank products The market has seen this behavior before. In 2017, JPMorgan CEO Jamie Dimon publicly called Bitcoin a fraud. $BTC dropped sharply. Soon after, reports showed JPMorgan clients were buying Bitcoin for their rich clients. Public comments and actual positioning do not always match. So what does all of this mean? • The MSCI announcement created the first wave of concern • The market was already fragile at that time • JPMorgan’s note came at the exact moment when MSTR and BTC were weak • More traders now believe there may be short side pressure • JPMorgan and Morgan Stanley are releasing IBIT linked products at the same time • Saylor stepped in to give clarity and stabilize the situation All of this now point to one thing: the flows and the timing around $MSTR are not random. Understanding this behavior is important, because this is how large institutions influence the market long before the retail crowd notices.

English
5
9
53
16.8K
Roman Melnikov retweetledi
Sandeep | CEO, Polygon Foundation (※,※)
This is a great formula for life. 70% build 20% write 10% fight Thanks @balajis
Balaji@balajis

For years, the conventional wisdom was to put 100% of your energy into building, and to ignore politics. The problem was that those who put 100% of their energy into fighting, particularly politicians, would then take everything you’d built. Like the video game Civilization. If you put everything into science and commerce, while the next door tribe puts it all into swordsmen and slogans, they’ll waltz over and walk over you. The temptation is then to respond in kind, by putting 100% of your resources into fighting. This does work to ward off the initial assault. But without building you become just like your violent and unproductive neighbors. Everyone becomes a violent, dumb tribesman. And technological progress grinds to a halt. This is the downward spiral that characterizes the darkest ages of history. Indeed, perhaps some day, long in the past, that violent neighboring tribe that’s trying to steal from you once fell into a similar downward spiral thanks to *their* enemies. And lost their productive capacity, and had only tribal raiding to fall back on, resulting in their current state of contagious degradation. Think Planet of the Apes. There is an answer, though. It’s 70% build, 20% write, and 10% fight. Put most energy into building, some energy into writing, and one tenth towards fighting. Think of the 10% like your defense budget. Your technological progress provides the capital and growth to fund defense. Indeed, you may end up with >10X the resources of the vicious neighboring tribes. Because they focus solely on demonizing others rather than building up themselves. And if so, 10% of your resources will match 100% of theirs. Then, even by devoting 100% of their energy to fighting, they won’t be able to steal. Your defense is too strong. So they’ll simply need to build their own stuff. As will other tribes. Now everyone gets wealthier, and has less reason to fight. And this is the upward spiral.

