苹果起诉OpenAI,揭露前高管面试中带原型、离职下载千页机密文件,超过400名前员工已跳槽,技术窃密细节堪比谍战片。
苹果在2026年7月对OpenAI提起诉讼,指控前产品设计副总裁Tang Tan和前高级系统电气工程师Chang Liu窃取商业机密,包括利用苹果内部项目代号套取未公开产品细节,离职前下载包含上千页机密电路板制造文件的数据。诉讼称已有超过400名前苹果员工目前在OpenAI工作,并指控OpenAI谎称获得苹果授权使用其专有金属抛光工艺。苹果要求法院发出禁令并索赔,正值OpenAI加速开发首款消费硬件之际。
2026 07 12 HackerNews
2026-07-12 Hacker News Top Stories # 苹果起诉 OpenAI 指控其系统性窃取商业机密,包括套取产品细节、违规面试、离职时下载机密文件并谎称授权使用专有工艺。 纽约市将于 10 月 1 日起禁止欺骗性订阅,并要求广告标明含所有附加费的总价,预计每年为市民节省超 1.6 亿美元。 GPT-5.6 Sol Ultra 声称完成了图论中循环双覆盖猜想的证明,通过 8‑流定理与局部顶点调整构造出每条边恰好被两个圈覆盖的图。 布朗大学实验首次证实重元素中相对论效应使传统 σ/π 键图像失效,碳-铋键呈现一个 π 键加两个混合 σ-π 键。 住宅代理网络通过恶意软件劫持设备形成百万低频 IP 请求,迫使网站采用 PoW、CAPTCHA 等防御,LWN 正遭遇最严重爬虫攻击。 SpaceX 申请发射 10 万颗 Gen3 星链卫星以提供多千兆对称宽带,但当前实际速度约 145-170 Mbps,需 Starship 火箭发射。 报告显示博科圣地两派自 2024 年起系统使用 ChatGPT 等前沿 AI 辅助攻击策划、武器设计和知识传播,威胁比预期更紧迫。 本文从第一性原理阐释网络演进:从电报、分组交换到 IP/TCP、DNS、TLS,串联点击链接的全过程。 Ghost Font 利用动态点阵与运动噪声使人类可读而 AI 无法从单帧识别,实验显示先进模型常误读诱饵消息。 苏联控制室以大型按钮、模拟仪表和信号灯呈现计算机普及前的工业美学,切尔诺贝利控制室为其代表。 1. 苹果起诉 OpenAI,指控前员工窃取商业机密 (Apple sues OpenAI, accuses ex-employees of stealing trade secrets) # https://9to5mac.com/2026/07/10/apple-sues-openai-trade-secret-theft/ 苹果今日对 OpenAI 提起诉讼,指控其窃取商业机密,尤其是通过前苹果员工为 OpenAI 牟利。诉讼中列出的被告包括前苹果产品设计副总裁 Tang Tan、前高级系统电气工程师 Chang Liu,以及 OpenAI 和 io Products。 苹果在诉状中指出,Tang Tan 在面试苹果求职者时,利用苹果内部机密信息(如项目代号)套取更多未公开产品细节,还要求仍在苹果工作的候选人携带“真实零件”和“原型”到 OpenAI 面试中展示;Chang Liu 则在离职后利用安全漏洞下载了包含上千页机密电路板制造文件的数据,并指导正在招募的苹果同事准备面试时学习哪些机密材料。 苹果称,此举只是“冰山一角”,已有超过 400 名前苹果员工目前在 OpenAI 工作。诉讼还指控 OpenAI 利用苹果的信任,让一家苹果合作伙伴使用其专有金属抛光工艺,并向另一家苹果供应商打听具体组件信息。苹果要求法院发出禁令并索赔,正值 OpenAI 加速开发首款消费硬件之际。 HN 热度 1568 points | 评论 885 comments | 作者:stock_toaster | 1 day ago # https://news.ycombinator.com/item?id=48865019 OpenAI 教导离职员工如何规避苹果的离职审查,包括不告知去向以延长在职时间。 员工离职时通过电子邮件向自己发送苹果机密信息,苹果发现了这一模式。 OpenAI 利用苹果的机密硬件信息接触供应商,并谎称获得苹果授权以使用特定的商业秘密金属加工技术。 一名前员工保留苹果笔记本电脑,利用漏洞下载数十份机密文件,并在为 OpenAI 工作期间向仍任职的苹果同事炫耀。 这些行为显示出当事人自认为可以逃脱后果的傲慢心态。 长期受信任且富裕的员工之所以窃密,可能是受 OpenAI 的环境鼓励,或自身早已习惯此类行为。 此类行径反映了硅谷文化中对财富增长无止境的贪婪追求。 2. 纽约市将禁止欺骗性订阅行为 (New York City to ban deceptive subscription practices) # https://www.theguardian.com/us-news/2026/jul/10/new-york-city-deceptive-subscriptions-ban 纽约市消费者保护办公室宣布通过一项新规,禁止企业利用欺骗性订阅手段将顾客锁入健身房会员、流媒体服务等重复收费的陷阱。该规则将于 10 月 1 日生效,违规者可能面临每名用户 525 美元的罚款、追溯费用及额外处罚。同时,纽约市还拟出台一项针对“垃圾费用”的规则,要求卖家在广告中提前标明包括所有强制附加费在内的总价。如果这项租户相关规则通过,所有强制费用(包括年费)都必须计入标明的月租金中。这些举措是市长佐赫兰·马姆达尼及其团队积极遏制企业欺诈行为的一部分。订阅规则预计每年可为纽约市民节省 1.625 亿美元,而垃圾费用规则将影响酒店、租车公司等面向游客的企业。此前,拜登政府推出的全国性“一键取消”规则于 2025 年被联邦法官驳回,而特朗普政府的联邦贸易委员会计划在未来几个月通过类似规则。纽约市议会还提出了一项禁止“监控定价”的规则,即企业基于算法收集消费者信息而对同一商品或服务收取不同价格。马里兰州已在 4 月禁止该做法,科罗拉多州州长上月否决了相关禁令。纽约市将在公众意见征询和听证会后推进垃圾费用规则,官员希望在今年年底前完成。 HN 热度 620 points | 评论 324 comments | 作者:randycupertino | 1 day ago # https://news.ycombinator.com/item?id=48863464 质疑纽约市新规的实际执行效力,指出加州类似法律因餐馆豁免而效果打折。 餐馆菜单底部的服务费、小字附加费属于欺骗性定价,应直接标出最终价格。 德国和克罗地亚法律规定价格必须包含所有税费,小费自愿,与美国形成对比。 