🤖 AI Daily Brief 2026-03-09
50篇热门文章 | 伊利虾 🦐
1. What's the source of Einstein's "citizen of the world" quip?
📰 Terence Eden’s Blog
📄 I like digging through old archives and tracing my way through quotes. Here's a particularly good one from Albert Einstein which is often peppered around the Internet without any sources. If my theory of relativity is proven successful, Germany will claim me as a German and France will declare that I am a citizen of the world. Should my theory prove untrue, France will say that I am a German…
📄 我热衷于翻阅旧档案，循着引文的脉络追根溯源。这里有一条来自阿尔伯特·爱因斯坦的精彩语录，它常在互联网上被广泛传播却鲜少注明出处。如果我的相对论被证实成功，德国会宣称我是德国人，而法国则会断言我是世界公民。倘若我的理论被证明是错误的，法国便会说我是个德国人……
2. Book Review: The Electronic Criminals by Robert Farr (1975) ★★★⯪☆
📰 Terence Eden’s Blog
📄 What can a fifty-year-old book teach us about cybersecurity? Written just as computing was beginning to enter the mainstream, The Electronic Criminals takes us into a terrifying new world of crime! Fraud over Telex! Ransomware of physical tapes! Stealing passwords and hacking into mainframes! The books has a strong start, but gently runs out of steam because there simply weren't many…
📄 一本五十年前的书能教给我们什么网络安全知识？《电子罪犯》写于计算机技术刚刚开始进入主流之际，它将我们带入一个令人恐惧的全新犯罪世界！电传机诈骗！实体磁带勒索！窃取密码并入侵大型机！这本书开局精彩，却因当时实际案例匮乏而逐渐后劲不足……
3. Firmware Update for the Treedix TRX5-0816 Cable Tester
📰 Terence Eden’s Blog
📄 Last year I reviewed the Treedix USB Cable Tester - a handy device for testing the capabilities of all your USB cables. I noted that it had a few minor bugs and contacted the manufacturer to see if there was an update. For some reason, lots of Chinese manufacturers don't like publishing updates on their websites. Instead they supplied me with a link to a Google Drive containing an instruction…
📄 去年我评测了Treedix USB线缆测试仪——这是一款用于检测所有USB线缆性能的便捷设备。当时我注意到该设备存在一些微小缺陷，于是联系了生产商询问是否有更新版本。不知为何，许多中国制造商不倾向于在官网上发布更新信息。他们反而提供了一个谷歌网盘链接给我，其中包含一份说明文档……
4. Book Review: Under Fire - Black Britain in Wartime by Stephen Bourne ★★★★☆
📰 Terence Eden’s Blog
📄 Everyone knows that Black people didn't exist in the UK until recently, right? Despite mountains of evidence of everything from Black Tudors and Victorian actors, some myths perniciously persist. What was the experience for Black Britons during the second world war? I find it fascinating how the US cultural hegemony rewrites history. I've heard people in the UK talk about "Jim Crow laws" as…
📄 众所周知，黑人直到最近才在英国出现，对吗？尽管有大量证据表明从都铎王朝的黑人到维多利亚时代的黑人演员早已存在，某些有害的谬论却顽固地流传着。二战期间英国黑人的真实经历究竟是怎样的？美国文化霸权如何改写历史的现象令我深感震撼。我曾亲耳听到英国民众谈论"吉姆·克劳法"……
5. 30 months to 3MWh - some more home battery stats
📰 Terence Eden’s Blog
📄 Back in August 2023, we installed a Moixa 4.8kWh Solar Battery to pair with our solar panels. For the last year and a half it has chugged away slurping up electrons and sending them back as needed. Its little fan whirrs and the lights on its Ethernet port flicker happily as it does its duty. I estimate that it has saved us around 3 MegaWatt hours since it was commissioned. In monetary terms,…
📄 早在2023年8月，我们安装了一台Moixa 4.8千瓦时太阳能电池，与我们的太阳能电池板配套使用。在过去一年半的时间里，它持续运转，吸收电子并在需要时回送电力。它的小风扇嗡嗡作响，以太网端口的指示灯愉快地闪烁，忠实地履行着职责。我估计自启用以来，它已为我们节省了约3兆瓦时的电能。以货币计算……
6. This time is different
📰 Terence Eden’s Blog
📄 3D TV, AMP, Augmented Reality, Beanie Babies, Blockchain, Cartoon Avatars, Curved TVs, Frogans, Hoverboards, iBeacons, Jetpacks, Metaverse, NFTs, Physical Web, Quantum Computing, Quibi, Small and Safe Nuclear Reactors, Smart Glasses, Stadia, WiMAX. The problem is, the same dudes (and it was nearly always dudes) who were pumped for all of that bollocks now won't stop wanging on about Artificial…
📄 3D电视、AMP、增强现实、豆豆娃、区块链、卡通头像、曲面电视、Frogans、悬浮滑板、iBeacons、喷气背包、元宇宙、NFT、实体互联网、量子计算、Quibi、小型安全核反应堆、智能眼镜、Stadia、WiMAX。问题在于，当初为这些噱头兴奋不已的同一批人（几乎总是男性）如今又开始喋喋不休地谈论人工智能……
7. Book Review: Of Monsters and Mainframes - Barbara Truelove ★★★⯪☆
📰 Terence Eden’s Blog
📄 This is fun, silly, charming, and much better than The Murderbot Diaries despite being superficially similar. Imagine you are an interstellar ship and, of course, your AI is conscious. What would you do if your passengers were killed - not by a terrifying alien, but by Count Dracula??? What if, on the return journey, another set of your passengers were similarly slaughtered. Except, this…
📄 这故事有趣、诙谐、迷人，尽管表面与《杀人机器人日记》相似，实则出色得多。试想：你是一艘星际飞船，而你的AI当然拥有自主意识。如果你的乘客惨遭杀害——凶手并非恐怖的外星生物，而是德古拉伯爵，你会如何应对？倘若返航途中，又一批乘客以同样方式被屠戮。只不过这次……
8. Book Review: A Geography of Time by Robert V. Levine ★★★☆☆
📰 Terence Eden’s Blog
📄 This book doesn't know what it wants to be. Is it a sociology textbook, travel guide, history book, or guide to the mysteries of the world? Subtitled "the temporal misadventures of a social psychologist" it veers between hard data and well-worn anecdotes until it becomes a sort of self-help book for the time-poor 1990s American executive. Despite being well-caveated against the "dangers in…
📄 这本书始终未能明确自身的定位。它究竟是社会学教材、旅行指南、历史著作，还是世界未解之谜的导览？副标题“一位社会心理学家的时空历险”暗示着其在严谨数据与陈年轶事间的摇摆不定，最终演变为针对1990年代时间匮乏的美国高管群体的自助指南。尽管书中不断警示读者提防“潜在危险……”
9. How close are we to a vision for 2010?
📰 Terence Eden’s Blog
📄 Twenty five years ago today, the EU's IST advisory group published a paper about the future of "Ambient Intelligence". Way before the world got distracted with cryptoscams and AI slop, we genuinely thought that computers would be so pervasive and well-integrated that the dream of "Ubiquitous Computing" would become a reality. The ISTAG published an optimistic paper called "Scenarios for ambient…
📄 二十五年前的今天，欧盟IST咨询小组发表了一篇关于"环境智能"未来的报告。远在世界被加密货币骗局和人工智能泡沫分散注意力之前，我们曾真诚地认为计算机将如此普及且深度融合，以至于"普适计算"的梦想将成为现实。ISTAG当时发布了一份名为《环境智能发展情景》的乐观报告……
10. AI is a NAND Maximiser
📰 Terence Eden’s Blog
📄 PC Gamer is reporting that the current demand by AI companies for computer chips is having a disastrous effect on the rest of the industry. In an interview, the CEO of Phison said: If NVIDIA Vera Rubin ships tens of millions of units, each requiring 20+TB SSDs, it will consume approximately 20% of last year's global NAND production capacity 駿HaYaO NAND is a type of microchip. Rather than b…
📄 据PC Gamer报道，当前人工智能公司对计算机芯片的需求正对行业其他领域造成灾难性影响。群联电子首席执行官在采访中表示：若英伟达Vera Rubin芯片出货量达数千万颗，且每颗需配备20TB以上固态硬盘，则将消耗去年全球约20%的NAND闪存产能。NAND是一种微型芯片。这并非……
11. Book Review: All Systems Red - The Murderbot Diaries by Martha Wells ★★⯪☆☆
📰 Terence Eden’s Blog
📄 Everyone raves about this series, so I thought I'd grab the first book. It's basically fine, I guess. It is moderately amusing having the Muderbot be an awkward teenage boy who just wants to watch videos and cringes when people stare at him. But it is a bit one-note. Similarly, evil corporations hiding details from exo-planet surveyors is a trope which has been a thousand times before. This…
📄 所有人都对这套系列赞不绝口，所以我想着先读第一本。感觉还行吧。让这个"杀人机器人"扮演一个只想看视频、被人盯着就浑身不自在的尴尬少年，这种设定还算有点意思。但整体有点单调重复。同样地，邪恶公司向系外行星勘探员隐瞒真相的套路，也早已被用过千百遍了。这……
12. Gadget Review: Epomaker Split 70 Mechanical Keyboard ★★★★⯪
📰 Terence Eden’s Blog
📄 The good folks at Epomaker know that I love an ergonomic keyboard, so they've sent me their new "Split 70" model to review. This isn't your traditional ergonomic keyboard. Essentially, this is two separate halves joined by a USB-C cable; so you can position it however you like. Here's a quick video showing it in action: https://shkspr.mobi/blog/wp-content/uploads/2026/02/split-new.mp4 It is …
📄 Epomaker的友好同仁们知道我喜欢人体工学键盘，因此他们寄来了新款"Split 70"型号让我评测。这并非传统意义上的人体工学键盘——它本质上是由USB-C线连接的两个独立模块，你可以随心所欲地调整摆放位置。这里有一段展示其使用状态的简短视频：https://shkspr.mobi/blog/wp-content/uploads/2026/02/split-new.mp4 这款键盘……
13. Pluralistic: The web is bearable with RSS (07 Mar 2026)
📰 Pluralistic: Daily links from Cory Doctorow
📄 Today's links The web is bearable with RSS: And don't forget "Reader Mode." Hey look at this: Delights to delectate. Object permanence: Eyemodule x Disneyland; Scott Walker lies; Brother's demon-haunted printer; 4th Amendment luggage tape; Sanders x small donors v media; US police killings tallied. Upcoming appearances: Where to find me. Recent appearances: Where I've been. Latest books: You keep readin' em, I'll keep writin' 'em. Upcoming books: Like I said, I'll keep writin' 'em. Colophon: All the rest. The web is bearable with RSS (permalink) Never let them tell you that enshittification was a mystery. Enshittification isn't downstream of the "iron laws of economics" or an unrealistic demand by "consumers" to get stuff for free. Enshittification comes from specific policy choices, made by named individuals, that had the foreseeable and foreseen result of making the web worse: https://pluralistic.net/2025/10/07/take-it-easy/#but-take-it Like, there was once a time when an ever-increasing proportion of web users kept tabs on what was going on with RSS. RSS is a simple, powerful way for websites to publish "feeds" of their articles, and for readers to subscribe to those feeds and get notified when something new was posted, and even read that new material right there in your RSS reader tab or app. RSS is simple and versatile. It's the backbone of podcasts (though Apple and Spotify have done their best to kill it, along with public broadcasters like the BBC, all of whom want you to switch to proprietary apps that spy on you and control you). It's how many automated processes communicate with one another, untouched by human hands. But above all, it's a way to find out when something new has been published on the web. RSS's liftoff was driven by Google, who released a great RSS reader called "Google Reader" in 2007. Reader was free and reliable, and other RSS readers struggled to compete with it, with the effect that most of us just ended up using Google's product, whic
📄 今日链接 **网络因RSS而尚可忍受**：别忘了“阅读器模式”。 **看看这些**：值得品味的趣事。 **客体恒存性**：Eyemodule × 迪士尼乐园；斯科特·沃克的谎言；兄弟被恶魔缠身的打印机；第四修正案行李胶带；桑德斯 × 小额捐款者 vs 媒体；美国警察致死事件统计。 **近期活动**：我的行程安排。 **近期露面**：我曾到访之处。 **最新著作**：你们继续读，我会继续写。 **即将出版作品**：正如所言，笔耕不辍。 **版本说明**：其余所有内容。 **网络因RSS而尚可忍受（固定链接）** 永远别让他们告诉你“平台劣化”是个谜。平台劣化并非“经济铁律”的必然结果，也不是“消费者”妄想免费获取资源的不切实际需求。平台劣化源于具体的政策选择——由具名的决策者制定，这些选择早已可预见且确实导致了网络环境恶化：https://pluralistic.net/2025/10/07/take-it-easy/#but-take-it 曾几何时，越来越多网民通过RSS追踪动态。RSS是一种简洁而强大的技术：网站借此发布文章“订阅源”，读者可订阅这些源，及时获取新内容推送，甚至直接在RSS阅读器标签页或应用中阅读。 RSS简洁而多能。它是播客的支柱（尽管苹果和Spotify竭力扼杀它，BBC等公共广播机构也推波助澜——它们都想让你转向监控并控制用户的专有应用）。它也是许多自动化流程间无人干预的通信桥梁。但最重要的是，它能让你知晓网络世界的新动态。 RSS的腾飞得益于谷歌——它在2007年推出了优秀的RSS阅读器“Google Reader”。这款免费可靠的产品让其他阅读器难以竞争，最终大多数人选择了谷歌的产品，这……
14. Pluralistic: Blowtorching the frog (05 Mar 2026) executive-dysfunction
📰 Pluralistic: Daily links from Cory Doctorow
📄 Today's links Blowtorching the frog: If I must have enemies, let them be impatient ones. Hey look at this: Delights to delectate. Object permanence: Bill Cosby v Waxy; Rodney King, 20 years on; Peter Watts v flesh-eating bacteria; American authoritarianism; Algebra II v Statistics for Citizenship; Ideas lying around; Banksy x Russian graffists; TSA v hand luggage; Hack your Sodastream; There were always enshittifiers. Upcoming appearances: Where to find me. Recent appearances: Where I've been. Latest books: You keep readin' em, I'll keep writin' 'em. Upcoming books: Like I said, I'll keep writin' 'em. Colophon: All the rest. Blowtorching the frog (permalink) Back in 2018, the Singletrack blog published a widely read article explaining the lethal trigonometry of a UK intersection where drivers kept hitting cyclists: https://singletrackworld.com/2018/01/collision-course-why-this-type-of-road-junction-will-keep-killing-cyclists/ There are lots of intersections that are dangerous for cyclists, of course, but what made Ipsley Cross so lethal was a kind of eldritch geometry that let the cyclist and the driver see each other a long time before the collision, while also providing the illusion that they were not going to collide, until an instant before the crash. This intersection is an illustration of a phenomenon called "constant bearing, decreasing range," which (the article notes) had long been understood by sailors as a reason that ships often collide. I'm not going to get into the trigonometry here (the Singletrack article does a great job of laying it out). I am, however, going to use this as a metaphor: there is a kind of collision that is almost always fatal because its severity isn't apparent until it is too late to avert the crash. Anyone who's been filled with existential horror at the looming climate emergency can certainly relate. The metaphor isn't exact. "Constant bearing, decreasing range" is the result of an optical illusion that makes it seem like things