English
51
23
365
49.8K
Roman Melnikov retweetledi
Dr Martin Hiesboeck
Dr Martin Hiesboeck@MHiesboeck·
You may know me as a fierce critic of Solana and Ethereum roll ups. If Solana is such a success, if Etherum is so big and L2s so fast and profitable, what does one lonely academic’s voice matter eh? It’s all about the money isn’t it? Institutions of print millions into ecosystem that are immature, unable to scale and dangerously hacker-friendly. If I can’t convince you, maybe this will make you think twice. 🚨 A critical new research paper by a team led by my former colleague and accepted at ACM CCS 2025, exposes a devastating, low-cost vulnerability that threatens to freeze and halt major Layer 2 rollups and single sequencer chains. The target of these attacks should be obvious. It’s the sequencers. If our goal is decentralization eliminating the single point of failure, how can we support systems with such a critical flaw — single centralized sequencer. Where did blockchain development go so wrong? A sequencer is the piece of the puzzle that orders all transactions inside a block. And transaction ordering is the key to security and reliability of any chain. If someone jumps the queue (MEV) the system becomes skewed and unfair. If you interfere with the ordering you can suffocate the chain just like in a DDoS attack. What my fellow researchers in Hong discovered isn’t another generalized bug—it's a novel class of Denial-of-Sequencing (DoS) attacks that directly targets the most mission-critical component of every rollup: the sequencer. The researchers demonstrate how attackers can craft "non-packable" transactions that are insidious: 😐 They bypass the L2's legality check, appearing perfectly valid. 😐 They successfully reach the sequencer, consuming vital CPU and I/O resources. 😐 They are ultimately discarded after execution, but the attacker pays zero gas cost. 🥶 The sequencer's compute power is rapidly exhausted, leading to disabled block production and a complete freeze on all user transactions. This isn't theoretical. The paper specifically demonstrates the ability to halt for Polygon zkEVM end-to-end and severely slow down Arbitrum. Alarmingly, existing fuzzers and detection tools completely failed to catch these vulnerabilities. Even though not tested in detail for this paper, the vulnerability affects all EVM’s, zkEVM’s and single or permissioned sequencer chains. As rollups move into institutional and high-value transaction layers, the security and reliability of sequencers is now mission-critical. 🛡️ Get the Full Defense Strategy The paper cited is very academic, but we have an easier to read version for @UpholdOTC customers available. Actionable solutions explained include If you are working on rollup security, sequencer design, or runtime validation, this paper is not optional—it is a required read to protect your L2 ecosystem. If you are an investor — know what you are buying.
Dr Martin Hiesboeck tweet media
English
13
39
133
11.6K
eliott
eliott@eliottmea·
Didn't hyperliquid override its own oracle for Jelly Jelly markets ? Don't they have only 6 validators and hence can frontrun and mev as much as possible given geographical colocation ? Can anyone explain the impossibility of reorgs on their L1 (without access to open source code) ? hyperliquid doesn't even remotely live up to what decentralization means.
English
7
10
73
3.2K
Flood
Flood@ThinkingUSD·
If you don’t get why DEXs are crucial to Crypto’s survival look no further than 10/10. We still have ZERO context about what happened, why people were compensated and who had massive drawdowns that day. We have no idea if CEXs are still solvent. Stop guessing, choose Hyperliquid
English
83
65
859
94.2K
NVIDIA GeForce
NVIDIA GeForce@NVIDIAGeForce·
🟢 GEFORCE DAY IS BACK 🟢 To celebrate, we're giving away TWO GeForce RTX 5080 Founders Edition GPUs, signed by NVIDIA CEO Jensen Huang. Want one? Comment "GeForce Day" for a chance to WIN & stay tuned for more!
NVIDIA GeForce tweet media
English
58.4K
3.6K
47.6K
5.9M
Roman Melnikov
Roman Melnikov@dimdumon·
@CatGodSandHive @tcoratger Beside the perf itself the very interesting property is by default quantum security ( does not exist for EC based systems ).
English
1
0
0
18
CatGod
CatGod@CatGodSandHive·
@tcoratger Similar question but more specific. What's the difference with pairing friendly snarks and their performance metrics (proof size, verification time, etc.)
English
1
0
0
64
Thomas Coratger
Thomas Coratger@tcoratger·
(1/14) [ZK Whiteboard Sessions - S3M3: Lattice-based SNARKs, w/ Vadim Lyubashevsky] Let's unpack this session on lattice-based proofs. We'll explore how they work, from the foundational SIS problem to advanced techniques using polynomial rings. 🧵
English
5
7
47
2.8K
Shai ❤️ Deshe 💜 Wyborski 💙
The claim that pre-ZK bridging cannot be trustless is untrue. It is challenging and definitely can be done. The correct claim is that pre-ZK means withdrawals are delegated to another consensus and inherit it's trust assumptions. You can bridge trustlessly using any decentralized consensus mechanism as long as you believe it has the required honest (super)majority and spread. Doing so is challenging, as you need a designated, properly incentivised voting mechanism that is sufficiently active to retain decentralization (and fancy cryptography to stitch everything together securely). But it is by no means impossible. ZK simplifies things a great deal, as it can pile all the required trust on the base layer. But it is not the only way.
English
7
11
102
8.2K
Ori Newman
Ori Newman@OriNewman·
@DesheShai Do you know of any protocol where the set of bridge managers is permissionless that can achieve that?
English
2
0
13
1.2K
Denis Mashkevich
Denis Mashkevich@madenis·
Why did they call Rhodes electric “piano”? The sound is way more similar to a xylophone
English
1
0
2
311
Wei Dai
Wei Dai@_weidai·
If I told you a closed-source chain with no audits of the core components is going to be a $50B protocol, would you believe me? No?... Wrong! That's the new reality now. I wouldn't be surprised if more follow suit. Hyperliquid
English
24
6
127
18.2K
avery.apt 🇺🇸
avery.apt 🇺🇸@AveryChing·
Fastest block times in the world at just under 100ms - only on @Aptos. Single digits will come for the global trading engine.
avery.apt 🇺🇸 tweet media
English
43
36
192
38.3K
Evaldas. Code first
Evaldas. Code first@lunfardo314·
I personally designed and implemented from ground up L1 and L1+L2 (in production), so no need to ask of I know something. Just to simplify things to you, pls pay attention to your own mentioning of “trusted L1 node”. This is exactly to the point: you need to trust it! It means, it can fake anything and you will take it as truth. Or it can just be silent, or “a bit out if topic” and so on. Meanwhile crypto is about trustless. A particular node cannot be trustless, so for that we have a permissionless network of unknown set of nodes which do not trusts each other, just follow protocol and that guarantees eventual consensus. In many L2 designs it is conveniently forgotten that you have to trust your connection with L1. So, in your picture, each L2 node’s trust assumptions is trusting a particular L1 node. Hence, not trustless -> not decentralized. Even if that L1 node belongs to you, it can lie to L2 for many reasons out of your control. Because L2 protocols is not a part of L1 protocol, that is different thing. Hope that helps
English
4
0
2
194
Evaldas. Code first
Evaldas. Code first@lunfardo314·
Interesting. It seems the guys don’t take into account the trivial fact that peception of the state/sequence of a permissionless L1 is subjective for any participant, be it L1 or L2 node. Always. If you take one L2 node which perceives L1 state, it will not be deterministic until it settles. The correct solution of the problem would be to take perceptions of several L1 nodes and then run a consensus on L2 (preferably - leaderless)
Pavel Emdin@emdin