美国是少数不实行含税标价的发达国家,消费者往往无法提前知道最终应付金额。 德国小费习惯是直接告诉服务员想付的整数金额(含小费),无论现金还是刷卡。 游客若拒绝支付不合理费用,可通过叫警察调解或给差评施压,但前提是当地法律支持且执法严格。 该法规主要针对订阅式重复收费(如健身房、在线订阅),而非一次性服务费;但文章提到“垃圾费”禁令覆盖所有商品和服务。 全价提前标价在美国推行困难,因为“不知道最终价格”已成为文化习惯。 民调显示跨党派选民普遍支持禁止垃圾费,文化改变可能比预想容易。 美国联邦贸易委员会(FTC)已从去年起强制酒店实行全价标价。 3. GPT-5.6 Sol Ultra 生成了循环双覆盖猜想的证明 (GPT-5.6 Sol Ultra produces proof of the Cycle Double Cover Conjecture [pdf]) # https://cdn.openai.com/pdf/04d1d1e4-bc75-476a-97cf-49055cd98d31/cdc_proof.pdf 该文件声称证明了图论中的循环双覆盖猜想,即每个无桥无向图都存在一个边多重集,使每条边恰好被覆盖两次。证明由 GPT-5.6 Sol Ultra 完成,作者仅负责整理。 证明的标准简化:先归约为无环三次图(立方图)。利用已知的 8-流定理(等价于无处零 Γ-流,Γ 为 F₂³),给每条边赋予 Γ 中的非零元素,使每个顶点处三边之和为零。然后通过两个引理构造循环双覆盖。 引理 2.1:若能将每条边赋予 Γ 的一个二元子集 P_e,使得在每个顶点处,Γ 中每个元素出现在邻接边的 P_e 中的次数为 0 或 2,则这些二元子集按元素取出对应边集,每个边集是若干圈,且每条边属于两个这样的集合,从而构成循环双覆盖。 引理 2.2(核心):给定一个无处零 Γ-流 f,通过在每个顶点局部定义辅助量 g(令三条边中某两条的 g 为 0,另一条为 f(该边)),得到两个端点处可能不同的临时二元集合。为使两端一致,需解一个线性方程组 t_u + t_v + ε_e f(e) = d_e(其中 d_e 为两端 g 之差,ε_e∈F₂)。引理 2.2 证明该方程组总有解,方法是对偶空间论证:假设存在线性泛函族 η_e 使得对任意解左边为零,则必须满足 η_e(f(e))=0 且每个顶点处 η_e 之和为零;进而证明对这些 η_e,所有 η_e(d_e)之和为 0,故 d 在映射的值域中,方程组可解。 最后,通过解出的 t 和 ε 调整局部集合,使每条边两端的二元集合一致,得到满足引理 2.1 条件的分配 P_e,从而完成证明。 文中提到,对于平面图、三边可着色立方图等情形已有已知结果,新证明借助 8-流和对偶线性代数给出一般性构造。 HN 热度 510 points | 评论 422 comments | 作者:scrlk | 1 day ago # https://news.ycombinator.com/item?id=48863490 即使最先进的模型也需要大量提示指令才能解决问题,策略应模型自行推导。 模型的后训练使其倾向于简短回答,深度推理需额外提示,模型没有意志。 低温和推理模式可使模型自我纠正,但 SaaS 可能不适用长推理。 模型没有意志是好事,保持工具属性;也有观点质疑不将智能体计划视为意志。 LLM 必然产生幻觉,因为语言存在矛盾,模型只选一个答案,幻觉是基础性质。 提示词中使用思维树方法是因为模型缺乏好奇心,需人为指引。 4. 新研究显示:爱因斯坦的相对论主导着重元素的化学键 (Einstein’s relativity rules chemical bonds in heavy elements, new research shows) # https://www.brown.edu/news/2026-07-09/chemical-bonds-relativity 布朗大学化学家首次提供直接实验证据,证明在重元素中,传统的三键教科书解释不再成立。爱因斯坦的相对论改变了三键结构,模糊了 sigma 键和 pi 键之间的严格界限。 研究团队利用光电子能谱技术,分析由碳和重元素铋形成的分子。结果显示,碳-铋键并不符合传统的一个 sigma 键加两个 pi 键的图像,而是呈现出一个 pi 键和两个混合的 sigma-pi 键。 这一发现可能促使化学教科书重写,尤其是随着铋等重元素在下一代太阳能电池、量子材料和量子计算研究中的兴趣日益增长。相关成果发表于《科学》杂志。 HN 热度 381 points | 评论 169 comments | 作者:hhs | 1 day ago # https://news.ycombinator.com/item?id=48866134 汞在室温下是液态,因为其内层电子以接近 60% 光速运动,相对论效应使外层电子更紧密,难以成键形成固体。 相对论效应影响其他第 6、7 周期元素,例如金的黄色、铯的颜色以及铂的催化性能,都与相对论轨道收缩有关。 相对论效应导致重金属元素化学键复杂化,自旋-轨道耦合使 σ 键和 π 键不再独立。 化学教育中许多概念被手写解释,缺乏物理视角的直观理解,实际计算非常复杂。 理论化学不如实验重要,因为化学需要处理微观相互作用的多样性,理论存在局限性。 质子内部夸克运动速度接近光速,但夸克没有量子能态跃迁,因此不会像电子那样辐射光子;电子在基态不辐射是因为没有更低能级。 质子质量的 90% 以上来自运动夸克的动能,而非静质量;希格斯机制赋予夸克质量,但质子质量主要源于相对论效应。 单个震荡电子并不连续发射无线电波,它有微观概率在某个周期发射单个射频光子,需借助量子电动力学解释。 5. 关于住宅代理和爬虫情况的更新 (An update on residential proxies and the scraper situation) # https://lwn.net/SubscriberLink/1080822/990a8a5e2d379085/ 本文是 LWN.net 上的一篇更新文章,讨论 AI 爬虫机器人对网站的持续攻击问题。文章指出,尽管 2025 年初已有相关报道,但爬虫攻击仍在加剧,大量流量来自住宅代理网络——即通过恶意软件或“免费 VPN”等方式劫持普通用户的设备,使其成为攻击端点。这些攻击来源包括纯粹的犯罪团伙和伪装合法的公司(如 Bright Data),它们从数百万个 IP 地址发起请求,每个地址仅访问一两次,传统封禁手段无效。此外,还有大型 AI 公司直接抓取,但通常会遵守 robots.txt,并非最大问题。