x.qntx.eth @gitctrlx

Vitalik 在文中提出一种基于期权的合成资产设计，替代传统清算机制的稳定币。核心思想是将1 ETH拆分为正向（P）和反向（N）两种期权，到期时根据预言机价格决定各自ETH分配，从而避免清算。用户可通过持有深度实值期权并定期再平衡来获得稳定的指数敞口（如USD）。相比清算机制，该方案仅依赖慢速预言机，减少实时预言机的安全风险，并在极端行情下表现为缓慢的二次偏离而非突然清算。 翻译如下：learnblockchain.cn/article/26231

vitalik.eth @VitalikButerin

3 days ago

Building index-tracking assets on top of options instead of debt ethresear.ch/t/building-ind… What if the use options as the base of defi, instead of CDPs and liquidations? So instead of extreme price movements creating a sharp and global "you get liquidated" effect, instead your

211 174 1K 340K 486

See also: my criticism of the idea that ZK alone, even if well implemented, is sufficient to make "identity verification" pro-freedom: vitalik.eth.limo/general/2025/0… (And some alternative ideas for how to solve underlying goals)

Mullvad.net @mullvadnet

3 days ago

The EU age verification app is presented as “completely anonymous”. But the risk is that member states (the countries are supposed to create their own versions of the open-source EU app) use it to introduce identity verification that makes it impossible to post anonymously on

65 759 3K 254K 499

One of the many things I dislike about the style of "make AI go well" discourse from frontier AI companies is how nationalist the whole thing has gotten. In the 2010s, it was: "we're here to benefit all of humanity" In the 2020s, "we're here to benefit all of 4% of humanity" And even the Good People are buying into this frame completely😢 And they expect humanity to go along, because "come on, be a realistic adult, it's either us or Chiiina" or something like that (And on the EU side, you get "[X] with European values", which too often seems to mean "the same stuff they're doing, but with Us instead of Them in charge") Very big-dog-small-dog energy, morally speaking.

Polymarket @Polymarket

3 days ago

NEW: Bernie Sanders proposes the government take 50% of OpenAI & Anthropic to give the public a “direct ownership stake.”

598 234 4K 869K 321

数字游民 收入来源 与 地理位置 解耦 生存成本 与 固定基础设施 解耦 心理稳定性 与 外部环境一致性 解耦

有点味道了

finally found the real cappuccino

释永信在佛珠上刻有 24 个助记词对应 1.3 亿美元的比特币，和随身携带以太坊钱包的 USB 所以大师也看准了，除了比特币和以太坊之外的加密货币不值得投资

大公司死于太讲礼貌。 无数企业的衰退，正是死于对制度的过度“礼貌”， 凌晨1点17分，十万员工被一封邮件突然砸醒。 标题只有硬邦邦的三个字：生产力。 这里没有任何管理梯队的宏大叙事。 只有直接对组织架构动刀子的四条铁律。 来，学下怎么重塑一台最高效的机器。 在会场上做不出直接贡献？别待着，马上离场。 内部的某条规定毫无逻辑？别盲从，直接无视。 传个话需要绕劈六个管理层级？砍掉中间链条，点对点直线触达。 在这里，任何人都可以绕过权限直接联系任何人，不管是谁，包括马斯克自己。 庞大的组织是怎么走向平庸的？ 因为无数企业的衰退，正是死于对制度的过度“礼貌”，那些层层设防的规矩完美到了没人敢去质疑。 几百年来，人类极其迷信科层制，一层层汇报，一级级审批。我们本想用规则换取稳定，但也因为规则，组织长出了无数不解决问题的寄生节点。一个指令传递六次，执行早就变异了，这要怎么快？ 面对臃肿的体系，常规的管理者在干什么？ 在修补，在打补丁，在发明新的考核系统去管理旧的审批系统。 而马斯克的做法呢？ 全盘框选，直接“删除”。 与其耗费心智去维护、去管理庞大的系统，不如从物理意义上彻底抹除一切不必要的环节。 直接剔除脂肪，只留神经元互联。 这就是特斯拉能保持绝对统治级速度的底层逻辑：当规则成了阻力，别去适应它，粉碎它。

Introducing Claude Opus 4.8: it builds on Opus 4.7 with sharper judgment, more honesty about its own progress, and the ability to work independently for longer than its predecessors. Available today at the same price.

Say “hi” to SoFiUSD (SoFiD) 👋 The first stablecoin issued by a U.S. national bank and redeemable 1:1 for cash or cash equivalents. Rolling out now, it’s built for how money moves today: fast, flexible, 24/7.