Second PR to Reth from Igra Labs. Critical for true sub-second finality support. BlockDAG reorgs within 1-2 second spans are routine. Correctly mapping these to a deterministic EVM sequence is non-trivial -- without them based rollup breaks under real BlockDAG conditions.

English
4
1
10
2.5K
Roman Melnikov
Roman Melnikov@dimdumon·
@lunfardo314 @emdin Eh - I am kaspa L1 chief architect ! So please 🙏 be polite and stop saying what you are ) To the point - run BTC node and you will find yourself in same position no ?
English
1
0
2
152
Roman Melnikov
Roman Melnikov@dimdumon·
Imagine for simple test of your ideas that L2 sits inside L1 node - still think that this is a problem ? Another point - pay attention that L2 has its own P2P network where L2 blocks are relayed - L2 nodes usually don’t use it since their own L1 provides them this block faster , but if L1 node has some problem it is switched on. Another question - do you understand that each L2 has its own trusted L1 node ?
English
1
0
5
157
Evaldas. Code first
Evaldas. Code first@lunfardo314·
@dimdumon @emdin Please, read more carefully. I am not talking about Kaspa L1. I am talking about fault tolerance of L1-L2 connection, which, when stops, stops the whole L2, does not matter how fault tolerant L1 is. I am surprised you guys miss those fundamental aspects of distributed systems
English
1
0
1
204
Denis Mashkevich
Denis Mashkevich@madenis·
This! was completely unexpected Lijiang, Yunnan
Denis Mashkevich tweet media
English
3
0
14
903