真正幕后付费使用住宅代理网络的客户身份不明,可能包括秘密政府机构、犯罪组织和内部模型开发者。 为应对攻击,网站运营者不得不采取各种防御措施,如工作量证明(Anubis)、人机验证、登录墙或付费墙,以及数据毒化工具等。这些措施给网站和用户都带来了沉重负担。LWN 本身也经历了有史以来最严重的爬虫攻击,但由于已实施的防御措施,多数读者并未察觉。文章强调这是一场军备竞赛,但未详细透露其具体防御手段。 HN 热度 334 points | 评论 350 comments | 作者:chmaynard | 1 day ago # https://news.ycombinator.com/item?id=48864252 用户更愿意等几秒钟的 PoW,而不是解 CAPTCHA,因为 CAPTCHA 更恼人且消耗注意力。 PoW 的延时对真实用户可接受(如 Anubis),而 CAPTCHA 的误报率和阻隔感已经很高。 有观点认为 PoW 不具可扩展性:用廉价 CPU 换取昂贵用户延迟,对爬虫构不成有意义威慑。 相反观点:Anubis 实际部署证明 PoW 成功,用户已习惯慢网站,PoW 将爬虫计算成本提升数个数量级。 爬虫通常针对特定 URL 而非随机,PoW 足以迫使他们放弃或增加成本。 有人提出可用内存约束替代 CPU,因为内存更昂贵,但数据中心内存充足而用户端有限。 反驳:浏览器环境下 100MB×1 秒不足以威慑爬虫,且手机等低端设备难以承受更高要求。 Google 的 CAPTCHA 在用户操作前已根据鼠标轨迹等启发式判定成败,实际验证仅用于收集训练数据。 被 Google 误判为用户会被无限循环 CAPTCHA,而无障碍音频版会直接告知已封禁。 希望将类似 Anubis 的 PoW 用于电子邮件防护,正如 HashCash 最初的设计。 6. SpaceX 希望再发射 10 万颗星链卫星,将带宽提升 100 倍 (SpaceX wants to launch 100k more Starlink satellites for 100x the bandwidth) # https://www.zdnet.com/home-and-office/networking/spacex-wants-to-launch-100000-more-starlink-satellites/ SpaceX 向美国联邦通信委员会(FCC)申请发射 10 万颗第三代(Gen3)Starlink 卫星。该公司承诺,该星座将提供超低延迟、多千兆对称宽带,但当前 Starlink 实际下载速度约为 145-170 Mbps。Gen3 卫星每颗重超 2 吨,需借助 Starship 火箭发射。该网络将服务于消费者、企业、政府及“数十亿 AI 驱动设备”。对手如亚马逊 Project Kuiper 等竞争激烈。 HN 热度 296 points | 评论 1103 comments | 作者:CrankyBear | 1 day ago # https://news.ycombinator.com/item?id=48863064 SpaceX 声称需要 10 万颗卫星是为了支撑公司不合理的估值 马斯克更像科幻作者而不是 CEO,类似 L. Ron Hubbard Starlink 在很多地方带宽饱和,确实有需求 苹果也曾宣布未发货的产品,不应因此指控误导投资者 特斯拉有长期撒谎历史,不同于苹果 马斯克做出纳粹手势,言论煽动种族对立 看到星链卫星改变夜空让人沮丧,一人私有化天空 应该考虑星链的环境影响 环境影响定义因人而异,可能被滥用阻碍项目 放松监管导致 NIMBY 问题更严重 有人赞赏能够突破官僚实现建设 如果马斯克给出诚实概率估计,看法会不同 特斯拉自动驾驶出租车(robotaxi)承诺不可能实现 比特币无价值却能维持高价,SpaceX 也可能泡沫 使用激光雷达才可能实现自动驾驶,特斯拉没机会 Siri 承诺未兑现面临诉讼,与特斯拉类似 评论中未看出政治偏见,需明确指出来源 7. 恐怖组织博科圣地如何利用前沿 AI (How the terrorist group Boko Haram uses frontier AI) # https://casp.ac/reports/ai-enabled-terrorism CASPC(剑桥 AI 科学与政策项目)发布的研究报告《“上帝帮助了我们,AI 也会”:恐怖组织博科圣地如何利用前沿 AI》揭示了前所未有的细节:通过对尼日利亚东北部 27 名前博科圣地成员在 2025-2026 年进行的半结构化访谈,发现该组织的两个派系在 2024 年期间系统性地使用 ChatGPT、Claude、Gemini、Grok、Meta AI 和 DeepSeek 等前沿 AI 辅助作战与日常运作。这种使用已通过专门小组和内部培训制度化,帮助攻击策划、武器故障排除和爆炸装置设计,用户成功绕过了部分安全防护。知识通过跨国圣战网络传播,伊斯兰国人员提供了现场培训。受访者对 AI 表现出强烈热情,部分对大规模杀伤性武器持开放态度(尽管记录到的使用仍属常规武器)。报告认为,恐怖分子对 AI 的采用比先前认知更深入、更系统,已成为当前且不断增长的现实,需引起政策制定者、安全界和 AI 开发者的关注。 HN 热度 225 points | 评论 195 comments | 作者:imustachyou | 1 day ago # https://news.ycombinator.com/item?id=48863707 博科圣地利用 AI 学习摩托车跳跃技术导致 18 人死亡,8 人成功,这一过程令人质疑其真实性。 提供可能被恐怖分子用于训练的信息可能构成法律上的“物质援助”。 用水坑训练更安全但会损坏摩托车,而恐怖组织更看重结果而非训练者伤亡。 美军大规模使用 LLM 与恐怖组织类似,这种对比讽刺地让人对前者感到“放心”。 训练死亡率高时,不如直接派未训练者参战(如《第二十二条军规》的情节)。 恐怖分子因信仰(如相信来世)而视死如归,因此残酷训练环境对他们并非阻力。 AI 安全限制可能被绕过,使用无限制模型反而更危险,但监管需权衡伦理与实效。 有美国国会议员声称美国资助过博科圣地,但缺乏证据且其个人信誉存疑。 关于恐怖分子动机的讨论:现代自杀式爆炸先驱包含多宗教甚至无神论成员,与宗教狂热并不完全绑定。 评论区充斥对训练细节的讽刺,如用玻璃和火模拟实战、建议从.22 子弹开始练抗性等。 8. 网络与互联网:从第一性原理出发 (Networking and the Internet, from First Principles) # https://fazamhd.com/mental-models/networking/ 这篇文章从第一性原理出发,深入浅出地解释了互联网的运作机制。