NashQ 🦣

Joining @CelestiaOrg as a Research Engineer 5 months ago has been a game-changer. The team is amazing, every single person is a wealth of knowledge. I'm grateful for the support and encouragement I receive, my questions are always entertained.

With 1Tb/s, we solved throughput. Let's see what comes next...

We got detroit become human in real life before gta 6

@checkmatexxxxxx @nickwh8te @dankrad @_Enoch @christine_dkim @l2beat The funds that are issued on the L2 are safe, because of the Celestia model and sampling. But the funds that are issued on Ethereum and bridged up have a different trust assumption because of Blobstream. A smart contract cannot do DAS, so you have an honest majority assumption.

@nickwh8te @dankrad @_Enoch @christine_dkim @l2beat Dam I just got cooked I thought this was basically impossible I thought that risk was basically void because of Celestia model

Worth reading by @christine_dkim Don't have unrealistic expectations. L2 at the moment is such a broadly defined category that it barely means anything; L2s can even compete with Ethereum. To properly understand the positioning of a project, try to understand these properties: - does it compete with Ethereum's core business (that's DeFi at the moment) - does it use the same tech and will it contribute it - will it burn ETH or otherwise accrue value to it (e.g. buying services from Ethereum L1) These will be a much better guide to how good a project is for Ethereum than whether it's an L1 or L2 :)

Q: "Did you copy everything?" A: "Agreed different color would help."

@sreeramkannan @eigencloud 💜 @celestia is hiring engineers as well, and would be a great fit for engineers coming from Eigen imo DMs open.

@nickwh8te Your logic: We abuse "decentralized" to mean X But centralized still has X Therefore we should not call it centralized. Imo we could just be specific if rollups have X or not and call out abuse of "decentralized". It's only a semantics/optics discussion, so great bait.

Instead of calling them "centralized sequencers" we should call them "single sequencers" A single sequencer is not "centralized" because it can still preserve most of the key qualities of decentralized systems such as: - verifiability - liveness - permissionlessness It only sacrifices censorship resistance but even then it can still inherit some CR from the L1 via mechanisms like inboxes TL;DR single sequencers can still be very decentralized, so calling them "centralized" is a psyop

@bertcmiller @DistributedMarz 4. Yeah, I read the paper you linked, but it did not categorize what spam is like you did. I wondered if they have a different category of spam. Excited to see more data on this !

@bertcmiller @DistributedMarz 3. By sophisticated spam, I meant trying to outmaneuver the auction with other spam. My point was that solving this would not induce other non-spam demand on Base. For that, you would need apps that abuse these low fees. ( and yes we are both excited for them ;))

Some thoughts after reading "MEV and the Limits of Scaling". Cool results and well explained, so its a nice read. 1) Is spam good for metrics and revenue for Base so they want to keep it? Higher base fee means more rev. "A full Base block is great no matter whats inside."

@0xBreadguy @yangl1996 That means 1 transaction is 1400 bytes ! That's quite large. Do you know what information is encoded in that transaction, that would offload work on nodes syncing? Is it maybe a list of state accessed by the transaction, so you can execute them in parallel?

Bro we need more DA capacity. Yesterday. Just listened to @yangl1996 talk about how we're going to leverage EigenDA and it's clear DA is going to be up-only for performant chains. Previously, we had ~20MB/s needed for ~100k TPS. Well, we revised: ~28MB/s for ~20k TPS. 7x increase! It's because we're exploring additional data production which will offload work from our node network, allowing them to stay synced without as much work. ...and 20k is not a far-fetched number for throughput. Hyperliquid is already at ~10k TPS as a singular application. A high-TPS, general-purpose platform could double that easily.

@0xkydo Decreasing blocktime could be another solution. Also internalizing arbitrage could be something viable. x.com/NashQueue/stat…

@NashQueue Isn’t the only viable strategy here is to increase the base price for everyone?

This looks like a huge mispricing of gas. The first line of defense would be to adjust that. What is the reason for it being so mispriced? I get Solana because there they only do CU and DA is free but I am confused why that's also the case for OP-Stack Rollups.

Highly recommend reading the article yourself. Very excited to see what kind of solution they will cook up. cc @DistributedMarz @bertcmiller writings.flashbots.net/mev-and-the-li…

4) Could the reason be latency? OP chains are slower in comparison to Orbit chains meaning they are more mispriced and a target for arbitrage later in the block. For Arbitrum its mostly top-of-block. A latency race rather than blind signing? What if you just go very very fast?

@0xkydo The logical conclusion would be to price congested state differently. Straw man: have a fixed size queue of all touched state during a block. Touching a state increments it by 1. Decay over some blocks. Expensive state sits at the top of the queue and is priced accordingly.

@0xkydo Maybe. The one thing that the these arb transactions have in common is that they touch the same state. Is mostly just swaps touching the same pools for arbitrage. Meaning that you don't have congestion on the global level because blobs are free but on the local level.

@true_wondertan @grndd_systems @celestia Someone has been cooking here

I'm happy to share that we've started Grounded Systems (@grndd_systems) - our studio for brave crypto application experiments with @celestia underneath. Go directly to the quote-tweeted thread and discover what is being built, our team, vision, and looming experiments. If you'd like to learn more about my journey to get here, please continue reading. Since before I joined Celestia in 2021, I've seen blockchains as a technology that can change the world for better. I know that sounds extremely cliche but that's how naive I was :) However, the problem back then was clear that technology itself was not ready for mass scale, so I got captivated by solving the infrastructure and getting into promising L1 early. Fast forward to 2025, after years in trenches building Celestia and studying mammoths, I have a firm conviction that low-latency abundant blockspace is the solution and almost the end game. And thence Grounded Systems was born which is my little lab for experimenting with radical product ideas, utilizing all that blockspace with some novel crypto primitives to eventually find a product that would fulfill my inner ambition. Don't get me wrong, tho, I am not leaving the protocol!!! I see the Berlin Wall between core and application developers in other major L1s, and, as a founding coredev I don't want to repeat this part of history for Celestia. I believe coredevs are privileged and somewhat obligated to fill the gap between core and app development, serving as a role model for how to build native apps. Celestia's case is especially problematic, considering the endless options it provides with its full-stack control. That's basically it. Thanks for reading. If you also see blockchains more then they are currently, closely watch @grndd_systems for ideas being cooked.