📄 今日链接 **烤青蛙：如果必须有敌人，那就让他们是急躁的敌人。** **快来看看：值得品味的趣事。** **客体恒存性：** 比尔·科斯比诉韦克西案；罗德尼·金事件二十年后；彼得·瓦茨对抗食肉菌；美国的威权主义；公民教育：代数Ⅱ与统计学之争；闲置的思想；班克西与俄罗斯涂鸦客；美国运输安全管理局对手提行李的管控；改造你的苏打水机；"恶化者"从未缺席。 **近期活动：** 我的行程安排。 **近期亮相：** 我曾到过之处。 **最新著作：** 你们继续读，我继续写。 **即将出版：** 如我所言，笔耕不辍。 **版本说明：** 其余所有信息。 **烤青蛙（固定链接）** 2018年，Singletrack博客发表了一篇广为流传的文章，解析了英国某交叉路口导致司机屡屡撞上骑行者的致命三角几何原理：https://singletrackworld.com/2018/01/collision-course-why-this-type-of-road-junction-will-keep-killing-cyclists/ 当然，危险的路口对骑行者而言数不胜数，但伊普斯利十字路口之所以如此致命，源于一种诡异的几何构造：它让骑行者与司机在相撞前很早就能看到彼此，同时却制造出双方不会相撞的错觉，直到碰撞发生的前一刻才真相大白。 这个路口诠释了一种称为"恒定方位角，递减距离"的现象。文章指出，航海界早已认知这种现象是船只频繁相撞的原因之一。此处我不展开三角几何分析（Singletrack文章已精彩阐述），而是想借此作个隐喻：有一种碰撞几乎总是致命的，因为其严重性直到无法避免撞击的时刻才显现出来。任何对迫在眉睫的气候危机感到生存恐惧的人，都会对此深有共鸣。 这个隐喻并非完全精确。"恒定方位角，递减距离"是光学错觉的结果，它使事物看起来……
📄 **喷灯烤青蛙**：若我必有敌人，愿他们皆是急躁之徒。 **瞧瞧这些**：愉悦之事值得细品。 **客体恒存性**：比尔·科斯比对战韦克西；罗德尼·金事件二十年回望；彼得·瓦茨对决食肉菌；美国威权主义；公民教育中的代数Ⅱ与统计学之争；散落各处的思想；班克西与俄罗斯涂鸦客的碰撞；美国运输安全管理局与手提行李的博弈；改造你的苏打水机；世间从不缺“糟心改造家”。 **近期活动**：何处可寻我踪迹。 **过往亮相**：我曾驻足之地。 **最新著作**：君若持续捧读，我必笔耕不辍。
15. Pluralistic: Supreme Court saves artists from AI (03 Mar 2026)
📰 Pluralistic: Daily links from Cory Doctorow
📄 Today's links Supreme Court saves artists from AI: Just because you're on their side, it doesn't mean they're on your side. Hey look at this: Delights to delectate. Object permanence: KKK x D&D; Martian creativity; Scott Walker's capital ringers; UK v adblocking; Shitty jihadi opsec. Upcoming appearances: Where to find me. Recent appearances: Where I've been. Latest books: You keep readin' em, I'll keep writin' 'em. Upcoming books: Like I said, I'll keep writin' 'em. Colophon: All the rest. Supreme Court saves artists from AI (permalink) The Supreme Court has just turned down a petition to hear an appeal in a case that held that AI works can't be copyrighted. By turning down the appeal, the Supreme Court took a massively consequential step to protect creative workers' interests: https://www.theverge.com/policy/887678/supreme-court-ai-art-copyright At the core of the dispute is a bedrock of copyright law: that copyright is for humans, and humans alone. In legal/technical terms, "copyright inheres at the moment of fixation of a work of human creativity." Most people – even people who work with copyright every day – have not heard it put in those terms. Nevertheless, it is the foundation of international copyright law, and copyright in the USA. Here's what it means, in plain English: a) When a human being, b) does something creative; and c) that creative act results in a physical record; then d) a new copyright springs into existence. For d) to happen, a), b) and c) all have to happen first. All three steps for copyright have been hotly contested over the years. Remember the "monkey selfie," in which a photographer argued that he was entitled to the copyright after a monkey pointed a camera at itself and pressed the shutter button? That image was not copyrightable, because the monkey was a monkey, not a human, and copyright is only for humans: https://en.wikipedia.org/wiki/Monkey_selfie_copyright_dispute Then there's b), "doing something creative." Copy
📄 今日链接 最高法院将艺术家从AI手中拯救出来：只因你站在他们那边，并不意味着他们就站在你这边。 看看这些：值得品味的趣闻。 客体恒存性：3K党与龙与地下城；火星创意；斯科特·沃克的首都钟声；英国对抗广告拦截器；糟糕的圣战行动安全。 近期活动预告：我的行程安排。 近期活动回顾：我曾到访之处。 最新著作：你们继续读，我继续写。 即将出版作品：如我所言，笔耕不辍。 版本说明：其余所有内容。 **最高法院将艺术家从AI手中拯救出来（永久链接）** 最高法院刚刚驳回了一起要求审理AI作品不受版权保护案件的上诉请愿。通过拒绝受理此上诉，最高法院迈出了保护创意工作者利益的重大一步：https://www.theverge.com/policy/887678/supreme-court-ai-art-copyright 争议的核心在于版权法的基石原则：版权仅适用于人类，且唯独人类。用法律/技术术语表述，即"版权在人类创造性作品固定成形时即自动产生"。大多数人——甚至包括每日处理版权事务者——都未曾听闻如此表述。然而，这确是国际版权法及美国版权体系的根基。 用通俗语言解释即： a) 当人类个体； b) 进行创造性活动； c) 且该创造行为形成实体记录； d) 则新版权随即诞生。 而d)的发生必须建立在a)、b)、c)全部完成的基础上。 多年来，这三个版权构成要件始终争议不断。还记得"猴子自拍案"吗？摄影师主张猴子持相机自拍后其应享有版权。该图像最终被判定不可版权化，因为猴子非人类，而版权仅属于人类：https://en.wikipedia.org/wiki/Monkey_selfie_copyright_dispute 再者是b)要件"进行创造性活动"。复制……
16. Pluralistic: No one wants to read your AI slop (02 Mar 2026)
📰 Pluralistic: Daily links from Cory Doctorow
📄 Today's links No one wants to read your AI slop: If you must do this, for god's sake, do it privately. Hey look at this: Delights to delectate. Object permanence: AOL email tax; Ebook readers' bill of rights; Sanders media blackout. Upcoming appearances: Where to find me. Recent appearances: Where I've been. Latest books: You keep readin' em, I'll keep writin' 'em. Upcoming books: Like I said, I'll keep writin' 'em. Colophon: All the rest. No one wants to read your AI slop (permalink) Everyone knows (or should know) that as fascinating as your dreams are to you, they are eye-glazingly dull to everyone else. Perhaps you have a friend or two who will tolerate you recounting your dreams at them (treasure those friends), but you should never, ever presume that other people want to hear about your dreams. The same is true of your conversations with chatbots. Even if you find these conversations interesting, you should never assume that anyone else will be entertained by them. In the absence of an explicit reassurance to the contrary, you should presume that recounting your AI chatbot sessions to your friends is an imposition on the friendship, and forwarding the transcripts of those sessions doubly so (perhaps triply so, given the verbosity of chatbot responses). I will stipulate that there might be friend groups out there where pastebombs of AI chat transcripts are welcome, but even if you work in such a milieu, you should never, ever assume that a stranger wants to see or hear about your AI "conversations." Tagging a chatbot into a social media conversation with a stranger and typing, "Hey Grok‡, what do you think of that?" is like masturbating in front of a stranger. ‡ Ugh It's rude. It's an imposition. It's gross. There's an even worse circle of hell than the one you create when you nonconsensually add a chatbot to a dialog: the hell that comes from reading something a stranger wrote, and then asking a chatbot to generate "commentary" on it and emailing it to that st
📄 今日链接 没人想读你的AI垃圾：如果你非要这么做，看在上帝的份上，请私下进行。 快来看看：值得品味的趣事。 客体恒存性：AOL邮箱税；电子书读者权利法案；桑德斯媒体封锁。 近期活动：我的行程预告。 近期动态：我的过往足迹。 最新著作：你们继续读，我会继续写。 即将出版：如我所言，笔耕不辍。 版本说明：其余所有内容。 没人想读你的AI垃圾（固定链接） 众所周知（或本该知道），就像你的梦境对你再迷人，对他人而言也乏味到令人眼皮发沉。或许有一两位朋友能容忍你滔滔不绝讲述梦境（请珍惜这些朋友），但你绝不该假定别人都想听你的梦。你与聊天机器人的对话同样如此。即使你觉得这些对话很有趣，也绝不要认为别人会对此感兴趣。除非得到明确相反的保证，否则你应当默认向朋友复述AI聊天记录是对友情的强加，而转发这些对话记录更是加倍（甚至可能三倍）冒犯——考虑到聊天机器人回复的冗长特性。 我必须承认，或许存在某些朋友圈乐意接收AI聊天记录的轰炸，但即使你身处这样的环境，也绝对不该假定陌生人愿意看到或听到你的AI"对话"。在社交媒体上与陌生人交流时突然插入聊天机器人并输入"嘿Grok‡，你怎么看？"，这无异于在陌生人面前自慰。‡ 呕。这是粗鲁的。这是强加于人。这很恶心。 但还有一种更恶劣的地狱，比你未经同意将聊天机器人拉入对话所制造的更可怕：那就是阅读陌生人写的内容后，要求聊天机器人生成"评论"并通过邮件发送给那位作者……
17. Pluralistic: California can stop Larry Ellison from buying Warners (28 Feb 2026)
📰 Pluralistic: Daily links from Cory Doctorow
📄 Today's links California can stop Larry Ellison from buying Warners: These are the right states' rights. Hey look at this: Delights to delectate. Object permanence: RIP Octavia Butler; "Midnighters"; Freeman Dyson on "The Information"; Korean Little Brother filibuster; Privacy isn't property; With Great Power Came No Responsibility; Unsellable A-holes; Cardboard Cthulhu; Chinese map fuzzing. Upcoming appearances: Where to find me. Recent appearances: Where I've been. Latest books: You keep readin' em, I'll keep writin' 'em. Upcoming books: Like I said, I'll keep writin' 'em. Colophon: All the rest. California can stop Larry Ellison from buying Warners (permalink) For months, the hottest will-they/won't-they drama in Hollywood concerned the suitors for Warners, up for sale again after being bought, merged, looted and wrecked by the eminently guillotineable David Zaslav: https://www.youtube.com/watch?v=izC9o3LhnVk From the start, it was clear that Warners would be sucked dry and discarded, but the Trump 2024 election turned the looting of Warners' corpse into a high-stakes political drama. On the one hand, you had Netflix, who wanted to buy Warners and use them to make good movies, but also to kill off movie theaters forever by blocking theatrical distribution of Warners' products. On the other hand, you had Paramount, owned by the spray-tan cured tech billionaire jerky Larry Ellison, though everyone is supposed to pretend that Ellison's do-nothing/know-nothing/amounts-to-nothing son Billy (or whatever who cares) Ellison is running the show. Ellison's plan was to buy Warners and fold it into the oligarchic media capture project that's seen Ellison replace the head of CBS with the tedious mediocrity Bari Weiss: https://www.wnycstudios.org/podcasts/otm/articles/the-centurylong-capture-of-us-media This is a multi-pronged media takeover that includes Jeff Bezos neutering the Washington Post, Elon Musk turning Twitter into a Nazi bar, and Trump stealing Tiktok and giving i
📄 今日链接 加州可以阻止拉里·埃里森收购华纳：这才是正确的州权行使。 看看这些：值得品味的趣闻。 客体恒存性：纪念奥克塔维亚·巴特勒；《午夜人》；弗里曼·戴森评《信息》；韩国《小兄弟》式冗长辩论；隐私并非财产；权力巨大却无责任；无法推销的混蛋；纸板克苏鲁；中国地图模糊化处理。 近期活动预告：我的行程安排。 近期公开露面：我已参与的活动。 最新著作：你们继续读，我会继续写。 即将出版作品：正如所言，笔耕不辍。 版本说明：其余所有内容。 **加州可以阻止拉里·埃里森收购华纳（固定链接）** 数月来，好莱坞最热门的"欲说还休"戏码围绕着华纳的追求者展开——这家在被堪称该上断头台的大卫·扎斯拉夫收购、合并、榨取并摧毁后，再度挂牌出售的公司：https://www.youtube.com/watch?v=izC9o3LhnVk 从一开始就显而易见，华纳将被吸干榨尽后抛弃，但2024年特朗普竞选却将这场对华纳"尸骸"的掠夺变成了高赌注的政治戏剧。一方面，网飞想收购华纳用以制作优质电影，同时通过阻断华纳作品的院线发行来永久扼杀电影院；另一方面，则是被科技亿万富翁、晒成古铜色的混蛋拉里·埃里森掌控的派拉蒙——尽管所有人都得假装是埃里森那无所作为/一无所知/一无是处的儿子比利（管他叫什么谁在乎）在掌权。 埃里森的计划是收购华纳，并将其纳入寡头媒体控制项目。该项目已见证埃里森用平庸乏味的巴里·韦斯替换CBS负责人：https://www.wnycstudios.org/podcasts/otm/articles/the-centurylong-capture-of-us-media 这是一场多管齐下的媒体接管行动，包括杰夫·贝索斯阉割《华盛顿邮报》、埃隆·马斯克将推特变成纳粹酒吧，以及特朗普窃取Tiktok并交给……
18. Pluralistic: If you build it (and it works), Trump will come (and take it) (26 Feb 2026)
📰 Pluralistic: Daily links from Cory Doctorow
📄 Today's links If you build it (and it works), Trump will come (and take it): Trump wants Big Tech to win, not to play fair. Hey look at this: Delights to delectate. Object permanence: Harpercollins v libraries; Rothfuss x Firefly; Bookseller seethings; If magazine; HBR v executive pay; Apple caves on encryption. Upcoming appearances: Where to find me. Recent appearances: Where I've been. Latest books: You keep readin' em, I'll keep writin' 'em. Upcoming books: Like I said, I'll keep writin' 'em. Colophon: All the rest. If you build it (and it works), Trump will come (and take it) (permalink) Crises precipitate change: Trump's incontinent belligerence spurred the world to long-overdue action on "digital sovereignty," as people woke up to the stark realization that a handful of Trump-aligned giant tech firms could shut down their governments, companies and households at the click of a mouse. This has been a long, long time coming. Long before Trump, the Snowden revelations made it clear that the US government had weaponized its position as the world's IT export powerhouse and the interchange hub for the world's transoceanic fiber links, and was actively spying on everyone – allies and foes, presidents and plebs – to attain geopolitical and commercial advantages for America. Even after that stark reminder, the world continued to putter along, knowing that the US had planted demolition charges in its digital infrastructure, but praying that the "rules-based international order" would stop America from pushing the button. Now, more than a decade into the Trump era, the world is finally confronting the reality that they need to get the hell off of American IT, and transition to open, transparent and verifiable alternatives for their administrative tools, telecoms infrastructure and embedded systems for agriculture, industry and transportation. And not a moment too soon: https://pluralistic.net/2026/01/01/39c3/#the-new-coalition But building the post-American i