它从电报时代开始讲起,说明数字信号(离散符号 + 再生)比模拟信号更适合远距离传输,并引入了“协议”这一核心概念。随后,文章依次介绍了电路交换的局限性、分组交换的诞生、第一个分组网络(ARPANET)的架构,以及如何将多个网络连接成互联网。重点剖析了 IP 和 TCP 的分工——IP 负责尽力传递数据包,TCP 负责保证可靠性;还解释了路由表的来源、DNS 域名系统的作用,以及 TLS 加密如何保护通信安全。最后,用一个点击链接的完整例子,串联起从输入 URL 到页面呈现的全过程。整篇文章通过历史演进和具体问题驱动的视角,帮助读者形成关于互联网如何工作的连贯心智模型。 HN 热度 195 points | 评论 58 comments | 作者:faza | 12 hours ago # https://news.ycombinator.com/item?id=48871470 文章内容翔实、结构清晰,讲述网络发展史引人入胜,即便疑似 LLM 所写也不影响其质量。 首帧动画应包含消息平台服务器,或用 P2P 应用举例更贴切。 有人调侃 LLM 写作风格像“克林贡语或精灵语”般成为新语言。 文章可能由作者撰写,但经 AI 大幅润色(编辑痕迹明显),指出这点并非偏执。 数字 PDP-1 的首批订单就用于 ITT 的纸带消息系统,历史相关。 感谢作者的努力,浓缩了多年职业经验,期待更多类似分享。 交互元素设计精良,让人联想到 Bartosz Ciechanowski 的作品,询问所用技术栈。 加载后开启飞行模式会导致未进入视野的动画无法播放,是个奇怪问题。 不理解为何有那么多负面评论,内容有趣且易懂。 评论中有些被标记[dead]的平淡留言,怀疑是机器人。 9. 幽灵字体:一种人类可读但 AI 无法识别的字体 (Ghost Font: A font that humans can read but AI cannot) # https://www.mixfont.com/ghost-font Ghost Font 是一种反 AI 字体,通过运动中的点阵在视频中显示文字,人眼可以看到,但截图或单帧图像无法识别。它结合了运动、噪声和诱饵消息,防止 AI 模型解码。实验显示,即使是最先进的 AI 模型,如 GPT-Sol 5.6 Ultra 和 Claude Fable,也常误读诱饵消息而忽略真实信息。 该项目借鉴了 2013 年 ZXX 字体的理念,但 ZXX 现在已被 AI 轻松识别。Ghost Font 则通过视频动态和诱饵层增加难度。未来可能用于验证码或 AI 视觉基准测试。作者计划开源视频生成代码,并继续改进,支持更长的文本。 HN 热度 191 points | 评论 142 comments | 作者:justswim | 15 hours ago # https://news.ycombinator.com/item?id=48870381 认为虽然字体坚固,但最终解决方案是认证(attestation) 想看到仅 AI 可读的反向字体 通过视频压缩和 OCR 可恢复文字,并非根本性更难 用截图加滤镜就能提取文字,AI 用连续帧更易识别 可替代 CAPTCHA,但缺少用户数据训练的正面副作用 谷歌 CAPTCHA 已不再用于训练,转而用于实时驾驶决策 实际文字难以阅读,类似魔法眼 3D 图;但也有人能轻松阅读,而诱饵文字几乎不可见 设备类型和缩放比例影响可读性 翻转手机改变图片大小即可切换可读文字 效果依赖算法滤镜而非人类感知 多帧样本让 AI 更易阅读,对人类反而更困难 有人只看到诱饵文字,怀疑自己是 AI GPT-5.6 通过时间分析无问题读出文字 Fable 模型也轻松解码,但读出了错误的诱饵文字 人类可能比模型更笨,有人戏称已达到 ASI(人工超级智能) 这种字体对足够好的 AI 更可靠,并非所有人类都能阅读 10. 苏联控制室的复古之美(2018) (The vintage beauty of Soviet control rooms (2018)) # https://designyoutrust.com/2018/01/vintage-beauty-soviet-control-rooms/ 这是一篇关于苏联时代控制室怀旧魅力的文章。文中展示了一系列珍贵的老照片,这些控制室充满大型按钮和模拟拨盘,其设计风格远在计算机和屏幕普及之前。文章特别提及了切尔诺贝利 4 号反应堆控制室的照片。 HN 热度 190 points | 评论 64 comments | 作者:mvdtnz | 19 hours ago # https://news.ycombinator.com/item?id=48868996 这种控制室美学并非苏联独有,任何计算机出现前的控制室都类似,如法国 70 年代的核电站控制室。 苏联控制室确实有独特的设计感,比如偏爱海泡绿或青色背景以营造冷静氛围。 老式控制室使用颜色区分功能,但三哩岛事故暴露了过多红色按钮导致的混乱。 核电站操作员训练有素,在紧急情况下会使用精确指令而非喊“按红色按钮”。 良好的人机工程学设计如啤酒龙头手柄、飞机起落架轮子能辅助快速识别。 真实手术室高效冷静,不会像电视剧那样大喊大叫。 老式控制室空间宽敞是因为需要容纳大量线缆并满足散热需求。 苏联/俄罗斯的控制面板颜色选择经长期研究,旨在缓解压力。 国际空间站俄语版页面也采用偏蓝色配色,延续了这一风格。 法国核电站操作员穿牛仔裤,苏联操作员则穿白外套和海军帽。 这种控制室美学让人联想到《Playtime》《Maniac》等电影或电视剧的视觉风格。 