⚡️ TODAY: Vitalik Buterin shares updates to his self-sovereign LLM setup, noting the growing intersection between Ethereum infrastructure and AI. "We should have models finetuned for Ethereum-related use cases as well."

keyboard_arrow_left Previous keyboard_arrow_right Next

Vitalik Buterin on why consortium blockchains have mostly failed “The original vision of consortium blockchains — the idea that you have 5 banks or major companies that come together and create their own chain — has been mostly a failure. I think the reason why is it ends up inheriting most of the disadvantages of centralization and most of the disadvantages of decentralization at the same time.” “The first five banks join, and they all feel like ‘Yay, we’re building a system together. We can all be part of it.’ But then once bank #6 and bank #7 and bank #29 come in, they’re joining a system where there’s already an established power structure, established participants, and basically to them it still feels like they’re joining some kind of centralized thing that’s controlled by a cartel.” “You don’t actually gain the benefits of true openness that people are looking for. You don’t have Etherscan. You don’t have a connection to an open, public network. At the same time, if you build on one of those systems, you have to figure out how to program distributed systems; you lose privacy — you might think you still have some, but the reality is you’re putting your data on a network where the only people that get to see it are you and all your closest competitors. From a privacy perspective, it actually doesn’t make much sense.” “The compromise between centralization and decentralization that actually makes a lot more sense is: you have an application and today that application is a server. You can keep your server, but instead, we’re going to add scaffolding on top to give users extra security guarantees. You put Merkle roots on chain. You put proofs on chain. And you give your users assurance that whatever is happening inside of your system is actually following the rules. So you have high scale, high performance, and you optimize for a minimal delta for existing centralized infrastructure deployment. Keep your existing infrastructure the way it is, and you just add a side car that makes the roots and the proofs.” Source: @arbitrum

SpaceX has almost finished writing V1.0 of an in-house AI training stack in C that exact-maps to 220k GB300s with 800G NICs, making heavy use of pipeline parallelism and getting as close to bare metal as possible. The potential speed improvement vs JAX for large training runs is over an order of magnitude.

Updates since then: * Deepseek v4 is out. There *is* a 2-bit quant that can run within 90 GB ( huggingface.co/antirez/deepse… ), and it works, however it's only fast on Apple hardware (I've head ~35 tok/s). On AMD, it's ~7 tok/s. IMO actually taking the effort to properly support more than one hardware manufacturer is a great example of the difference between mere "decentralized AI" and genuine "CROPS AI". I hope we can become better at this. * github.com/vbuterin/messa… also has alpha telegram support now. However, the path to adding your account is quite janky * github.com/Luce-Org/luceb… looks promising as a way to run "dense" models (eg. Qwen 27B) more efficiently. It's janky, but on my 5090 laptop it seems to be ~2x more tok/s than llama.cpp * VoxTerm (local AI recording, no third-party servers) continues to be developed github.com/dmarzzz/VoxTerm And there's a lot more projects coming on the horizon. One other thing that has been on my mind is that there's actually a lot of intersection between "CROPS ethereum access layer" and "CROPS AI". For example, we want a ZK way to make (paid) calls to remote LLMs. But if we have this, then it's just as useful for solving another problem: private RPC reads in Ethereum. Another example: application-specific finetuned LLMs. Leanstral ( mistral.ai/news/leanstral ; I get ~38 tok/s on AMD) fits into < 70 GB, but can hold its own against 1T models on writing Lean code. Things like this are a huge boon for writing more secure code ( vitalik.eth.limo/general/2026/0… ). We should have models finetuned for Ethereum-related use cases as well.

vitalik.eth @VitalikButerin

2 months ago

My self-sovereign / local / private / secure LLM setup, April 2026 vitalik.eth.limo/general/2026/0…

695 672 5K 1.3M 5K

吴说获悉，a16z crypto 表示，并非所有代币化资产都真正具备链上金融属性。目前债券是规模最大的代币化资产类别，市值约 152 亿美元，但仅约 5% 被用于 DeFi；贵金属代币情况也类似，大部分资产仅停留在链上。相比之下，再保险代币约 84% 的供应量部署于 DeFi，私募信贷则约为 33%。a16z crypto 认为，目前许多所谓“代币化”更接近于资产数字化，即仅将记录迁移至区块链，而未真正实现链上金融系统的可组合性。wublock123.com/news/a16z-cryp…

Crypto stocks. We may be headed full-on to a Snow Crash cyber-punk future with no long-term personal relationships and digital value embedded in all of us directly correlated to the value provided to a society that increasingly devalues humanity. This may be the point in time that needs to be stopped from going forward by some future being. #snowcrash

CZ：我曾怀疑 RWA 和稳定币，但它们都增长很快 币安创始人赵长鹏（CZ）@cz_binance 在 2026 年 5 月 9 日接受 Crypto In America 频道采访时表示，自己过去曾对 RWA 和稳定币持怀疑态度，但 RWA 概念出现约一年半后已快速发展，稳定币在 10 年前也不被他看好，如今已成为约 2000 亿美元规模的产业。CZ 还提到，互联网已发展数十年，但 SpaceX 仍可能成为全球最大的单一互联网服务提供商，说明成熟技术中依然可能出现新的机会，加密行业也一样。 来源：Crypto In America

Facet 联合创始人 Tom Lehman 提议将 EIP-8182 纳入以太坊 Hegota 升级。该提案拟通过协议级 shielded pool，为 ETH 与 ERC-20 引入原生隐私转账功能。方案采用 UTXO 设计，无管理员密钥、暂停机制或代理合约，并允许用户直接向现有以太坊地址或 ENS 名称进行私密转账。Lehman 表示，此举旨在解决应用层隐私方案匿名集不足的问题，为钱包与应用提供统一隐私池。（The Block）wublock123.com/news/tom-lehma…