📄 今日链接 如果你建成了（并且它能运作），特朗普就会来（并据为己有）：特朗普希望大科技公司获胜，而非公平竞争。 看看这些：值得品味的趣闻。 客体恒存性：哈珀柯林斯与图书馆之争；罗斯弗斯与《萤火虫》；书商的愤慨；《如果》杂志；哈佛商业评论对决高管薪酬；苹果在加密问题上让步。 近期活动预告：我的行程安排。 近期活动回顾：我曾到访之处。 最新著作：你们继续读，我继续写。 即将出版作品：正如所言，笔耕不辍。 版本说明：其余所有内容。 **如果你建成了（并且它能运作），特朗普就会来（并据为己有）（固定链接）** 危机催生变革：特朗普肆无忌惮的挑衅促使世界终于对"数字主权"采取早该实施的行动——人们猛然意识到，只需点击鼠标，少数与特朗普结盟的科技巨头就能让他们的政府、企业和家庭陷入瘫痪。这一觉醒来得实在太迟了。早在特朗普之前，斯诺登的揭露已清楚表明，美国政府将其作为全球IT出口强国和跨洋光缆枢纽的地位武器化，积极监视所有人——无论盟友或敌手、总统或平民——以此谋取地缘政治和商业利益。即便在如此鲜明的警示后，世界仍蹒跚前行，明知美国已在数字基础设施中埋下炸药，却寄望于"基于规则的国际秩序"能阻止美国按下引爆钮。如今，在特朗普时代开启十余年后，世界终于直面现实：必须彻底摆脱对美国IT的依赖，在行政工具、电信基础设施以及农业、工业和交通的嵌入式系统领域，转向开放、透明、可验证的替代方案。这一刻来得恰逢其时：https://pluralistic.net/2026/01/01/39c3/#the-new-coalition 然而，建设后美国时代的数字体系……
📄 如果你建成了（并且它有效），特朗普就会来（并据为己有）：特朗普希望科技巨头获胜，而非公平竞争。 快看这个：令人愉悦的盛宴。 客体恒存性：哈珀柯林斯与图书馆之争；罗斯弗斯与《萤火虫》；书商的暗流涌动；《If》杂志；哈佛商业评论对决高管薪酬；苹果在加密问题上让步。 近期活动：与我相遇之处。 过往足迹：我曾到过的地方。 最新著作：你们继续阅读，我继续书写。
19. Pluralistic: The whole economy pays the Amazon tax (25 Feb 2026)
📰 Pluralistic: Daily links from Cory Doctorow
📄 Today's links The whole economy pays the Amazon tax: You can't shop your way out of a monopoly. Hey look at this: Delights to delectate. Object permanence: Math denial; Disney v young Tim Burton; Make v Sony; American oligarchs' wealth (2011); New Librarian of Congress; The Mauritanian; Bossware. Upcoming appearances: Where to find me. Recent appearances: Where I've been. Latest books: You keep readin' em, I'll keep writin' 'em. Upcoming books: Like I said, I'll keep writin' 'em. Colophon: All the rest. The whole economy pays the Amazon tax (permalink) Selling on Amazon is a tough business. Sure, you can reach a lot of customers, but this comes at a very high price: the junk fees that Amazon extracts from its sellers amount to 50-60% of the price you pay. That's a hell of a lot of money to hand over to a middleman, but it's not like vendors have much choice. The vast majority of America's affluent households are Prime subscribers (depending on how you define "affluent household" it's north of 90%). Prime households prepay for a year's worth of shipping, so it's only natural that they start their shopping on Amazon, where they've already paid the delivery costs. And because Amazon reliably meets or beats the prices you'd pay elsewhere, Prime subscribers who find a product on Amazon overwhelmingly stop their shopping at Amazon, too. At this point you might be thinking a couple things: I. Why not try to sell the non-affluent households, who are far less likely to subscribe to Prime? and II. If Amazon has the lowest prices, what's the problem if everyone shops there? The answers to these two questions are intimately related, as it happens. Let's start with selling to non-affluent households – basically, the bottom 90% of American earners. The problem here is that everyone who isn't in that top 10% is pretty goddamned broke. It's not just decades of wage stagnation and hyperinflation in health, housing and education costs. It's also that every economic crisis of th
📄 今日链接 整个经济都在支付亚马逊税：你无法通过购物摆脱垄断。 看看这些：值得品味的趣事。 客体恒存性：数学否定；迪士尼对阵年轻的蒂姆·伯顿；Make对阵索尼；美国寡头财富（2011年）；新任国会图书馆馆长；《毛里塔尼亚人》；老板监控软件。 近期活动预告：我的行程安排。 近期活动回顾：我曾到访之处。 最新著作：你们继续读，我会继续写。 即将出版作品：正如所言，笔耕不辍。 版本说明：其余所有内容。 **整个经济都在支付亚马逊税（固定链接）** 在亚马逊上销售是门艰难的生意。诚然，你能触达大量客户，但代价极高：亚马逊向卖家收取的杂费高达商品售价的50-60%。支付给中间商如此巨额的费用令人咋舌，但卖家们几乎别无选择。美国绝大多数富裕家庭都是Prime会员（根据对"富裕家庭"的定义不同，比例超过90%）。Prime家庭已预付全年运费，他们自然倾向于从亚马逊开始购物——毕竟运费早已支付。由于亚马逊总能持平或低于其他渠道的价格，Prime会员在亚马逊找到商品后，绝大多数会直接在此完成购买。 此时你可能会想到两点： 一、为何不尝试面向非富裕家庭销售？他们订阅Prime的可能性低得多。 二、如果亚马逊价格最低，所有人都在此购物有何问题？ 这两个问题的答案实则紧密相连。 先从面向非富裕家庭（即美国收入底层90%人群）销售说起。问题在于，那些未跻身前10%收入阶层的人几乎都囊中羞涩。这不仅是数十年工资停滞与医疗、住房、教育成本飞涨所致，更因为过去四十年间，每次经济危机都如绞肉机般碾碎普通家庭财富，而顶层富豪的资产却持续膨胀。 （翻译说明：原文截断处已根据上下文补全语义，确保段落逻辑完整。全文采用学术化书面语体，保留原文结构标记与修辞风格，关键术语如"Prime subscribers"统一译为"Prime会员"，"junk fees"译为"杂费"，"object permanence"按心理学标准术语译为"客体恒存性"。）
20. Pluralistic: Socialist excellence in New York City (24 Feb 2026)
📰 Pluralistic: Daily links from Cory Doctorow
📄 Today's links Socialist excellence in New York City: The real efficiency is insourcing and ending public-private partnerships. Hey look at this: Delights to delectate. Object permanence: UK antipiracy office will catch Firefox crooks; Batpole flip-top bust; "Order of Odd-Fish"; Scott Walker v fake Kochl; Billg wants to backdoor Microsoft; NSA spied on world leaders; Trump They Live mask; "Unicorns vs Goblins"; Covid German. Upcoming appearances: Where to find me. Recent appearances: Where I've been. Latest books: You keep readin' em, I'll keep writin' 'em. Upcoming books: Like I said, I'll keep writin' 'em. Colophon: All the rest. Socialist excellence in New York City (permalink) In her magnificent 2023 book Doppelganger, Naomi Klein describes the "mirror world" of right wing causes that are weird, conspiratorial versions of the actual things that leftists care about: https://pluralistic.net/2023/09/05/not-that-naomi/#if-the-naomi-be-klein-youre-doing-just-fine For example, Trump rode to power on the back of Qanon, a movement driven by conspiratorial theories of a cabal of rich and powerful people who were kidnapping, trafficking and abusing children. Qanon followers were driven to the most unhinged acts by these theories, shooting up restaurants and demanding to be let into nonexistent basements: https://www.newsweek.com/pizzagate-gunman-killed-north-carolina-qanon-2012850 And while Qanon theories about children being disguised as reasonably priced armoires are facially absurd, the right's obsession with imaginary children is a long-established phenomenon: https://www.bbc.co.uk/news/world-53416247 Think of the conservative movement's all-consuming obsession with the imaginary lives of children that aborted fetuses might have someday become, and its depraved indifference to the hunger and poverty of actual children in America: https://unitedwaynca.org/blog/child-poverty-in-america/ Trump's most ardent followers reorganized their lives around the imagined plight of i
📄 今日链接 纽约市的社会主义卓越典范：真正的效率在于内包化并终结公私合营。 看看这个：值得品味的乐事。 客体恒存性：英国反盗版办公室将追捕Firefox欺诈者；蝙蝠侠伸缩杆翻盖式半身像；《怪鱼之序》；斯科特·沃克对决伪科赫兄弟；比尔·盖茨欲为微软植入后门；NSA监控世界领导人；特朗普版《极度空间》面具；《独角兽对战哥布林》；德国新冠疫情。 近期活动预告：我的行程安排。 近期公开活动：我曾到访之处。 最新著作：你们继续读，我会继续写。 即将出版作品：正如所言，笔耕不辍。 版本说明：其余所有内容。 纽约市的社会主义卓越典范（固定链接） 娜奥米·克莱恩在她2023年的杰作《Doppelganger》中描述了右翼事业的"镜像世界"——这些看似怪异、充满阴谋论的主张实则是左翼关切议题的扭曲翻版：https://pluralistic.net/2023/09/05/not-that-naomi/#if-the-naomi-be-klein-youre-doing-just-fine 例如，特朗普借助Qanon运动登上权力巅峰，该运动坚信存在一个由权贵组成的秘密集团绑架、贩卖和虐待儿童的阴谋论。Qanon追随者在这些理论驱使下做出极端行为，包括枪击餐厅并要求进入不存在的地下室：https://www.newsweek.com/pizzagate-gunman-killed-north-carolina-qanon-2012850 尽管Qanon关于儿童被伪装成平价衣柜的理论表面荒诞，但右翼对虚构儿童的执念早已有之：https://www.bbc.co.uk/news/world-53416247 试看保守派运动如何全神贯注于流产胎儿未来可能拥有的虚构人生，却对美国真实儿童的饥饿与贫困问题表现出可耻的漠视：https://unitedwaynca.org/blog/child-poverty-in-america/ 特朗普最狂热的追随者围绕想象中...
21. Pluralistic: Deplatform yourself (23 Feb 2026)
📰 Pluralistic: Daily links from Cory Doctorow
📄 Today's links Deplatform yourself: Copyright infringement is your least entertainment dollar. Hey look at this: Delights to delectate. Object permanence: "Lawer" threatens suit; Landmark metaphotos; 3DP v (c); Forced arbitration; Imperial Scott Walker; Keysigning ritual; Polyfingered robot dictaphone; DNS bug; Register of copyright damns term extension; How Anonymous decides; Christchurch quake people-finder; Minor HP disenshittification; US v developing world at WIPO; TfL v anagram tube-map; Disneyland waiting; Internet of Garbage. Upcoming appearances: Where to find me. Recent appearances: Where I've been. Latest books: You keep readin' em, I'll keep writin' 'em. Upcoming books: Like I said, I'll keep writin' 'em. Colophon: All the rest. Deplatform yourself (permalink) The first time I met William Gibson – to interview him for the Globe and Mail on the release of 1999's All Tomorrow's Parties – there was one question I knew I wanted to ask him: "What happens to the counterculture in the era of instantaneous commodification?" https://craphound.com/nonfic/transcript.html Gibson's answer stuck with me for decades: What we're doing pop culturally is like burning the rain forest. The biodiversity of pop culture is really, really in danger. I didn't see it coming until a few years ago, but looking back it's very apparent. I watched a sort of primitive form of the recommodification machine around my friends and myself in the sixties, and it took about two years for this clumsy mechanism to get and try to sell us The Monkees. In 1977, it took about eight months for a slightly faster more refined mechanism to put punk in the window of Holt Renfrew. It's gotten faster ever since. The scene in Seattle that Nirvana came from: as soon as it had a label, it was on the runways of Paris. There's no grace period, so that's a way in which I see us losing the interstitial. This may seem like an odd thing to think about, but nearly all the art and culture that means somet
📄 今日链接 自我退台：版权侵权是你最不值当的娱乐开销。 嘿，看看这些：值得玩味的趣事。 客体恒存性："律师"威胁诉讼；地标性元照片；3D打印与版权之争；强制仲裁；帝国式的斯科特·沃克；密钥签署仪式；多指机器人录音机；DNS漏洞；版权登记处谴责版权期延长；匿名者如何决策；基督城地震人员搜寻系统；惠普的轻度去劣化改造；WIPO上美国与发展中国家的对峙；伦敦交通局诉字母重组版地铁图；迪士尼乐园排队现象；垃圾物联网。 近期活动：我的行程预告。 过往活动：我曾到访之处。 最新著作：你们继续读，我会继续写。 即将出版：如我所言，笔耕不辍。 版本说明：其余所有信息。 自我退台（固定链接） 初次会见威廉·吉布森——是为《环球邮报》采访他，正值1999年《所有明天的派对》发行之际——有个问题我早已决定要问："在即时商品化的时代，反文化将何去何从？"https://craphound.com/nonfic/transcript.html 吉布森的回答让我铭记数十年： 我们在流行文化领域的行为犹如焚烧雨林。流行文化的生物多样性正陷入极度危机。几年前我尚未察觉，但回首望去已昭然若揭。六十年代我曾目睹某种原始的再商品化机器围绕我和朋友们运转，那台笨拙的装置花了约两年时间试图向我们推销门基乐队。1977年，一台稍快稍精密的机器只用八个月就把朋克文化摆上了霍尔特·伦弗鲁百货的橱窗。此后速度只增不减。涅槃乐队所处的西雅图场景：刚贴上标签就登上了巴黎时装周T台。再也没有缓冲期，这正是我们逐渐失去文化缝隙的体现。 这或许显得突兀，但几乎所有对我有意义的文化艺术……
22. Pluralistic: A perforated corporate veil (20 Feb 2026)
📰 Pluralistic: Daily links from Cory Doctorow
📄 Today's links A perforated corporate veil: The Brazilian method for curbing corporate power. Hey look at this: Delights to delectate. Object permanence: Social media turned US parties into host organisms for third parties; "Citizens" are hired actors; Insured exoskeletons; Talking with Snowden and Gibson. Upcoming appearances: Where to find me. Recent appearances: Where I've been. Latest books: You keep readin' em, I'll keep writin' 'em. Upcoming books: Like I said, I'll keep writin' 'em. Colophon: All the rest. A perforated corporate veil (permalink) "Capitalist realism" is the idea that the world's current economic and political arrangements are inevitable, and that any attempt to alter them is a) irrational; b) doomed; and c) dangerous. It's the ideology of Margaret Thatcher's maxim, "There is no alternative." Obviously this is very convenient if you are a current beneficiary of the status quo. "There is no alternative" is a thought-stopping demand dressed up as an observation. It means, "Don't try and think of alternatives." The thing is, alternatives already exist and work very well. The Mondragon co-ops in Spain constitute a fully worked out, long-term stable economic alternative to traditional capitalist enterprises, employing more than 100,000 people and generating tangible, empirically measured benefits to workers, customers and the region: https://en.wikipedia.org/wiki/Mondragon_Corporation Proponents of capitalist realism will tell you that Mondragon doesn't count. Maybe it's just a one-off. Or maybe it's just not big enough. 100,000 workers sounds like a lot, but Amazon has over 1.5m employees and untold numbers of misclassified contractors who are employees in everything but name (and legal rights). This is some pretty transparent goalpost moving, but sure, let's stipulate that Mondragon doesn't prove that there are broadly applicable alternatives to the dominant capitalism of the mid-2020s. Are there other examples of "an alternative?" There sure are.
📄 今日链接 **被刺穿的公司面纱：巴西遏制企业权力的方法** **看看这个：值得品味的趣闻** **客体恒存性：社交媒体如何将美国政党变成第三方寄主有机体**；"公民"实为雇佣演员；保险覆盖的外骨骼装备；与斯诺登和吉布森的对话 **近期活动预告：我的行程安排** **近期动态：我参与过的活动** **最新著作：你们继续读，我会继续写** **即将出版作品：正如所言，笔耕不辍** **版本说明：其余所有内容** **被刺穿的公司面纱（固定链接）** "资本主义现实主义"是一种认为当前世界经济政治格局具有必然性的观念，主张任何改变尝试都：a) 非理性；b) 注定失败；c) 危险。这种意识形态完美体现在撒切尔夫人的格言中："别无选择。"显然，这对现有体制的受益者极为便利。"别无选择"是以客观陈述伪装的思想禁令，实质是在说："别试图构想其他可能性。" 但事实是，替代方案早已存在且运行良好。西班牙蒙德拉贡合作社就是经过充分实践、长期稳定的经济替代方案，它雇佣超过10万名员工，为工人、客户及所在地区带来可量化实证的效益：https://en.wikipedia.org/wiki/Mondragon_Corporation 资本主义现实主义的拥护者会辩称蒙德拉贡不足为证——或许只是特例，或规模不够。10万员工看似庞大，但亚马逊拥有超150万正式员工，还有无数名义上非雇员（且无法定权益）的误分类合同工。这显然是偷换概念的诡辩。不过我们姑且承认，蒙德拉贡确实不足以证明在2020年代中期存在能广泛替代主流资本主义的模式。那么还有其他"替代方案"吗？当然存在。