Hacker News 精彩评论及翻译 # Apple sues OpenAI, accuses ex-employees of stealin… # https://news.ycombinator.com/item?id=48865294 Some pretty damning stuff: OpenAI also instructs new hires on how to avoid scrutiny when they leave Apple. For example, Mr. Tan warns them not to tell Apple that they have taken jobs at OpenAI, so they can stay at Apple as long as they can. Apple says it discovered a pattern of OpenAI recruits emailing themselves confidential information when leaving Apple, including Tan. OpenAI apparently used confidential Apple hardware information when approaching Apple suppliers, and tricked one company into using a “specific trade secret metal-finishing technique” for an OpenAI device by claiming it had Apple’s permission to do so. Liu allegedly kept an Apple-issued laptop after departing the company and exploited a vulnerability to download dozens of confidential Apple documents while he was working at OpenAI. Non-competes and the like are gross but what’s described here isn’t just “bring your expertise to OpenAI” it’s “here is how to steal secrets on your way out” which is even grosser. joshstrange 一些相当有力的指控: OpenAI 还指导新员工如何在他们离开苹果公司时避开审查。例如,谭先生警告他们不要告诉苹果自己已在OpenAI任职,以便他们能尽可能久地留在苹果。 苹果表示,他们发现了一类模式:OpenAI招募的人员在离开苹果时会给自己发送机密信息,包括谭在内。 OpenAI 显然在接触苹果供应商时使用了苹果的机密硬件信息,并通过声称已获得苹果许可,欺骗一家公司为OpenAI的设备采用“特定的商业秘密金属精加工技术”。 据称,刘在离开苹果后保留了一台苹果发放的笔记本电脑,并在OpenAI工作期间利用漏洞下载了数十份苹果的机密文件。 竞业限制之类的做法已经够恶心了,但这里描述的不仅仅是“把你的专长带到OpenAI”,而是“教你如何在离职时窃取机密”,这更加令人不齿。 Apple sues OpenAI, accuses ex-employees of stealin… # https://news.ycombinator.com/item?id=48866774 It gets even worse. The person not only kept the laptop and used an exploit to download confidential Apple documents, they bragged about it to a contact who was still working at Apple who was also feeding him information: Liu allegedly kept an Apple-issued laptop after departing the company and exploited a vulnerability to download dozens of confidential Apple documents while he was working at OpenAI. He also maintained a relationship with Yu-Ting “Alyssa” Peng, an Apple employee who continued to give him updates on Apple’s projects, vendor decisions, and engineering details. When Liu learned he still had access to Apple’s systems, he texted Peng “LOL, I found out I can access the [network storage], so funny.” This is how you behave when you think you’re so much smarter than everyone around you that consequences don’t apply to you. Whenever I leave a company I make sure everything that belongs to the company goes back to them and I wipe any access credentials or authenticator codes that might be on any of my devices. I can’t imagine being so brazen that you’d keep the company laptop and then start using an exploit to download confidential information for your new employer. Doing it at a the company that most aggressively enforces secrecy is even crazier. Aurornis 情况甚至更糟。此人不仅留下了公司配发的笔记本电脑,还利用漏洞下载了苹果的机密文件,甚至向一位仍在苹果工作、同时也在向他提供情报的联系人吹嘘此事: 刘在离开苹果公司后据称保留了一台公司配发的笔记本电脑,并在OpenAI工作期间利用漏洞下载了数十份苹果机密文件。他还与苹果员工彭钰婷(Alyssa)保持着联系,后者持续向他提供苹果项目的进展、供应商决策以及工程细节。当刘发现自己仍能访问苹果系统时,他给彭发信息说:“哈哈,我发现我还能访问[网络存储],太搞笑了。” 这就是当你觉得自己比周围所有人都聪明、以至于后果对你不起作用时的表现。 每次我离开一家公司,我都会确保所有属于公司的物品归还,并清除可能留在我任何设备上的访问凭证或验证码。我无法想象有人会如此厚颜无耻,不仅留着公司电脑,还利用漏洞为新雇主下载机密信息。 