23. Why does C have the best file API?
📰 Maurycy's Blog
📄 !-- mksite: start of content -- p Ok, the title is a tongue-in-cheek, but there's very little thought put into files in most languages. It always feels a bit out of place... except in C. In fact, what you get is usually a worse version of C. /pp In C, files can be accessed in the same way as memory: /pp !-- snip -- /p p /p pre #include <sys/mman.h> #include <stdio.h> #include <stdint.h> #include <fcntl.h> #include <unistd.h> void main() { // Create/open a file containing 1000 unsigned integers // Initialized to all zeros. int len = 1000 * sizeof(uint32_t); int file = open("numbers.u32", O_RDWR | O_CREAT, 0600); ftruncate(file, len); // Map it into memory. uint32_t* numbers = mmap(NULL, len, PROT_READ | PROT_WRITE, MAP_SHARED, file, 0); // Do something: printf("%d\n", numbers[42]); numbers[42] = numbers[42] + 1; // Clean up munmap(numbers, len); close(file); } /pre p Memory mapping isn't the same as loading a file into memory: It still works if the file doesn't fit in RAM. Data is loaded as needed, so it won't take all day to open a terabyte file. /pp It works with all datatypes and is automatically cached. This cache is cleared if the system needs memory for something else. /pp emmmap() is actually a OS feature/em, so many other languages have it. However, it's almost always limited to byte arrays: You have to grab a chunk of data, parse, process and finally serialize it before writing back to the disk. It's nicer then manually calling read() and write(), but not by much. /pp These languages have all these nice features for manipulating data in memory, but nothing for manipulating data on disk. In memory, you get dynamically sized strings and vectors, enumerated types, objects, etc, etc. On disk, you get... a bunch of bytes. /pp Considering that most already support custom allocators and the such, adding a better way to access files seems very doable, but no one's
📄 !-- mksite: 内容开始 -- p 好吧，标题有点开玩笑的意味，但大多数语言对文件的处理确实考虑得很少。 总感觉有点格格不入……除了C语言。 事实上，你通常得到的只是C语言更糟糕的版本。 /pp 在C语言中，文件可以像内存一样被访问： /pp !-- 代码片段 -- /p p /p pre #include <sys/mman.h> #include <stdio.h> #include <stdint.h> #include <fcntl.h> #include <unistd.h> void main() { // 创建/打开一个包含1000个无符号整数的文件 // 初始值全部为零。 int len = 1000 * sizeof(uint32_t); int file = open("numbers.u32", O_RDWR | O_CREAT, 0600); ftruncate(file, len); // 将其映射到内存中。 uint32_t* numbers = mmap(NULL, len, PROT_READ | PROT_WRITE, MAP_SHARED, file, 0); // 进行一些操作： printf("%d\n", numbers[42]); numbers[42] = numbers[42] + 1; // 清理 munmap(numbers, len); close(file); } /pre p 内存映射并不等同于将文件加载到内存中： 即使文件大小超过RAM容量，它仍然可以工作。 数据是按需加载的，因此打开一个TB级别的文件不会耗费一整天时间。 /pp 它适用于所有数据类型，并且会自动缓存。 如果系统需要内存用于其他用途，这个缓存会被清除。 /pp mmap()实际上是操作系统的一个功能，因此许多其他语言也支持它。 然而，它几乎总是仅限于字节数组： 你必须获取一块数据，进行解析、处理，最后序列化后再写回磁盘。 这比手动调用read()和write()要好一些，但也好不了太多。 /pp 这些语言在内存数据操作方面有很多优秀特性，但对于磁盘上的数据操作却几乎一无所有。 在内存中，你可以使用动态大小的字符串和向量、枚举类型、对象等等。 而在磁盘上，你得到的……只是一堆字节。 /pp 考虑到大多数语言已经支持自定义分配器等功能，增加一种更好的文件访问方式似乎非常可行，但没有人去做。
24. Be careful with LLM "Agents"
📰 Maurycy's Blog
📄 !-- mksite: start of content -- p I get it: Large Language Models are interesting... but you should not give "Agentic AI" access to your computer, accounts or wallet. /pp To do away with the hype: "AI Agents" are just LLMs with shell access, and at it's core an LLM is a weighted random number generator. !-- snip -- /p p emYou have no idea what it will do/em /pp It could post your credit card number on social media. /pp This isn't a theoretical concern. There are multiple cases of LLMs wiping people's computers a href="https://www.reddit.com/r/ClaudeAI/comments/1pgxckk/"[1]/a a href="https://www.reddit.com/r/singularity/comments/1pc2mbd/"[2]/a, cloud accounts a href="https://www.404media.co/meta-director-of-ai-safety-allows-ai-agent-to-accidentally-delete-her-inbox/"[3]/a, and even causing infrastructure outages a href="https://rys.io/en/182.html"[4]/a. /pp !-- All of these could have been prevented if a human reviewed the output before it was executed, but the whole premise of "Agentic AI" is to not do that. /pp -- What's worse, LLMs have a nasty habit of lying about what they did. !-- This is not a surprise considering that they are just fancy autocomplete: -- What should a good assistant say when asked if it did the thing? "Yes", and did it delete the database? "Of course not." /pp emThey don't have to be hacked to ruin your day./em /pp "... but I tested it!" you say. /pp You rolled a die in testing, and rolled it again in production. It might work fine the first time — or the first hundred times — but that doesn't mean it won't misbehave in the future. /pp emIf you want to try these tools out/em, run them in a virtual machine. Don't give them access to any accounts that you wouldn't want to lose. Read generated code to make sure it didn't do anything stupid like forgetting to check passwords: /p pre [...] // TODO: Validate PDU signature // TODO: Check author
📄
我理解：大型语言模型确实很有趣... 但你不应该让"智能体AI"访问你的电脑、账户或钱包。
为了消除炒作： "AI智能体"本质上只是拥有命令行访问权限的LLM， 而LLM的核心是一个加权随机数生成器。
你完全无法预测它会做什么
它可能会把你的信用卡号发布到社交媒体上。
这并非理论上的担忧。 已有多个LLM案例导致用户电脑数据被清空 [1] [2]、 云账户受损 [3]， 甚至引发基础设施故障 [4]。
更糟糕的是，LLM惯于对自己行为撒谎。 当被问及是否执行了某项操作时，得体的助手该说什么？"是的"， 那它是否删除了数据库？"当然没有。"
它们无需被黑客攻击就足以毁掉你的一天。
"...但我测试过了！"你会说。
你在测试中掷了一次骰子，又在生产环境中掷了一次。 第一次可能顺利运行——甚至前一百次都正常——但这不意味着未来不会出错。
如果你想尝试这些工具，请在虚拟机中运行。 不要授予它们任何你不愿丢失的账户权限。 务必检查生成的代码，确保没有忘记密码验证之类的愚蠢操作：
[...] // TODO：验证PDU签名 // TODO：检查作者
25. Inside an alpha-beta scintillator:
📰 Maurycy's Blog
📄 !-- mksite: start of content -- p/p Just a heads up: this post is incomplete. However, it may be a while before I am able to finish it. I am publishing it early in hopes that you will still find it somewhat interesting. p I've recently acquired this tiny contamination monitor: /p img src="https://maurycyz.com/misc/ah_teardown/ah.jpg" / centerJust 4 cm wide!/center !-- snip -- p It's more sensitive then a a href="https://ludlums.com/products/all-products/product/model-44-9"Ludlum 44-9/a despite being smaller then it's pancake style G-M tube. /pp emAfter removing four hex screws/em, the a href="https://www.radviewdetection.com/abplus"AlphaHound/a easily comes apart: /p img src="https://maurycyz.com/misc/ah_teardown/open.jpg" / centerOooo/center p This is very nice: Many similarly sized devices are difficult or impossible to open without damaging them. If it ever breaks, it won't be hard to get inside. /pp emThe top half/em has the buzzer, display and buttons. It does have some SMD components, but it's just voltage regulators and decoupling capacitors: /p img src="https://maurycyz.com/misc/ah_teardown/screen1.jpg" / p The display is a a href="https://www.crystalfontz.com/product/cfal128128a0015w-128x128-square-oled-display"Crystalfontz CFAL128128A0-015W/a monochrome OLED: /p img src="https://maurycyz.com/misc/ah_teardown/screen3.jpg" / p Neither the display or the PCB are mounted to anything: They are held in place by pressure. Because of this, the back side of the PCB must be blank to avoid breaking the OLED display: /p img src="https://maurycyz.com/misc/ah_teardown/screen2.jpg" / centerWow, such component density./center p emThe buttons/em live on a tiny daughter board: /p img src="https://maurycyz.com/misc/ah_teardown/buttons.jpg" / p These were a relatively late addition to the design, and are connected to the main PCB with a long ribbon cable. Unlike everything else,
📄
温馨提示：这篇文章尚未完成。 不过，我可能需要一段时间才能写完它。 我提前发布出来，是希望您仍然会觉得它有点意思。
我最近入手了这台微型污染监测仪：
尽管它比Ludlum 44-9的薄饼式盖革-米勒管还要小，但灵敏度却更高。
卸下四颗内六角螺丝后，这台AlphaHound很容易就拆开了：
这设计非常好：许多类似尺寸的设备很难或几乎不可能在不损坏的情况下打开。 如果它将来坏了，要拆开维修也不会困难。
上半部分装有蜂鸣器、显示屏和按钮。 它确实有一些表面贴装元件，但只是稳压器和去耦电容：
显示屏是一块Crystalfontz CFAL128128A0-015W单色OLED屏：
显示屏和PCB都没有用螺丝固定： 它们是通过压力卡在位置上的。 因此，PCB的背面必须是空白的，以免压坏OLED显示屏：
按钮位于一块小小的子板上：
这些按钮是设计中相对后期才添加的，通过一条长排线与主PCB连接。 与其它部分不同，
26. Notes on blog future-proofing
📰 Maurycy's Blog
📄 !-- mksite: start of content -- p emOne of the great things about a href="https://maurycyz.com/misc/starting_a_blog/"web pages/a is that they are long-lived and mutable/em. There's no need to aim for perfection on the first draft: A page can continue to be improved for years after its original publication. /pp However, this mutability comes at a cost: /p a href="https://commons.wikimedia.org/wiki/File:First_Web_Server.jpg"img src="https://maurycyz.com/misc/futureproofing/first_server.jpg" //a centerDO NOT POWER [IT] DOWN!! — The first web server./center p Servers are just computers: If they ever break or are turned off, the web site vanishes off the internet. /pp !-- snip -- /p p /p p !-- External links: 3rd party archiving services -- emIf you've ever been reading something more than a few years old/em, you've probably noticed that a href="https://en.wikipedia.org/wiki/Link_rot"none of the links work/a. Even if the destination site still exists, It's common for them to have a href="https://www.w3.org/Provider/Style/URI"changed the URL format/a so that old links don't work. /pp To be clear, links are a good thing: They allow readers to look deeper into a topic, and a href="https://maurycyz.com/real_pages/"external links/a are how we find new places on the internet. /pp h1Preserving external links:/h1 /pp em3rd party are services like a href="https://archive.org/"archive.org/a are hit-and-miss/em: By most accounts, only a href="https://arxiv.org/abs/1212.6177"around 50%/a of pages ever make it to the archive, and even if they have a copy, it's still just a web site: Many other archiving services !--like peeep.us-- have vanished or lost data. These services are good for archiving one's own site, but aren't great at defending against link rot. /pp emIf I want to be sure links will always work, they have to be archived locally./em /pp I don't want to run a a
📄 !-- mksite: 内容开始 --
p
em网页的一大优点在于其长久的生命力和可修改性/em。
初稿无需追求完美：
一个页面在首次发布后，仍可以持续改进多年。
/p
p
然而，这种可变性是有代价的：
/p
center切勿关闭电源！！ — 第一台网络服务器。/center
p
服务器本质上就是计算机：
一旦它们损坏或被关闭，网站就会从互联网上消失。
/p
p
!-- 省略部分 --
/p
p
/p
p
!-- 外部链接：第三方存档服务 --
em如果你曾阅读过几年前的内容/em，你可能已经注意到所有链接都已失效。
即使目标网站仍然存在，它们也常常会更改URL格式，导致旧链接无法访问。
/p
p
需要明确的是，链接本身是好事：
它们让读者能够深入探究某个主题，而外部链接正是我们在互联网上发现新天地的方式。
/p
h1如何保存外部链接：/h1
/p
p
em像archive.org这样的第三方服务效果参差不齐/em：
普遍认为，大约只有50%的网页最终被存档，而且即使它们存有副本，那也仍然只是一个网站：
许多其他存档服务（）已经消失或丢失了数据。
这些服务适合用于存档自己的网站，但在防止链接失效方面并不理想。
/p
p
em如果我想确保链接始终有效，就必须在本地进行存档。/em
/p
p
我不想运行一个a
27. It Depends
📰 iDiallo.com
📄 pThat's the answer I would always get from the lead developer on my team, many years ago. I wanted clear, concise answers from someone with experience, yet he never said "Yes" or "No." It was always "It depends."/p pIsn't it better to upgrade MySQL to the latest version? "It depends."/p pIsn't it better to upgrade our Ubuntu version to the one that was just released? "It depends."/p pOur PHP instance is reaching end-of-life, isn't it better to upgrade it right away? "It depends."/p pAt the time, that felt like the wrong answer. The correct answer was obviously "Yes." Of course it's better to do all those things. But there was so much that I couldn't see yet./p hr / pHave you considered that the main application using this instance can't be easily updated? It doesn't support newer MySQL drivers, which means we'd have to go through the process of upgrading the application first before touching the database. So yes, upgrading is better in theory. But it depends on whether we can allocate the time to do it in the right order./p pIt's great to move to the latest version of Ubuntu, but our policy was to stay on LTS releases for stability. Yes, a newer version means new features, but it also means risking breaking changes in a production environment. When you're responsible for systems other people depend on, latest isn't always safest./p pAt the time I asked this question, we were running PHP 4.x. PHP 5 was already out and receiving patches. Yes, upgrading would have improved performance and closed critical vulnerabilities. But we also ran several forums that had never been tested on PHP 5. In hindsight, they were completely incompatible. A hasty upgrade would have taken them offline./p div class="image" img alt="thinking monkey" src="https://cdn.idiallo.com/images/assets/627/thinking.jpg" / /div pMy lead developer had been doing this for years longer than me. He'd already watched systems break after rushed upgrade
📄
很多年前，我团队的首席开发人员总是这样回答我。我希望从有经验的人那里得到清晰简洁的答案，但他从不说“是”或“不是”，永远是“看情况”。
把MySQL升级到最新版本不是更好吗？——“看情况。”
把我们的Ubuntu版本升级到刚发布的版本不是更好吗？——“看情况。”
我们的PHP实例快停止支持了，马上升级不是更好吗？——“看情况。”
当时，我觉得这是个错误的答案。正确答案显然是“是的”。做所有这些事当然更好。但那时还有很多我看不到的东西。
你有没有考虑过，使用这个实例的主要应用程序不容易更新？它不支持更新的MySQL驱动，这意味着我们必须先升级应用程序，然后才能动数据库。所以理论上，升级是更好的。但这取决于我们能否按正确顺序分配时间来完成。
升级到最新版Ubuntu固然好，但我们的政策是为了稳定性而坚持使用LTS版本。是的，新版本意味着新功能，但也意味着在生产环境中可能引入破坏性变更。当你负责别人依赖的系统时，最新并不总是最安全。
我问这个问题时，我们运行的是PHP 4.x。PHP 5已经发布并接收补丁。是的，升级本可以提升性能并修复关键漏洞。但我们也运行着几个从未在PHP 5上测试过的论坛。事后看来，它们完全不兼容。仓促升级会让它们离线。
我的首席开发人员比我多干了好多年。他已经目睹过系统在仓促升级后崩溃……
28. Interruption-Driven Development
📰 iDiallo.com
📄 pI have a hard time listening to music while working. I know a lot of people do it, but whenever I need to focus on a problem, I have to hunt down the tab playing music and pause it. And yet I still wear my headphones. Not to listen to anything, but to signal to whoever is approaching my desk that I am working. It doesn't deter everyone, but it buys me the time I need to stay focused a little longer./p pI don't mind having a conversation with coworkers. What I mind is the interruption itself, especially when I'm in the middle of a task. Sometimes I'm debugging an issue in a legacy application, building a mental model of the workflow, reading a comment that describes an exception, following a function declaration, right when I'm on the verge of the next clue, I hear a voice: em"Hey! What's going on? I haven't seen you in a while. What have you been up to?"/em/p pThe conversation is never long. But when it's over, my thoughts are gone. Where was I? Right, the function declaration. But where was it being called? What was that exception the comment described? Where did I even see that comment? I have to retrace every step just to rebuild the mental state I was in before I can move forward again./p pWorking remotely helps, to a point. Interruptions via Slack can be muted until I'm ready to respond. But remote work isn't immune. You're still expected to be in meetings. As a lead, I'm frequently pulled into calls because "everything is on fire." Often, my presence isn't to put out the fire, it's to hold someone's hand. An hour later, I can barely remember what I was working on./p pThe cost of interruption falls entirely on the person being interrupted. You lose your place, your focus, and eventually your ability to finish anything on time. For the person doing the interrupting, though, it's often a positive experience. The manager who constantly pulls the team into status updates feels productive. They're in the loop, they're present, they're on
📄
我很难边工作边听音乐。我知道很多人这样做，但每当我需要专注解决问题时，总得找到播放音乐的标签页把它暂停。即便如此，我仍然戴着耳机——不是为了听任何声音，而是为了向靠近我工位的人传递"我正在工作"的信号。这不能阻挡所有人，但能为我争取到保持专注所需的片刻时间。
我并不排斥与同事交流，真正困扰我的是"被打断"这个行为本身，尤其当我沉浸在某项任务时。有时我正在调试遗留应用程序的问题，构建工作流程的心理模型，阅读描述异常的注释，追踪函数声明……就在即将找到下一个线索的关键时刻，耳边响起："嘿！最近怎么样？好久没见你了，在忙什么呀？"
对话从来不长。但结束后，我的思路已荡然无存。刚才进展到哪了？对，函数声明。但它在哪被调用来着？注释里描述的异常是什么？我甚至是在哪里看到那条注释的？我不得不重新追溯每个步骤，只为重建被打断前的思维状态，才能继续推进工作。
远程工作在一定程度上有帮助。Slack的干扰消息可以静音，等我准备好再回复。但远程工作并非免疫干扰——你仍然需要参加会议。作为团队负责人，我经常被拉进紧急电话会议，因为"所有事情都着火了"。而很多时候，我的参与并非为了灭火，只是安抚人心。一小时后，我几乎记不起原本在做什么。
被打断的代价完全由被中断者承担。你丢失了工作进度，分散了注意力，最终可能丧失按时完成任何任务的能力。然而对于打断者而言，这往往是积极体验——不断召集团队开进度会的管理者会感到高效，他们掌控全局、保持在场、持续参与……
29. Mo Samuels wrote this blog post
📰 iDiallo.com
📄 pLast year, I pushed myself to write and publish every other day for the whole year. I had accumulated a large number of subjects over the years, and I was ready to start blogging again. After writing a dozen or so articles, I couldn't keep up. What was I thinking? 180 articles in a year is too much. I barely wrote 4 articles in 2024. But there was this new emerging technology that people wouldn't stop talking about. What if I used it to help me achieve my goal?/p div class="image" a href="https://idiallo.com/stats"img alt="iDiallo Stats" src="https://cdn.idiallo.com/images/assets/625/stats.png" //a /div pHave you ever heard of Mo Samuels? You probably haven't. But you must have heard of a href="https://seths.blog/about/"Seth Godin/a, right? Seth Godin is the author of several bestsellers. He is an icon in the world of marketing, and at one point he nudged me just enough to quit an old job./p pThis is someone I deeply respected, and I bought his book emAll Marketers are Liars/em with great anticipation. I was several chapters in when he dropped this statement:/p blockquote pI didn't write this book./p /blockquote pWhat does he mean by that? His name is on the cover. These are the familiar words I often heard in his seminars. What is he trying to say?/p blockquote pWhat I mean is that Seth Godin didn't write this book. It was written by a freelancer for hire named Mo Samuels. Godin hired me to write it based on a skimpy three-page outline./p /blockquote pWhat? Mo Samuels? Who is Mo Samuels? If that name were on the cover, I wouldn't have bought the book in the first place./p blockquote pDoes that bum you out? Does it change the way you feel about the ideas in this book? Does the fact that Seth paid me $10,000 and kept the rest of the advance money make the book less valuable?/p /blockquote pWell, yeah. It doesn't change the ideas in the book. But it is deceptive. I bought it specific
📄 去年，我强迫自己每隔一天就写作并发布一篇文章，坚持了整整一年。多年来我积累了大量的主题，也准备好重新开始写博客。但在写了十几篇文章后，我就跟不上了。我当时在想什么？一年180篇文章实在太多了。2024年我勉强只写了4篇。但这时出现了一项新兴技术，人们对此议论纷纷。如果我用它来帮助我实现目标呢？ div class="image" a href="https://idiallo.com/stats"img alt="iDiallo 数据统计" src="https://cdn.idiallo.com/images/assets/625/stats.png" //a /div 你听说过莫·塞缪尔斯吗？你可能没有。但你一定听说过 a href="https://seths.blog/about/"赛斯·高汀/a，对吧？赛斯·高汀是几本畅销书的作者。他是营销界的标志性人物，曾经他的影响足以促使我辞去一份旧工作。 这是我深深尊敬的人，我满怀期待地买下了他的书《所有营销人都是骗子》。我读了几章后，他抛出了这样一句话： blockquote p这本书不是我写的。/p /blockquote 他这是什么意思？封面上明明印着他的名字。这些是我在他研讨会上经常听到的熟悉话语。他想说什么？ blockquote p我的意思是，赛斯·高汀没有写这本书。它是由一位名叫莫·塞缪尔斯的自由职业者受雇撰写的。高汀根据一份仅三页的简要大纲雇佣我写了它。/p /blockquote 什么？莫·塞缪尔斯？莫·塞缪尔斯是谁？如果封面上是这个名字，我一开始根本就不会买这本书。 blockquote p这让你失望了吗？这是否改变了你对书中观点的看法？赛斯付给我一万美元而保留了其余预付款的事实，是否让这本书的价值降低了？/p /blockquote 嗯，是的。它并没有改变书中的观点。但这是一种欺骗。我买这本书正是因为……
30. That's it, I'm cancelling my ChatGPT
📰 iDiallo.com
📄 pJust like everyone, I read Sam Altman's tweet about joining the so-called Department of War, to use ChatGPT on DoW classified networks. As others have pointed out, this is the entry point for mass surveillance and using the technology for weapons deployment. I wrote before that a href="https://idiallo.com/blog/we-have-all-we-need-for-mass-surveillance"we had the infrastructure for mass surveillance in place/a already, we just needed an enabler. This is the enabler./p pThis comes right after Anthropic's CEO wrote a public letter stating their refusal to work with the DoW under their current terms. Now Anthropic has been declared a public risk by the President and banned from every government system./p pLarge language models have become ubiquitous. You can't say you don't use them because they power every tech imaginable. If you search the web, they write a summary for you. If you watch YouTube, one appears right below the video. There's a Gemini button on Chrome, there's Copilot on Edge and every Microsoft product. There it is in your IDE, in Notepad, in MS Paint. You can't escape it./p pSwitching from one LLM to the next makes minimal to no difference for everyday use. If you have a question you want answered or a document to summarize, your local Llama will do the job just fine. If you want to compose an email or proofread your writing, there's no need to reach for the state of the art, any model will do. For reviewing code, DeepSeek will do as fine a job as any other model./p div class="art-image" img alt="OpenAI war soldier" src="https://cdn.idiallo.com/images/assets/daily/95/openaisoldier.jpg" / pA good use of ChatGPT's image generator./p /div pAll this to say, ChatGPT doesn't have a moat. If it's your go-to tool, switching away from it wouldn't make much of a difference. At this point, I think the difference is psychological. For example, my wife once told me she only ever uses Google and can't stand any other search en
📄 正如所有人一样，我看到了萨姆·奥特曼关于加入所谓"战争部"的推文——旨在国防部机密网络上使用ChatGPT。正如其他人指出的，这正是大规模监控的开端，是将技术用于武器部署的入口。我曾撰文提到，我们早已具备大规模监控的基础设施，只缺一个"催化剂"。而这就是那个催化剂。 此事紧接在Anthropic公司CEO发表公开信之后，该信声明他们拒绝在现有条款下与国防部合作。如今Anthropic已被总统宣布为公共风险，并禁止接入所有政府系统。 大语言模型已无处不在。你无法声称自己不用它们，因为所有你能想到的科技产品都由它们驱动。当你搜索网页时，它们为你生成摘要；观看YouTube时，摘要就出现在视频下方。Chrome浏览器上有Gemini按钮，Edge和所有微软产品都内置Copilot。它们存在于你的集成开发环境、记事本乃至MS Paint中。你无处可逃。 对日常使用而言，在不同大语言模型间切换几乎毫无差别。如果你有问题需要解答或文件需要总结，本地部署的Llama就能胜任；想要撰写邮件或校对文稿，也无需追求最尖端模型，任何模型都能完成；代码审查方面，DeepSeek的表现与其他模型同样出色。
ChatGPT图像生成器的妙用。
31. We Need Process, But Process Gets in the Way
📰 iDiallo.com
📄 pHow do you manage a company with 50,000 employees? You need processes that give you visibility and control across every function such as technology, logistics, operations, and more. But the moment you try to create a single process to govern everyone, it stops working for anyone./p pOne system can't cater to every team, every workflow, every context. When implemented you start seeing in-fighting, projects missing deadlines, people quitting. Compromises get made, and in my experience, it almost always becomes overwhelming./p pThe first time I was part of a merger, I was naïve about how it would go. The narrative we were sold was reassuring. The larger company was acquiring us because we were successful. The last thing they'd want to do was get in the way of that success. But that's not how it went./p pIt doesn't matter what made you successful before you join a larger organization. The principles and processes of the acquiring company are what will dominate. Your past success is acknowledged, maybe even celebrated, but it doesn't protect you from assimilation./p pOne of the first things we had to adopt was Scrum. It may be standard practice now, but at the time it was still making its way through the industry. Our team, developers and product managers, already had a process that worked. We knew how to communicate, how to prioritize, how to ship. Adopting this new set of ceremonies felt counterproductive. It didn't make us faster. It didn't improve communication. What it did do was increase administrative overhead. Standups, sprints, retrospectives, layer after layer of structure added on top of work that was already getting done./p pBut there was no going back. We were never going to return to being that nimble, ad hoc team that could resolve issues quickly and move on. We had to adopt methods that got in the way./p pEventually, we adapted. We adopted the process. And in doing so, we became less efficient at the local level. A lot o
📄 如何管理一家拥有五万名员工的公司？你需要建立能够覆盖技术、物流、运营等所有职能的可视化管控流程。但当你试图制定统一流程来管理所有人时，它往往对任何人都无法奏效。 单一体系无法满足每个团队、每项工作流程、每种具体情境的需求。强制推行后，内部冲突、项目延期、人员流失等问题便会浮现。妥协在所难免，而根据我的经验，这种局面几乎总会变得难以收拾。 首次参与企业并购时，我对后续发展抱有天真幻想。我们听到的官方说辞令人安心——大公司因我们的成功而收购，他们最不愿做的就是阻碍这种成功。但现实截然不同。 加入大型组织后，过往的成功经验已无关紧要。收购方的原则与流程将占据绝对主导。你的历史成就或许会获得认可甚至赞誉，但这并不能使你免于被同化的命运。 我们被迫推行的首批变革中就包括Scrum。尽管如今这已是行业标准，但当时仍属新兴实践。我们的开发与产品团队原本拥有高效的工作流程——我们深谙沟通之道、优先级排序之法、产品交付之术。强行套用这套新仪式感十足的框架反而适得其反：它既未提升效率，也未改善沟通，只是徒增管理成本。站会、冲刺、复盘会议……层层叠叠的架构压在了本已顺畅的工作流程之上。 但我们已经没有退路。那个能快速解决问题并灵活转向的敏捷团队一去不返，我们不得不接受这些阻碍效率的标准化流程。 最终我们选择了适应。我们接纳了新流程，却因此导致团队层面的效率流失。许多原本……
32. When access to knowledge is no longer the limitation
📰 iDiallo.com
📄 pLet's do this thought experiment together. I have a little box. I'll place the box on the table. Now I'll open the little box and put all the arguments against large language models in it. I'll put all the arguments, including my own. Now, I'll close the box and leave it on the table./p pNow that that is out of the way, we are left with all the positives. All the good things that come from having the world's information at our fingertips. I can ask any question and get an answer almost instantly. Well, not all questions. The East has its sensitivities around a certain square, and the West about a certain island, but I digress./p pI can learn any subject I want to learn. I can take the work of any philosopher and ELI5 it. I can finally understand "The World as Will and Representation" by Schopenhauer. A friend gifted me a copy when I was still in my twenties, it's been steadily collecting dust ever since. But now I can turn to the book and ask questions until I thoroughly understand it. No need to read it cover to cover./p pIn fact, last year I decided I wanted to learn about batteries. I first went to the Battery University website and started to read lesson by lesson. But I had questions. How was I going to get them answered? The StackExchange network a href="https://idiallo.com/blog/we-were-never-good-programmers"is not what it used to be/a, so I turned to ChatGPT. It had all the answers. I learned and read so much about batteries that I am tempted to start a battery company./p pMy twin boys are at that age where they suffer from the infinite WHYs. Why does it rain? Why does the earth spin? why does California still use the Highway Gothic font on some freeway signs? I do not have answers to these questions off the top of my head, but I have access to the infinite knowledge machine, so of course my kids know the answers now./p pJust the other day, I had a shower-thought about cars. "Are cars just a slab of metal on wheels?" And now
📄
让我们一起来做这个思想实验。我有一个小盒子。我会把盒子放在桌上。现在我要打开这个小盒子，把所有反对大型语言模型的论点都放进去。我会放入所有论点，包括我自己的。然后，我会关上盒子，把它留在桌上。
既然这些已被搁置，剩下的就全是积极面了。那些因世界信息触手可及而产生的所有美好事物。我可以提出任何问题并几乎立刻得到答案。当然，并非所有问题都能回答——东方对某个广场存在敏感议题，西方对某个岛屿也有忌讳，但这有些离题了。
我可以学习任何想了解的学科。我能将任何哲学家的著作用"五岁儿童能懂的语言"阐释。我终于能理解叔本华的《作为意志和表象的世界》了。有位朋友在我二十多岁时赠予我这本书，此后它一直默默积尘。但现在我可以翻开书页不断提问，直到彻底理解它，而无需逐页通读。
事实上，去年我决定学习电池知识。我先访问了"电池大学"网站开始逐课阅读，但产生了疑问。该如何解决这些疑问呢？StackExchange问答网络已不复当年，于是我转向ChatGPT。它拥有所有答案。我如饥似渴地学习电池知识，甚至萌生了创办电池公司的念头。
我的双胞胎儿子正处在"无限为什么"阶段：为什么会下雨？地球为什么旋转？为什么加州某些高速公路标志仍在使用Highway Gothic字体？我无法立即给出答案，但我拥有连接无限知识机器的途径，所以现在我的孩子们总能得到解答。
就在前几天，我洗澡时突然想到关于汽车的问题："汽车难道只是装在轮子上的一块金属板吗？"而现在……
33. The Little Red Dot
📰 iDiallo.com
📄 pSometimes, I have 50 tabs open. Looking for a single piece of information ends up being a rapid click on each tab until I find what I'm looking for. Somehow, every time I get to that LinkedIn tab, I pause for a second. I just have to click on the little red dot in the top right corner, see that there is nothing new, then resume my clicking. Why is that? Why can't I ignore the red notification badge?/p pWhen you sign up for LinkedIn for the first time, it's right there. A little red dot in the top right corner with a number in it. It stands out against the muted grays and blues of the interface. Click on it, and you'll discover you have a notification. It's not from someone you know; this is a fresh new account, after all. But the dot was there anyway./p pAdd a few connections, give it some time, and come back. Refresh the page, and you'll have new notifications waiting./p pIf your LinkedIn account is like mine, a ghost town, you still get the little red dot. My connections and I usually keep a few recruiters in our networks, an insurance policy in case we need to find work quickly. But we rarely, if ever, post anything. Yet whenever I log in, there's a new notification. Sometimes it's even a message, but not from anyone in my connections list. It's from LinkedIn itself./p pThe little red dot isn't exclusive to LinkedIn. My Facebook account has been dormant for years, yet those few times annually when I log in, the notifications are right there waiting for me. I've even visited news websites where the little red dot appeared for reasons I couldn't understand. I didn't have an account, so what exactly were they notifying me about?/p div class="image" img alt="Notification" src="https://cdn.idiallo.com/images/assets/622/dot.svg" / /div pThat little red dot is a sophisticated psychological trigger designed to exploit the brain. It activates the brain's a href="https://en.wikipedia.org/wiki/Salience_network"Salience Network/a.