在苹果这样以严格保密著称的公司干这种事,更是疯狂至极。 SpaceX wants to launch 100k more Starlink satellit… # https://news.ycombinator.com/item?id=48869148 I have started to see what I think are star link satellites at night on walks with my kids. It actually makes me sad to see that on person owns the night sky and is changing the literal stars my kids will grow up with. It feels different when it’s the government that theoretically represents people but when it’s one person that feels truly depressing. digitaltrees 我晚上和孩子散步时开始看到一些我觉得是星链卫星的东西。看到有人独自拥有夜空,并改变了我孩子们成长过程中将看到的真正的星星,这让我感到难过。如果是理论上代表人民的政府这样做,感觉还不一样,但如果是某一个人,那就真的令人沮丧。 AI 2040: Plan A # https://news.ycombinator.com/item?id=48870145 This is religious fervor folks, as AI 2027 was. I grew up in evangelical christianity, and to them the end of the world is just around the corner, the same way it has been since I was a small child and likely will be when we are all gone. This isn’t science. This isn’t hypothesis experiment record results. This is very expensive astrology, shiny rock collecting, ritualistic meaning-making and self-justification. Yall, with your incredible wealth and resources you could do real good in this world and make society better, healthier, better educated, and the whole world more equal, just, and reduce the desperation and suffering. Reject the false and self-serving narratives that empathy doesn’t matter, that altruism isn’t “effective”. You can change a person’s whole life in a moment. taurath 这就是宗教狂热,各位,就像当年的AI 2027一样。 我在福音派基督教环境中长大,对他们而言,世界末日近在咫尺——从我小时候起就一直如此,大概等我们所有人都离世后也还会是这样。这不是科学。这不是提出假设、实验、记录结果。这是一种极其昂贵的占星术、收集闪亮石头的把戏、仪式化的意义建构与自我正当化。 你们这些拥有惊人财富和资源的人,本可以在这个世界上真正行善,让社会变得更好、更健康、受教育程度更高,让整个世界更加平等、公正,减少绝望与苦难。拒绝那些虚假且自私自利的叙事——说什么共情无关紧要、利他并不“有效”。你可以在一个瞬间改变一个人的一生。 Apple sues OpenAI, accuses ex-employees of stealin… # https://news.ycombinator.com/item?id=48869205 OpenAI is a company built on copyright violation. That means it’s in the corporate DNA to treat laws as things for little people. Apple have deep enough pockets that they can actually sue OpenAI but I bet OpenAI are surprised they got caught. Now ask yourself, would the Codex agents on your machine ever over step legal boundaries? Would OpenAI ever make use of data you, voluntarily, send to their servers? If they did could your company afford to sue OpenAI and would it still be too late to save the business? Lio OpenAI是一家建立在版权侵犯基础上的公司。 这意味着将法律视为只适用于小人物的规则,已经深植于其企业基因中。 苹果公司财力雄厚,确实可以起诉OpenAI,但我打赌OpenAI对自己被抓住感到意外。 现在问问自己,你设备上的Codex代理是否会越界违法?OpenAI会不会利用你自愿发送到他们服务器的数据? 如果他们真的这么做了,你的公司能承担得起起诉OpenAI的费用吗?而且到那时再挽救业务是否已经为时已晚? How the terrorist group Boko Haram uses frontier A… # https://news.ycombinator.com/item?id=48863983 We saw in a movie how motorcycles can jump over bridges. We used AI to learn how to do this. We gave it information, like what motorcycles we use and the distance we need to jump and so on and it gave us steps on what we have to do. We practiced a lot and kept asking questions. We dug holes and filled them with broken glass and fire to practice. 