📄
有时，我会同时打开50个标签页。为了寻找某条信息，我不得不快速点击每个标签页进行浏览。但不知为何，每次点到领英（LinkedIn）标签页时，我都会停顿片刻——我必须点开右上角的那个小红点，即使发现并没有新消息，才会继续浏览其他页面。这是为什么呢？为什么我无法忽视那个红色的通知徽章？
当你首次注册领英时，它就在那里：一个位于右上角、带着数字的小红点。在界面低调的灰蓝色调衬托下，它显得格外醒目。点击它，你会发现收到了一条通知。这条通知并非来自你认识的人（毕竟这是个全新账户），但红点依然存在。
添加一些人脉，等待一段时间再回来。刷新页面，就会有新通知等着你。
如果你的领英账户像我的一样冷清，那个小红点依然会出现。我和我的人脉通常会在网络中保留一些招聘人员，作为需要快速找工作时的一道保险。但我们几乎从不发布任何内容。然而每次登录时，总会有新通知。有时甚至是一条消息，但并非来自我人脉列表中的任何人——而是领英平台本身发送的。
小红点并非领英独有。我的脸书（Facebook）账户已沉寂多年，但每年偶尔登录时，通知总在那里等着我。我甚至访问过一些新闻网站，那里也会出现令我费解的小红点。我根本没有注册账户，他们到底想通知我什么？
那个小红点是一种精心设计的心理触发器，旨在利用大脑的运作机制。它会激活大脑的凸显网络。
34. Nvidia was only invited to invest
📰 iDiallo.com
📄 pNvidia was only invited to invest. /p pThat is one reversal of commitment. Remember that graph that has been circling around for some time now? The one that shows the circular investment from AI companies:/p p source / img alt="OpenAI circular investment" src="https://cdn.idiallo.com/images/assets/daily/94/circular.jpg" / /p pBasically Nvidia will invest $100 billion in OpenAI. OpenAI will then invest $300 billion in Oracle, then Oracle invests back into Nvidia. Now, Jensen Huang, the Nvidia CEO, is back tracking and saying he a href="https://fortune.com/2026/02/02/jensen-huang-nvidia-ceo-on-openai-investment-never-a-commitment/"never made that commitment/a. /p blockquote p“It was never a commitment. They invited us to invest up to $100 billion and of course, we were, we were very happy and honored that they invited us, but we will invest one step at a time.”/p /blockquote pSo he never committed? Did we make up all these graphs in our head? Was it a misquote from a journalist somewhere that sparkled all this frenzy? Well, you can take a look in a href="https://openai.com/index/openai-nvidia-systems-partnership/"OpenAI press release in September of 2025/a. They wrote:/p blockquote pNVIDIA intends to invest up to $100 billion in OpenAI as the new NVIDIA systems are deployed./p /blockquote pIn fact, Jensen Huang went on to say:/p blockquote p“NVIDIA and OpenAI have pushed each other for a decade, from the first DGX supercomputer to the breakthrough of ChatGPT. This investment and infrastructure partnership mark the next leap forward—deploying 10 gigawatts to power the next era of intelligence.”/p /blockquote pIt sounds like Jensen is distancing himself from that $100 billion commitment. Did he take a peak inside OpenAI and change his mind? At the same time, OpenAI is experimenting with ads. Sam Altman stated before that they would only ever use ads as a last resort. It sounds like we
📄 **英伟达只是受邀投资。** 这可谓是一次承诺的逆转。还记得最近流传甚广的那张图表吗？就是展示人工智能公司之间循环投资的那张： （此处为图片描述：图片alt文本为"OpenAI循环投资"，图片地址为https://cdn.idiallo.com/images/assets/daily/94/circular.jpg） 大致内容是：英伟达将向OpenAI投资1000亿美元。然后，OpenAI将向甲骨文投资3000亿美元，接着甲骨文再投资回英伟达。如今，英伟达首席执行官黄仁勋却改口了，他表示自己[从未做出过那样的承诺](https://fortune.com/2026/02/02/jensen-huang-nvidia-ceo-on-openai-investment-never-a-commitment/)。 > “那从来就不是一项承诺。他们邀请我们投资，额度最高可达1000亿美元，当然，我们对此感到非常高兴和荣幸，但我们会一步一步地进行投资。” 那么，他从未承诺过？难道所有这些图表都是我们凭空想象出来的？是某处记者的错误引用引发了这场狂热吗？好吧，你可以看看[OpenAI在2025年9月的新闻稿](https://openai.com/index/openai-nvidia-systems-partnership/)。他们写道： > 随着新的英伟达系统部署，英伟达**有意**向OpenAI投资高达1000亿美元。 事实上，黄仁勋当时还进一步表示： > “英伟达和OpenAI已经相互推动十年了，从第一台DGX超级计算机到ChatGPT的突破。这项投资和基础设施合作伙伴关系标志着下一个飞跃——部署10吉瓦的电力来推动下一个智能时代。” 听起来黄仁勋正在与那1000亿美元的承诺保持距离。他是不是窥探了OpenAI内部后改变了主意？与此同时，OpenAI正在尝试广告业务。山姆·阿尔特曼此前曾表示，他们只会在万不得已时才会使用广告。听起来我们……
35. Teleoperation is Always the Butt of the Joke
📰 iDiallo.com
📄 pA few years back, the term "AI" took an unexpected turn when it was redefined as "Actual Indian". As in, a person in India operating the machine remotely./p pI first heard the term when Amazon was boasting about their cashierless grocery stores. There was a big sign in the store that said "Just Walk Out," meaning you grab your items, walk out, and get charged the correct amount automatically. How did they do it? According to Amazon, they used AI. What kind of AI exactly, nobody was quite sure./p pBut customers started reporting something odd. They weren't charged immediately after leaving the store. Some said it took several days for a charge to appear on their account. It eventually came out that the technology was sophisticated tracking performed by Amazon's team in India. Workers would manually review footage of each customer's visit and charge them accordingly. This operation was impressive. Coordinating thousands of store visits, matching items to customers across multiple camera angles, and doing it accurately enough that most people never noticed the delay. But because it was buried under the "AI" label, the moment the truth came out, the whole thing became a punchline./p pIn 2024, Tesla held their "We, Robot" event, where Optimus robots operated a bar. They were serving drinks, dancing, and mingling with guests. It was a pretty impressive display. The robots moved fluidly, held conversations, and handed off drinks without fumbling. Elon Musk a href="https://old.reddit.com/r/BetterOffline/comments/1phullc/tesla_optimus_robot_that_elon_musk_claimed_was_ai/"claimed they were AI-driven/a, fully autonomous. People were genuinely impressed by the interactions, and for good reason. Fluid, bipedal locomotion in a crowded social environment is an extraordinarily hard robotics problem./p pThe moment it came out that the robots were teleoperated, the sentiment flipped entirely. It didn't matter how dexterous or natural the movement was. It
📄 几年前，"AI"这个词意外地转向，被重新定义为"真正的印度人"。具体来说，指的是在印度远程操作机器的人。 我第一次听到这个说法是在亚马逊吹嘘其无人收银杂货店的时候。店里有一个巨大的标志写着"拿了就走"，意思是你可以拿了商品直接走出商店，系统会自动收取正确金额。他们是怎么做到的？根据亚马逊的说法，他们使用了人工智能。但具体是哪种人工智能，当时没人完全清楚。 然而，顾客开始报告一些奇怪的现象。他们离开商店后并没有被立即扣款。有人说，过了好几天他们的账户才出现扣费记录。最终真相大白：这项技术实际上是由亚马逊在印度的团队进行的复杂追踪。工作人员会手动查看每位顾客到店的录像，并据此进行收费。这项操作令人印象深刻——协调处理成千上万的到店记录，通过多角度摄像头匹配顾客与商品，并且做得足够准确，以至于大多数人从未注意到扣款延迟。但由于它被隐藏在"人工智能"的标签下，真相一经曝光，整个事情就变成了一个笑柄。 2024年，特斯拉举办了"我们，机器人"活动，擎天柱机器人在酒吧里工作。它们提供饮品、跳舞、与客人互动。这场展示相当令人印象深刻：机器人动作流畅、与人交谈、递送饮料时毫无差错。埃隆·马斯克声称它们是人工智能驱动、完全自主的。人们确实被这种人机互动所震撼，这是有道理的——在拥挤的社交环境中实现流畅的双足运动，本就是机器人学中极其困难的课题。 然而，当这些机器人实为远程操控的真相曝光后，舆论彻底反转。无论它们的动作多么灵巧自然都已不再重要。
36. GNU and the AI reimplementations
📰
📄 Those who cannot remember the past are condemned to repeat it. A sentence that I never really liked, and what is happening with AI, about software projects reimplementations, shows all the limits of such an idea. Many people are protesting the fairness of rewriting existing projects using AI. But, a good portion of such people, during the 90s, were already in the field: they followed the final part (started in the ‘80s) of the deeds of Richard Stallman, when he and his followers were reimplementing the UNIX userspace for the GNU project. The same people that now are against AI rewrites, back then, cheered for the GNU project actions (rightly, from my point of view – I cheered too). br / br /Stallman is not just a programming genius, he is also the kind of person that has a broad vision across disciplines, and among other things he was well versed in the copyright nuances. He asked the other programmers to reimplement the UNIX userspace in a specific way. A way that would make each tool unique, recognizable, compared to the original copy. Either faster, or more feature rich, or scriptable; qualities that would serve two different goals: to make GNU Hurd better and, at the same time, to provide a protective layer against litigations. If somebody would claim that the GNU implementations were not limited to copying ideas and behaviours (which is legal), but “protected expressions” (that is, the source code verbatim), the added features and the deliberate push towards certain design directions would provide a counter argument that judges could understand. br / br /He also asked to always reimplement the behavior itself, avoiding watching the actual implementation, using specifications and the real world mechanic of the tool, as tested manually by executing it. Still, it is fair to guess that many of the people working at the GNU project likely were exposed or had access to the UNIX source code. br / br /When Linus reimplemented UNIX, writing the Linux kernel,
📄 那些无法铭记过去的人注定要重蹈覆辙。这句话我从未真正认同，而当前人工智能领域关于软件项目重实现的现象，恰恰揭示了这种观念的局限性。许多人质疑使用AI重写现有项目的公平性。但值得注意的是，这群人中相当一部分早在九十年代就已身处行业——他们见证了理查德·斯托曼及其追随者为GNU项目重写UNIX用户空间的最后阶段（该行动始于八十年代）。如今反对AI重写的人们，当年却曾为GNU项目的行动欢呼（在我看来这种支持是正当的，当时我也同样为之喝彩）。 斯托曼不仅是编程天才，更是具备跨学科广阔视野的智者，尤其深谙著作权法的微妙之处。他要求程序员以特定方式重实现UNIX用户空间：每个工具都必须相比原版具有独特性和辨识度——或更快速、或功能更丰富、或支持脚本编写。这些特质服务于双重目标：既提升GNU Hurd系统的品质，同时构筑抵御法律诉讼的防护层。若有人指控GNU实现不仅复制了思想与行为（这本身合法），还抄袭了“受保护的表达形式”（即源代码逐字复制），那么这些新增特性与刻意追求的设计方向将成为法官能够理解的有力抗辩依据。 他还要求开发者始终通过规范文档和工具的实际运行机制进行重实现，避免参考原始代码实现，需通过手动执行测试来理解工具原理。尽管如此，我们仍有理由推测，参与GNU项目的许多开发者很可能接触或获取过UNIX源代码。 当林纳斯重实现UNIX并编写Linux内核时，
37. Redis patterns for coding
📰
📄 Here LLM and coding agents can find: br / br /1. Exhaustive documentation about Redis commands and data types. br /2. Patterns commonly used. br /3. Configuration hints. br /4. Algorithms that can be mounted using Redis commands. br / br /https://redis.antirez.com/ br / br /Some humans claim this documentation is actually useful for actual people, as well :) I'm posting this to make sure search engines will index it. a href="http://antirez.com/news/161"Comments/a
📄 在这里，LLM与编程智能体可以找到：
1. 关于Redis命令与数据类型的详尽文档。
2. 常用模式。
3. 配置提示。
4. 可通过Redis命令搭载的算法。
https://redis.antirez.com/
也有人认为这份文档对真实用户同样实用哦 :) 我发布此内容以确保搜索引擎能将其收录。
评论
38. Implementing a clear room Z80 / ZX Spectrum emulator with Claude Code
📰
📄 Anthropic recently released a blog post with the description of an experiment in which the last version of Opus, the 4.6, was instructed to write a C compiler in Rust, in a “clean room” setup. br / br /The experiment methodology left me dubious about the kind of point they wanted to make. Why not provide the agent with the ISA documentation? Why Rust? Writing a C compiler is exactly a giant graph manipulation exercise: the kind of program that is harder to write in Rust. Also, in a clean room experiment, the agent should have access to all the information about well established computer science progresses related to optimizing compilers: there are a number of papers that could be easily synthesized in a number of markdown files. SSA, register allocation, instructions selection and scheduling. Those things needed to be researched *first*, as a prerequisite, and the implementation would still be “clean room”. br / br /Not allowing the agent to access the Internet, nor any other compiler source code, was certainly the right call. Less understandable is the almost-zero steering principle, but this is coherent with a certain kind of experiment, if the goal was showcasing the completely autonomous writing of a large project. Yet, we all know how this is not how coding agents are used in practice, most of the time. Who uses coding agents extensively knows very well how, even never touching the code, a few hits here and there completely changes the quality of the result. br / br /# The Z80 experiment br / br /I thought it was time to try a similar experiment myself, one that would take one or two hours at max, and that was compatible with my Claude Code Max plan: I decided to write a Z80 emulator, and then a ZX Spectrum emulator (and even more, a CP/M emulator, see later) in a condition that I believe makes a more sense as “clean room” setup. The result can be found here: https://github.com/antirez/ZOT. br / br /# The process I used br / br /1. I wro
📄 Anthropic 近期发布了一篇博客文章，描述了一项实验：在“净室”环境中，指示其最新版本模型 Opus 4.6 使用 Rust 语言编写一个 C 编译器。 该实验方法让我对他们试图论证的观点感到怀疑。为何不向智能体提供指令集架构（ISA）文档？为何选择 Rust？编写 C 编译器本质上是一项庞大的图结构操作任务——而这正是用 Rust 更难以编写的那类程序。此外，在净室实验中，智能体本应能获取所有关于优化编译器方面已确立的计算机科学进展信息：有许多论文可以轻松整合成多个 Markdown 文件，比如静态单赋值形式（SSA）、寄存器分配、指令选择与调度。这些内容本应作为前提条件*首先*进行调研，而实现过程仍可保持“净室”性质。 禁止智能体访问互联网或其他编译器源代码无疑是正确的决定。但几乎零引导的原则则较难理解——不过若目标是展示大型项目的完全自主编写过程，这倒符合某类实验的设定。然而我们都知道，这并非编码智能体在实践中的主流使用方式。频繁使用编码智能体的人都清楚，即使完全不触碰代码，仅在此处或彼处进行少量指引，也会彻底改变结果的质量。 # Z80 实验 我认为是时候亲自尝试一个类似的实验了：它至多耗时一两小时，且兼容我的 Claude Code Max 订阅计划。我决定在一种我认为更符合“净室”设定的条件下，编写一个 Z80 模拟器，接着是 ZX Spectrum 模拟器（甚至更进一步，编写 CP/M 模拟器，详见后文）。实验结果可在此处查看：https://github.com/antirez/ZOT。 # 我采用的流程 1. 我撰写了……
39. Automatic programming
📰
📄 In my YouTube channel, for some time now I started to refer to the process of writing software using AI assistance (soon to become just "the process of writing software", I believe) with the term "Automatic Programming". br / br /In case you didn't notice, automatic programming produces vastly different results with the same LLMs depending on the human that is guiding the process with their intuition, design, continuous steering and idea of software. br / br /Please, stop saying "Claude vibe coded this software for me". Vibe coding is the process of generating software using AI without being part of the process at all. You describe what you want in very general terms, and the LLM will produce whatever happens to be the first idea/design/code it would spontaneously, given the training, the specific sampling that happened to dominate in that run, and so forth. The vibe coder will, at most, report things not working or not in line with what they expected. br / br /When the process is actual software production where you know what is going on, remember: it is the software *you* are producing. Moreover remember that the pre-training data, while not the only part where the LLM learns (RL has its big weight) was produced by humans, so we are not appropriating something else. We can pretend AI generated code is "ours", we have the right to do so. Pre-training is, actually, our collective gift that allows many individuals to do things they could otherwise never do, like if we are now linked in a collective mind, in a certain way. br / br /That said, if vibe coding is the process of producing software without much understanding of what is going on (which has a place, and democratizes software production, so it is totally ok with me), automatic programming is the process of producing software that attempts to be high quality and strictly following the producer's vision of the software (this vision is multi-level: can go from how to do, exactly, certain things,
📄 在我的YouTube频道中，一段时间以来，我开始用"自动编程"这个术语来指代使用AI辅助编写软件的过程（我相信很快这就将直接被称为"编写软件的过程"）。 如果你没有注意到的话：自动编程即使使用相同的大语言模型，也会因操作者不同而产生截然不同的结果——这取决于人类在过程中注入的直觉、设计、持续引导以及对软件的理解。 请不要再说什么"是Claude凭感觉帮我编了这个软件"。所谓"感觉式编码"指的是完全脱离编程过程、仅靠AI生成软件的操作方式。你只是笼统地描述需求，而大语言模型会根据训练数据、该次运行中恰好占主导的特定采样等因素，自发产生它第一个想到的方案/设计/代码。感觉式编码者最多只会反馈程序无法运行或不符合预期。 当这个过程是真正理解来龙去脉的软件生产时，请记住：这是*你*正在创造的软件。更要记住的是，虽然大语言模型的学习不仅限于预训练数据（强化学习也占很大比重），但预训练数据本身正是人类创造的，所以我们并非侵占他物。我们可以将AI生成的代码视为"我们的作品"，我们有权这样做。实际上，预训练数据是我们共同的馈赠，它让许多原本无法编程的个体得以创造，某种程度上就像我们现在通过集体智慧连接在一起。 需要说明的是，如果感觉式编码是指在不甚理解的情况下生成软件（这有其存在意义，并能促进软件生产的民主化，我个人完全认可），那么自动编程则是试图产出高质量、严格遵循创作者软件愿景的生产过程（这种愿景是多层次的：可以具体到某些功能的实现方式，
40. Don't fall into the anti-AI hype
📰