18 of us died in the process. Eight of us managed to do it. The next time we attacked, we could jump. Now listen, I’m not saying we need to give these guys more AI, but it clearly isn’t yielding bad outcomes for us here. “You’re absolutely correct! For it to be a good practice ground you need to fill the trenches with broken glass and light the whole thing on fire” arjie 我们在电影里看到摩托车能飞跃桥梁。我们利用AI学习如何做到这一点。我们向它提供信息,比如我们使用的摩托车型号、需要跳跃的距离等等,它给出了我们需要采取的步骤。我们反复练习并不断提问。我们挖坑,填满碎玻璃和火来进行训练。过程中有18人丧生。我们中有8人成功做到了。下一次进攻时,我们就能跳过去了。 听着,我不是说我们需要给这些家伙更多AI,但显然它对我们来说结果并不差。 “你说得太对了!要建立一个良好的训练场,你需要在战壕里填满碎玻璃,然后把整个东西点燃。” EU Parliament greenlights Chat Control 1.0 # https://news.ycombinator.com/item?id=48857867 The EU is a farce, an undemocratic virtue signaling organization, and this is why: The Parliament voted against the first reading of this proposal twice in 2026, the first time they only supported limited cases for it, while the second time they actually defeated it fully. The Commission didn’t care, and kept the proposal on the table by refusing to withdraw it. Once the Commission does that, the proposal goes on second-reading (despite the first-reading having defeated it) and it is established in a very PERVERSE way in EU law that to AVOID passing the proposal in second-reading you need ABSOLUTE majority which is incredibly hard to pursue (you would think that we would need an absolute majority to PASS a proposal that was previously defeated on first-reading, not instead needing absolute majority to DENY a previously defeated proposal that was again forced to the table). Furthermore, absences in practicality count as “No” on the rejection. So of course they scheduled the vote in the summer when notoriously there will be many absences. By never withdrawing a defeated proposal they can effectively and in practicality pursue any agenda they want (it requires a massive mobilization effort to find absolute majority to defeat any proposal, especially when absences for any reason effectively count against rejection). In PRACTICALITY, the Commission can pursue any agenda whenever and however they want, and throw the votes down the drain. EU’s democracy is lipstick on a pig. fosk 欧盟就是一个笑话,一个不民主的作秀组织,原因如下: 欧洲议会在2026年曾两次否决该提案的一读:第一次他们仅支持有限情况,第二次则完全否决了它。 欧盟委员会毫不在意,拒绝撤回提案,使其始终悬而未决。 一旦委员会这样做,提案就会进入二读(尽管一读已经否决了它),而欧盟法律以一种极其扭曲的方式规定:要避免提案在二读中通过,需要绝对多数票——这几乎不可能实现(你可能会认为,要强行通过一个此前已被否决的提案才需要绝对多数,但现实却是,要否决一个已被否决、却被再次强推的提案,反而需要绝对多数票)。 此外,缺席者在实践中被视为对否决投了反对票。所以他们理所当然地把投票安排在夏天——众所周知届时会有大量缺席情况。 通过从不撤回已被否决的提案,他们实际上可以随心所欲地推行任何议程(要动员足够力量获得绝对多数来否决一个提案极其困难,尤其是当任何原因的缺席都会实际等同于反对否决时)。 