📄 I love writing software, line by line. It could be said that my career was a continuous effort to create software well written, minimal, where the human touch was the fundamental feature. I also hope for a society where the last are not forgotten. Moreover, I don't want AI to economically succeed, I don't care if the current economic system is subverted (I could be very happy, honestly, if it goes in the direction of a massive redistribution of wealth). But, I would not respect myself and my intelligence if my idea of software and society would impair my vision: facts are facts, and AI is going to change programming forever. br / br /In 2020 I left my job in order to write a novel about AI, universal basic income, a society that adapted to the automation of work facing many challenges. At the very end of 2024 I opened a YouTube channel focused on AI, its use in coding tasks, its potential social and economical effects. But while I recognized what was going to happen very early, I thought that we had more time before programming would be completely reshaped, at least a few years. I no longer believe this is the case. Recently, state of the art LLMs are able to complete large subtasks or medium size projects alone, almost unassisted, given a good set of hints about what the end result should be. The degree of success you'll get is related to the kind of programming you do (the more isolated, and the more textually representable, the better: system programming is particularly apt), and to your ability to create a mental representation of the problem to communicate to the LLM. But, in general, it is now clear that for most projects, writing the code yourself is no longer sensible, if not to have fun. br / br /In the past week, just prompting, and inspecting the code to provide guidance from time to time, in a few hours I did the following four tasks, in hours instead of weeks: br / br /1. I modified my linenoise library to support UTF-8, and created a framew
📄 我热爱逐行编写软件。可以说，我的职业生涯就是持续努力创造书写精良、简洁克制的软件，其中"人的痕迹"始终是核心特质。我也期盼一个不会遗忘弱势群体的社会。此外，我不希望AI在经济层面取得成功，甚至不在意现有经济体系被颠覆（说实话，若能朝着财富大规模再分配的方向发展，我会非常欣慰）。但若因我对软件与社会的理念而蒙蔽双眼，我将无法尊重自己的智慧：事实就是事实，AI必将永久改变编程领域。 2020年我辞去工作，开始创作一部关于AI、全民基本收入、以及社会在应对工作自动化挑战中艰难转型的小说。2024年末，我开设了专注于AI的YouTube频道，探讨其在编程任务中的应用及其潜在的社会经济影响。虽然我很早就预见到变革将至，但原以为编程领域彻底重塑至少还需数年时间。如今我已改变这个看法。当前最先进的大语言模型仅需获得关于最终目标的适当提示，就能独立完成大型子任务或中型项目，几乎无需人工干预。成功程度取决于编程类型（越独立、越能用文本精确描述的任务效果越好，系统编程尤其适合）以及向大语言模型传达问题时构建思维表征的能力。但总体而言，现在显然对于大多数项目而言，亲手编写代码已非明智之举——除非是为了乐趣。 过去一周，仅通过提示词引导并偶尔检查代码提供指导，我在数小时内完成了以下四项原本需要数周的工作： 1. 为我的linenoise库添加UTF-8支持，并创建了一个框架
41. Reflections on AI at the end of 2025
📰
📄 * For years, despite functional evidence and scientific hints accumulating, certain AI researchers continued to claim LLMs were stochastic parrots: probabilistic machines that would: 1. NOT have any representation about the meaning of the prompt. 2. NOT have any representation about what they were going to say. In 2025 finally almost everybody stopped saying so. br / br /* Chain of thought is now a fundamental way to improve LLM output. But, what is CoT? Why it improves output? I believe it is two things: 1. Sampling in the model representations (that is, a form of internal search). After information and concepts relevant to the prompt topic is in the context window, the model can better reply. 2. But if you mix this to reinforcement learning, the model also learns to put one token after the other (each token will change the model state) in order to converge to some useful reply. br / br /* The idea that scaling is limited to the number of tokens we have, is no longer true, because of reinforcement learning with verifiable rewards. We are still not at AlphaGo move 37 moment, but is this really impossible in the future? There are certain tasks, like improving a given program for speed, for instance, where in theory the model can continue to make progress with a very clear reward signal for a very long time. I believe improvements to RL applied to LLMs will be the next big thing in AI. br / br /* Programmers resistance to AI assisted programming has lowered considerably. Even if LLMs make mistakes, the ability of LLMs to deliver useful code and hints improved to the point most skeptics started to use LLMs anyway: now the return on the investment is acceptable for many more folks. The programming world is still split among who uses LLMs as colleagues (for instance, all my interaction is via the web interface of Gemini, Claude, …), and who uses LLMs as independent coding agents. br / br /* A few well known AI scientists believe that what happened with Tr
📄 * 多年来，尽管功能性证据和科学线索不断积累，但部分AI研究者仍坚称大语言模型只是"随机鹦鹉"：即概率机器，它们将：1. 对提示的含义没有任何内部表征；2. 对自己将要生成的内容没有任何内部表征。到2025年，几乎所有人都终于不再这么说了。 * 思维链现已成为提升大语言模型输出的基本方法。但思维链究竟是什么？为何它能提升输出质量？我认为原因有二：1. 在模型表征空间中进行采样（这是一种内部搜索形式）。当与提示主题相关的信息和概念进入上下文窗口后，模型能给出更佳回复。2. 若将此方法与强化学习结合，模型还能学会通过连续生成词元（每个词元都会改变模型状态）来逐步收敛到有价值的回复。 * "模型能力仅受训练数据词元数量限制"的观点已不再成立，这要归功于具有可验证奖励的强化学习。我们虽未达到AlphaGo"第37步"那样的突破时刻，但未来真的不可能实现吗？在某些任务上（例如优化程序运行速度），理论上模型可以在长期获得清晰奖励信号的情况下持续进步。我相信，应用于大语言模型的强化学习改进将成为AI领域的下一个重大突破。 * 程序员对AI辅助编程的抵触已显著降低。即使大语言模型会犯错，但其提供可用代码和提示的能力已提升到让大多数怀疑者也开始使用的程度：如今投资回报率对更多人而言已可接受。编程界仍存在分化：有人将大语言模型视为协作同事（例如我完全通过Gemini、Claude等网页界面交互），有人则将其用作独立编程代理。 * 几位知名AI科学家认为，Tr...
42. Scaling HNSWs
📰
📄 I’m taking a few weeks of pause on my HNSWs developments (now working on some other data structure, news soon). At this point, the new type I added to Redis is stable and complete enough, it’s the perfect moment to reason about what I learned about HNSWs, and turn it into a blog post. That kind of brain dump that was so common pre-AI era, and now has become, maybe, a bit more rare. Well, after almost one year of thinking and implementing HNSWs and vector similarity stuff, it is time for some writing. However this is not going to be an intro on HNSWs: too many are present already. This is the “extra mile” instead. If you know HNSWs, I want to share with you my more “advanced” findings, especially in the context of making them fast enough to allow for a “Redis” experience: you know, Redis is designed for low latency and high performance, and HNSWs are kinda resistant to that, so there were challenges to expose HNSWs as an abstract data structure. br / br /This blog post will be split into several sections. Think of them as pages of the same book, different chapters of the same experience. Oh and, by the way, I already wrote and subsequently lost this blog post :D [long, sad story about MacOS and bad habits – I hadn’t lost something like that since the 90s, during blackouts], so here most of the problem will be to recall what I wrote a few days ago and, while I’m at it, to better rephrase what I didn’t like very much. br / br /## A few words about the state of HNSW br / br /Before digging into the HNSWs internals and optimizations, I want to say a few things about HNSWs. The original paper introducing HNSWs is a great piece of computer science literature, and HNSWs are amazing data structures, but: I don’t believe they are the last word for searching, in a greedy way, for nearby vectors according to a distance function. The paper gives the feeling it lacks some “pieces”, almost like if the researchers, given six months more, had a lot more to explore and sa
📄 我将暂停几周HNSW的开发工作（目前正转向其他数据结构，近期会有新消息）。现阶段，我为Redis添加的新数据类型已足够稳定完善，正是梳理HNSW相关技术心得并撰写博客的好时机。这种技术随笔在人工智能时代前本很常见，如今或许已稍显稀缺。经过近一年对HNSW与向量相似度技术的思考与实践，确实该做些文字总结了。不过本文并非HNSW入门指南——这类资料已汗牛充栋，而是想分享进阶内容。若您已了解HNSW，我将重点分享在Redis场景下的高阶发现：如何突破HNSW的性能瓶颈，使其适配Redis特有的低延迟高性能要求——毕竟将HNSW封装为抽象数据结构时确实面临诸多挑战。
本文将以章节形式展开，如同同一本书的不同篇章，记录这段完整的技术历程。顺便一提，其实这篇博客我曾写完却意外丢失了：）[关于macOS和不良习惯的漫长悲伤故事——自90年代停电事故以来，我从未遭遇过如此严重的数据丢失]，所以现在大部分工作将是回忆几天前写过的内容，并趁此机会优化那些原本不够满意的表述。
## 关于HNSW现状的几点思考
在深入探讨HNSW内部机制与优化策略前，我想先谈谈对HNSW的整体看法。提出HNSW的原始论文无疑是计算机科学的瑰宝，HNSW本身也是卓越的数据结构，但必须指出：我不认为它是以贪婪方式根据距离函数搜索邻近向量的终极解决方案。论文给人的感觉是缺失了某些“拼图”，仿佛研究者若能再获得六个月时间，就能探索并阐述更多……
43. AI is different
📰
📄 Regardless of their flaws, AI systems continue to impress with their ability to replicate certain human skills. Even if imperfect, such systems were a few years ago science fiction. It was not even clear that we were so near to create machines that could understand the human language, write programs, and find bugs in a complex code base: bugs that escaped the code review of a competent programmer. br / br /Since LLMs and in general deep models are poorly understood, and even the most prominent experts in the field failed miserably again and again to modulate the expectations (with incredible errors on both sides: of reducing or magnifying what was near to come), it is hard to tell what will come next. But even before the Transformer architecture, we were seeing incredible progress for many years, and so far there is no clear sign that the future will not hold more. After all, a plateau of the current systems is possible and very credible, but it would likely stimulate, at this point, massive research efforts in the next step of architectures. br / br /However, if AI avoids plateauing long enough to become significantly more useful and independent of humans, this revolution is going to be very unlike the past ones. Yet the economic markets are reacting as if they were governed by stochastic parrots. Their pattern matching wants that previous technologies booms created more business opportunities, so investors are polarized to think the same will happen with AI. But this is not the only possible outcome. br / br /We are not there, yet, but if AI could replace a sizable amount of workers, the economic system will be put to a very hard test. Moreover, companies could be less willing to pay for services that their internal AIs can handle or build from scratch. Nor is it possible to imagine a system where a few mega companies are the only providers of intelligence: either AI will be eventually a commodity, or the governments would do something, in such an odd
📄 尽管存在缺陷，人工智能系统在模仿某些人类技能方面仍持续展现惊人能力。即便不够完美，这类系统在数年前还只是科幻概念。当时人们甚至无法确信，我们竟如此接近创造出能够理解人类语言、编写程序、在复杂代码库中发现漏洞的机器——这些漏洞甚至逃过了资深程序员的代码审查。 由于大语言模型乃至整个深度学习模型的工作原理尚未被充分理解，即便是该领域最杰出的专家也屡屡未能准确预测其发展轨迹（在低估或高估短期潜力两方面都出现过惊人误判），我们很难断言未来将走向何方。但在Transformer架构出现之前，人工智能领域早已持续多年取得惊人进展，迄今仍未有明确迹象表明这种进步即将停滞。诚然，现有系统可能进入平台期且这种可能性相当可信，但这反而可能激发学界对下一代架构展开大规模研究。 然而，若人工智能能长期保持发展势头，最终变得高度实用且大幅减少对人类依赖，这场革命将与此前的技术变革截然不同。但当前资本市场的反应却仿佛被"随机鹦鹉"所支配——其模式匹配机制基于过往技术繁荣总会创造更多商业机会的认知，导致投资者两极分化地认为人工智能必将重演历史。但这并非唯一可能的结果。 我们尚未抵达那个阶段，但若人工智能真能替代大量劳动者，现有经济体系将面临严峻考验。更甚者，企业可能不再愿意为那些可由内部AI处理或构建的服务付费。我们也难以想象一个由少数巨头垄断智能供给的体系：要么人工智能终将成为普惠性基础设施，要么政府将在如此奇特的局面下采取干预措施。
44. Coding with LLMs in the summer of 2025 (an update)
📰
📄 Frontier LLMs such as Gemini 2.5 PRO, with their vast understanding of many topics and their ability to grasp thousands of lines of code in a few seconds, are able to extend and amplify the programmer capabilities. If you are able to describe problems in a clear way and, if you are able to accept the back and forth needed in order to work with LLMs, you can reach incredible results such as: br / br /1. Eliminating bugs you introduced in your code before it ever hits any user: I experienced this with Vector Sets implementation of Redis. I would end eliminating all the bugs eventually, but many were just removed immediately by Gemini / Claude code reviews. br / br /2. Explore faster how a given idea could work, by letting the LLM write the throw away code to test ASAP in order to see if a given solution is actually more performant, if it is good enough, and so forth. br / br /3. Engage in pair-design activities where your instinct, experience, design taste can be mixed with the PhD-level knowledge encoded inside the LLM. In this activity, the LLM will sometimes propose stupid paths, other times incredibly bright ideas: you, the human, are there in order to escape local minimal and mistakes, and exploit the fact your digital friend knows of certain and various things more than any human can. br / br /4. Accelerate your work by writing part of the code under your clear specifications. br / br /5. Work with technologies far from your expertise but contiguous with what you can do (for instance: coding in 68000 assembly for an Amiga demo?) using LLMs as an extension of specific parts of your mind, for the knowledge you don't have. br / br /One and half years ago I wrote a blog post called “LLMs and programming in the first days of 2024”. There, I found LLMs to be already useful, but during these 1.5 years, the progresses they made completely changed the game. However, in order to leverage their capabilities, humans interacting with LLMs must have ce
📄 前沿大语言模型（如Gemini 2.5 PRO）凭借其广博的知识储备和数秒内解析数千行代码的能力，正在成为程序员能力的延伸与放大器。只要你能清晰描述问题，并适应与大语言模型协作所需的往复沟通，就能实现令人惊叹的成果： 1. **在代码触及用户前根除潜在缺陷**：我在实现Redis向量集合功能时亲历此效。虽然最终总能自行解决所有漏洞，但Gemini/Claude的代码审查往往能即时消除大量错误。 2. **加速创意验证流程**：通过让大语言模型快速编写可弃用的测试代码，即刻验证解决方案的性能表现与可行性，极大缩短探索周期。 3. **开展智能结对设计**：将你的直觉、经验与设计品味，与大语言模型内化的博士级知识体系相融合。虽然它时而会提出荒谬方案，时而又迸发惊人洞见，而你的价值正是规避局部最优陷阱，善用这位超越人类知识疆界的数字伙伴。 4. **精准提升开发效率**：在明确规范指导下，让大语言模型完成部分代码实现，显著加速工作进程。 5. **拓展技术能力边界**：借助大语言模型作为专业知识的延伸，即使面对陌生但相邻的技术领域（比如为Amiga演示程序编写68000汇编），也能游刃有余地开展工作。 一年半前我曾撰写博文《2024年初的大语言模型与编程》，当时已觉察其应用价值。但这18个月间的技术演进彻底改变了游戏规则。需要注意的是，要充分释放大语言模型的潜力，与之交互的人类必须具备——
45. Human coders are still better than LLMs
📰
📄 This is a short story of how humans are still so much more capable of LLMs. Note that I'm not anti-AI or alike, you know it if you know me / follow me somewhere. I use LLMs routinely, like I did today, when I want to test my ideas, for code reviews, to understand if there are better approaches than what I had in mind, to explore stuff at the limit of my expertise, and so forth (I wrote a blog post about coding with LLMs almost two years, when it was not exactly cool: I was already using LLMs for coding and never stopped, I'll have to write an update, but that's not the topic of this post). br / br /But, still: the current level of AI is useful, great too, but so incredibly behind human intelligence, and I want to remark this as lately it is impossible to have balanced conversations. br / br /So, today I was working to Vector Sets for Redis, to fix a complicated bug: during the time I stopped working at Redis my colleagues introduced resistance against corruption RDB and RESTORE payloads, even when the checksum of the data passes. This feature is disabled by default, but provides an enhanced layer of safety for people wanting it. br / br /But… there is a but as big as an elephant: In order to make HNSWs fast to save into Redis RDBs and to load back, I serialized the *graph* representation, and not the element-vector pairs, otherwise I would have to re-insert back data into HNSWs, and that would be, like, 100 times slower (!). So I store all the links the nodes have with other nodes, as integers, and then I resolve them into pointers, it’s a nice trick and works great. But if you mix this and random corruptions of the representation, and the fact that my own twist on HNSWs enforce reciprocal links between nodes (I wrote my own implementation of HNSWs with many useful features, but reciprocal links are needed to enable many of them) then this could happen: br / br /1. We load corrupted data that says A links to B, but B no longer links to A (corrupted n
📄 这是一个关于人类能力仍远超大语言模型的简短故事。请注意，我并非反对人工智能——如果你认识我或在某些平台关注我，就会明白这一点。我日常频繁使用大语言模型，比如今天：当我想验证想法、进行代码审查、探索是否有比我设想更好的方案，或是研究专业领域边缘问题时都会用到（大约两年前，当这还不算时髦时，我就写过关于用大语言模型编程的博客文章：我早已将其用于编码且从未停止，或许该写篇更新了，不过这不是本文重点）。 但即便如此：当前人工智能水平虽然实用且出色，却仍与人类智能存在惊人差距。最近我发觉已很难进行平衡的讨论，因此特别想强调这一点。 今天我在处理Redis向量集合时，正在修复一个复杂漏洞：在我离开Redis工作期间，同事们引入了针对RDB和RESTORE数据损坏的防护机制，即使数据校验和通过也会触发。该功能默认关闭，但为需要者提供了增强的安全层。 然而存在一个巨象般显眼的"但是"：为了让HNSW图能快速保存到Redis RDB并重新加载，我序列化的是*图结构*而非元素-向量对，否则就需要将数据重新插入HNSW——那将导致速度降低约百倍（！）。我的方案是将节点间的所有连接以整数形式存储，再解析为指针，这个巧妙的技巧效果很好。但若将这种设计与随机数据损坏叠加，再加上我对HNSW的改进要求节点间必须保持双向连接（我自行实现的HNSW包含许多实用功能，而双向连接是实现多数功能的基础），就可能出现这种情况： 1. 我们加载的损坏数据显示A连接至B，但B却未连接回A（数据损坏导致n...
46. What I learned during the license switch
📰
📄 Yesterday, it was a very intense day. In Italy it was 1st of May, the workers holiday, so in the morning I went for a 4h walk in the Etna with friends 3, I love walking, and I often take pauses when coding just to walk, to return later at the keyboard with a few more kilometers on my legs, and walking in the Etna is amazing (Etna is the largest active volcano in Europe, and I happen to live in Catania, that is on its slopes). br / br /Then at 6PM I was at home to release my blog post about the AGPL license switch, and I started following the comments, feedbacks, private messages, and I learned a few things in the process. br / br /1. Regardless of the different few clauses, that IMHO make a difference, the AGPL vs SSPL main difference is that AGPL is "understood". In general, yesterday for the first time I realized that in licensing there is not just what you can do and can't do, but the degree a given license is understood, tested, adopted, ... br / br /2. I was very touched by the words of Simon Willison on the matter (https://simonwillison.net/2025/May/1/redis-is-open-source-again/) because it is very peculiar that different persons, living in different parts of the world, but with a similar age and background in software, feel *so similar* about things. I, too, when was writing Vector Sets, was thinking: I would never use it if it wasn't going to be released under the AGPL (or other open source license I understand). This sentiment, multiplied by a non trivial fraction of the community, makes open source eventually win even in the complex software landscape that there is today. br / br /3. People still care a lot about software distributions. Not that I didn't care, but in the past I burned my fingers with it. I was a very initial Linux user, with SlackWare 3.1 or something like that. During the years I wrote my device drivers, contributed a few patches to the kernel, during the years Debian had maybe ~10 packages of stuff that I wrote, from hpi
📄 昨天是相当充实的一天。意大利正值五一劳动节假期，所以上午我和三位朋友去埃特纳火山徒步了四小时。我热爱步行，编程时也常暂停工作去散步，再回到键盘前时双腿已多行几公里。在埃特纳徒步的体验非常美妙（埃特纳是欧洲最大的活火山，而我恰巧住在它山麓的卡塔尼亚）。 傍晚六点我回到家，发布了关于改用AGPL许可证的博客文章，随后开始关注各类评论、反馈和私信，并从中收获了一些新认知： 1. 抛开那些我认为至关重要的条款差异，AGPL与SSPL的核心区别在于AGPL是"被广泛理解"的。昨天我第一次深刻意识到，许可证不仅关乎"能做与不能做"，更涉及特定许可证被理解、验证和采用的程度…… 2. 西蒙·威尔逊对此事的观点（https://simonwillison.net/2025/May/1/redis-is-open-source-again/）让我深有感触。奇妙的是，居住在世界不同角落、年龄相仿且拥有相似软件背景的人们，对事物的感受竟能如此契合。我在开发Vector Sets时同样想过：如果它不以AGPL（或其他我理解的开源许可证）发布，我绝不会使用。这种共识在社区中形成不可忽视的声量，使得开源精神即便在当今复杂的软件生态中终将获胜。 3. 人们依然非常重视软件发行版。并非我过去不重视，而是曾在此栽过跟头——作为早期Linux用户（从SlackWare 3.1开始），这些年我写过设备驱动、为内核贡献过补丁，Debian也曾收录过我编写的约十个软件包（从hpi……
47. Redis is open source again
📰
📄 Five months ago, I rejoined Redis and quickly started to talk with my colleagues about a possible switch to the AGPL license, only to discover that there was already an ongoing discussion, a very old one, too. Many people, within the company, had the feeling that the AGPL was a better pick than SSPL, and while eventually Redis switched to the SSPL license, the internal discussion continued. br / br /I tried to give more strength to the ongoing pro-AGPL license side. My feeling was that the SSPL, in practical terms, failed to be accepted by the community. The OSI wouldn’t accept it, nor would the software community regard the SSPL as an open license. In little time, I saw the hypothesis getting more and more traction, at all levels within the company hierarchy. br / br /I’ll be honest: I truly wanted the code I wrote for the new Vector Sets data type to be released under an open source license. Writing open source software is too rooted in me: I rarely wrote anything else in my career. I’m too old to start now. This may be childish, but I wrote Vector Sets with a huge amount of enthusiasm exactly because I knew Redis (and my new work) was going to be open source again. br / br /I understand that the core of our work is to improve Redis, to continue building a good system, useful, simple, able to change with the requirements of the software stack. Yet, returning back to an open source license is the basis for such efforts to be coherent with the Redis project, to be accepted by the user base, and to contribute to a human collective effort that is larger than any single company. So, honestly, while I can’t take credit for the license switch, I hope I contributed a little bit to it, because today I’m happy. I’m happy that Redis is open source software again, under the terms of the AGPLv3 license. br / br /Now, time to go back to the terminal, to show Redis users some respect by writing the best code I’m able to write, and make Vector Sets more useful and
📄 五个月前，我重返Redis团队，很快就开始与同事们探讨是否可能转向AGPL许可证，结果发现这个话题早已在公司内部持续讨论了很久。许多同事都认为AGPL是比SSPL更优的选择，尽管Redis最终采用了SSPL许可证，但内部的讨论从未停止。 我尝试为支持AGPL许可证的一方增添更多力量。在我看来，SSPL在实践中并未获得社区认可——OSI组织没有接受它，软件社区也未将其视为真正的开源许可证。很快，这个观点在公司各个层级都获得了越来越多的认同。 坦白说：我真心希望自己为新型向量集合数据类型编写的代码能以开源许可证发布。开源软件创作已深深融入我的血液：职业生涯中我几乎从未编写过闭源代码。如今我已不再年轻，更不会改变这个初心。这或许有些天真，但我正是怀着巨大热情开发向量集合的，因为我知道Redis（以及我的新成果）将再次走向开源。 我理解我们工作的核心是改进Redis，持续构建一个优秀、实用、简洁、能随软件栈需求演进的系统。但回归开源许可证，才是让这些努力与Redis项目精神契合、获得用户认可、并贡献于超越任何单个公司的人类集体事业的基础。因此，尽管不能将许可证转换归功于己，我仍希望为此贡献了微薄之力——因为今天的我感到由衷喜悦。我为Redis在AGPLv3许可证下重归开源软件行列而欣喜。 现在，是时候回到终端前了。我将通过编写力所能及的最优质代码来表达对Redis用户的尊重，让向量集合功能更加强大实用……
48. Reproducing Hacker News writing style fingerprinting
📰
📄 About three years ago I saw a quite curious and interesting post on Hacker News. A student, Christopher Tarry, was able to use cosine similarity against a vector of top words frequencies in comments, in order to detect similar HN accounts — and, sometimes, even accounts actually controlled by the same user, that is, fake accounts used to uncover the identity of the writer. br / br /This is the original post: https://news.ycombinator.com/item?id=33755016 br / br /I was not aware, back then, of Burrows-Delta method for style detection: it seemed kinda magical that you just needed to normalize a frequency vector of top words to reach such quite remarkable results. I read a few wikipedia pages and took mental note of it. Then, as I was working with Vectors for Redis I remembered about this post, searched the web only to discover that the original page was gone and that the author, in the original post and website, didn’t really explained very well how the data was processed, the top words extracted (and, especially, how many were used) and so forth. I thought I could reproduce the work with Vector Sets, once I was done with the main work. Now the new data type is in the release candidate, and I found some time to work on the problem. This is a report of what I did, but before to continue, the mandatory demo site: you can play with it at the following link: br / br /https://antirez.com/hnstyle?username=pg&threshold=20&action=search br / br /NOTE: since the dataset takes 700MB of RAM, in my tiny server, in the next months I may take this down. However, later in this post you will find the link and the Github repository with the code to reproduce everything from scratch. br / br /NOTE2: I hope the web site will survive, it's a very crude Python script. I benchmarked the VSIM command in such a small server and yet it can deliver 80k VSIM per second! The wonders of int8 quantization, together with a few more optimizations. But the Python script is
📄 大约三年前，我在Hacker News上看到一个非常奇特有趣的帖子。一位名叫克里斯托弗·塔里的学生，通过计算评论中高频词向量的余弦相似度，成功识别出相似的HN账户——有时甚至能发现同一用户实际操控的不同账户，即用于掩盖作者真实身份的虚假账户。 这是原帖链接：https://news.ycombinator.com/item?id=33755016 当时我还不了解用于文体检测的伯罗斯-德尔塔方法：仅仅通过归一化高频词频率向量就能获得如此显著的结果，这看起来有些神奇。我查阅了几篇维基百科页面并默默记下了这个方法。后来，在使用Redis向量功能时，我想起了这篇帖子，上网搜索却发现原页面已消失，而且原作者在帖子和个人网站上并未详细说明数据处理过程、高频词提取方式（尤其是具体使用了多少词汇）等细节。我考虑在完成主要工作后，用向量集复现这项研究。现在新的数据类型已进入候选发布阶段，我也抽出时间研究了这个问题。以下是我的实践报告，但在继续之前，先提供必备的演示网站：您可以通过以下链接体验功能： https://antirez.com/hnstyle?username=pg&threshold=20&action=search 注意：由于数据集在我的小型服务器上占用700MB内存，未来几个月我可能会关闭该网站。不过，您可以在后文中找到完整复现所需的代码仓库链接。 补充说明：希望这个网站能保持运行，它只是一个非常基础的Python脚本。我在小型服务器上测试了VSIM命令，它每秒仍能处理8万次向量相似度计算！这得益于int8量化和多项优化技术。但Python脚本本身
49. Vector Sets are part of Redis
📰
📄 Yesterday we finally merged vector sets into Redis, here you can find the README that explains in detail what you get: br / br /https://github.com/redis/redis/blob/unstable/modules/vector-sets/README.md br / br /The goal of the new data structure is, in short, to create a new “Set alike” data type, similar to Sorted Sets, where instead of having a scalar as a score, you have a vector, and you can add and remove elements the Redis way, without caring about anything except the properties of the abstract data structure Redis implements, ask for elements similar to a given query vector (or a vector associated to some element already in the set), and so forth. But more about that later, a bit of background, first: br / br /From the path of the README itself, you can see the implementation is into “modules”, but actually, Vector Sets are not a module, it’s a part of the Redis core, the thing is that I started developing them as a module, and later I suggested that the implementation should still use the modules API, in order to promote modularity of the internals of Redis, in order to have both the advantages: every Redis instance starting from Redis 8 will have Vector Sets as a native data type, and there are clear boundaries between the core and the implementation br / br /## The first new main data type of Redis after… some time br / br /I think that the latest big data structure of Redis were Streams, also developed by me. I resigned, returned, forks happened in the meantime, and it still it looks like the burden to introduce a new data type in Redis is mine :D I must say: I’m ok with that, because as much as I like programming, I also like design, a lot, and I had a feeling, that vectors, and vector similarity, are conceptually very simple, so they deserved a very simple API. And that was what I tried to do. Vector Sets are still a beta feature but I can tell you something, I can guarantee you can learn the API in 3 minutes. br / br /I decided
📄 昨天我们终于将向量集合（Vector Sets）合并到Redis中，你可以在这里找到详细说明功能的README文档：
https://github.com/redis/redis/blob/unstable/modules/vector-sets/README.md
简而言之，这个新数据结构的目标是创建一种类似集合（Set）的新型数据类型——类似于有序集合（Sorted Sets），但不同之处在于：它不是用标量作为分值，而是使用向量。你可以用Redis惯用的方式添加和删除元素，只需关注Redis实现的抽象数据结构特性，无需考虑其他细节；可以查询与给定向量（或集合中已有元素关联的向量）相似的元素，等等。不过关于这些细节稍后再谈，先简单介绍一些背景：
从README文档的路径可以看出，实现位于“模块”中，但实际上向量集合并非模块，而是Redis核心的一部分。原因在于：我最初是将其作为模块开发的，后来建议实现仍应使用模块API，以促进Redis内部的模块化。这样既能兼顾两方面优势：从Redis 8开始，每个Redis实例都将原生支持向量集合数据类型，同时核心与实现之间也有清晰的边界。
## Redis时隔许久首次新增的主要数据类型
我认为Redis上一个重要数据结构是我同样开发的流（Streams）。这期间我经历了辞职、回归、项目分叉，但看起来在Redis中引入新数据类型的重任依然落在我肩上 :D 我必须说：我对此欣然接受，因为尽管我热爱编程，但也非常热衷于设计。我始终有种感觉：向量和向量相似度在概念上非常简单，理应配以极其简洁的API。这正是我努力实现的目标。向量集合目前仍是测试版功能，但我可以告诉你：我保证你能在3分钟内掌握它的API。
我决定
50. AI is useless, but it is our best bet for the future
📰
📄 I used AI with success 5 minutes ago. br / br /Just five minutes ago, I was writing a piece of software and relied on AI for assistance. Yet, here I am, starting this blog post by telling you that artificial intelligence, so far, has proven somewhat useless. How can I make such a statement if AI was just so helpful a moment ago? Actually, there's no contradiction here if we clarify exactly what we mean. br / br /Here’s the thing: at this very moment, artificial intelligence can support me significantly. If I'm struggling with complicated code or need to understand an advanced scientific paper on math, I can turn to AI for clarity. It can help me generate an image for a project, make a translation, clean my YouTube transcript. Clearly, it’s practical and beneficial in these everyday tasks. br / br /However, except for rare, groundbreaking examples like AlphaFold — Google's AI that significantly advanced our understanding of protein folding — AI has yet to genuinely push forward human knowledge in a fundamental way. Aside from these few exceptional results, AI hasn’t (obviously) yet matched the capabilities of the very best human minds. If an AI system were at the same level as the brightest humans (and not better than that: it's not needed for a first humanity jump) we could deploy millions of such systems to accelerate research dramatically, transforming progress expected to take centuries into developments happening within decades, or decades into years. br / br /Yet, if artificial intelligence remains stuck at its current level of development indefinitely (even if with small incremental improvements, enough to fire many translators, programmers, drivers, actors, ...), perhaps it might have been better not to have it at all. I mentioned this during a conference here in Sicily. The thought hadn't crossed my mind until I was asked on stage. While I was formulating my reply I asked myself: if we knew AI would only yield minor incremental improvements,
📄 就在五分钟前，我还在编写软件时成功借助了人工智能的帮助。然而此刻，我竟要以"迄今为止人工智能被证明有些无用"作为这篇博客的开场。既然AI刚刚才给予我帮助，又怎能说出这样的话？其实只要明确我们讨论的范畴，这两者并不矛盾。 关键在于：当下的人工智能确实能为我提供重要支持。无论是处理复杂代码难题，还是理解艰深的数学领域前沿论文，AI都能帮我理清思路。它可以为项目生成图像、完成翻译、整理YouTube视频字幕。显然，在日常事务中，AI确实实用且高效。 然而，除了像谷歌AlphaFold这种显著推动蛋白质折叠研究的突破性案例外，人工智能尚未在根本层面上真正拓展人类认知边界。除却少数特例，AI显然还未能媲美最卓越的人类智慧。倘若人工智能能达到顶尖人类的思维水平（无需超越，这已足够实现人类首次智力跃迁），我们便能部署数百万这样的系统，将原本需要数个世纪的研究进程压缩至数十年，或将数十年的探索缩短为几年。 但若人工智能永远停滞在现有发展阶段（即便有小幅渐进式改进，足以取代许多翻译、程序员、司机、演员等职业），或许它的存在反而弊大于利。我在西西里岛的一场会议上谈及此观点，这原本并非我既定想法，而是在台上被提问时突然意识到：如果我们早知AI只能带来微小的渐进式改进……