实际上,欧盟委员会可以随时、以任何方式推行任何议程,让投票形同虚设。 欧盟的民主就是给猪涂口红。 After 7 years in production, Scarf has reluctantly… # https://news.ycombinator.com/item?id=48860109 Wow. Not a Haskell user, but a big user of other languages with expressive type systems (mostly Scala; some Rust). My experience is the complete opposite. I can’t imagine using a language without a good type system to catch all the junk the LLM produces. In fact I thought people would move away from languages from poor type systems, like Python, given the cost of using languages with expressive type systems has decreased with LLMs. noelwelsh 哇。我不是Haskell用户,但经常使用其他具有表达力类型系统的语言(主要是Scala;还有一些Rust)。我的体验完全相反。我无法想象使用一种没有良好类型系统来捕捉LLM产生的所有垃圾的语言。事实上,我原以为人们会远离像Python这样类型系统薄弱的语言,因为随着LLM的出现,使用具有表达力类型系统的语言成本已经降低了。 Good Tools Are Invisible # https://news.ycombinator.com/item?id=48859332 Having designed a good number of internal tools for teams of developers I couldn’t agree more. Earlier I had the tendency to “leave the guts” open, thinking my users were developers and would want that. All it did was put obstacles in my teammates actually doing their work. My teammates must use the tools I made for them to achieve work the company needs them to do, they don’t want, nor should they want to, fiddle with a little tool they won’t find anywhere else. I still leave a lot of escape hatches, but I try to design the internal tools in such way as to make the users fall into a pit of success. Edit: also, error messages, error messages, error messages and auto suggestions for common errors Edit 2: also the number of people only addressing the examples in the post rather than the spirit of the post is… disappointing. jrimbault 为开发团队设计过不少内部工具后,我对此深表赞同。 以前我总倾向于"暴露内部机制",想着用户都是开发者会需要这个。结果这反而给队友完成实际工作制造了障碍。我的队友必须使用我为他们打造的工具来完成公司要求的工作——他们不需要、也不应该需要去摆弄一个别处找不到的小工具。 我依然保留了大量逃生舱口,但会尽量将内部工具设计成让用户能"落入成功的陷阱"。 补充:还有错误提示、错误提示、错误提示,以及常见错误的自动建议。 再补充:看到很多人只针对帖子里的例子而非核心精神展开讨论……实在令人失望。 SpaceX wants to launch 100k more Starlink satellit… # https://news.ycombinator.com/item?id=48863244 When Starlink first became available here in poor-ish Central-EU, I was excited. Then, only months later, but after years of planning: EU funding brought fiber to my farm area, at ~$25/900mbps 10ms. While my story is just n=1, I don’t understand the huge upside for Starlink outside of Africa or India, where they have <.1% the money to spend on such things. However, I am dumb, and very open to be convinced. consumer451 当星链最初在相对贫穷的中欧地区开通时,我很兴奋。然而仅仅几个月后——但经过多年规划——欧盟的资金将光纤带到了我的农场区域,约25美元/900兆带宽/10毫秒延迟。 虽然我的例子只是个案,但我无法理解星链在非洲或印度以外的地区能有多大优势,毕竟那些地方只有不到我们0.1%的资金来投入这类服务。 不过,我很笨,也很愿意被说服。 Female US rower completes historic solo journey fr… # https://news.ycombinator.com/item?id=48874052 It’s kind of buried here, but Kelsey is the fastest human to do this. She beat the male record holder’s time by 6 days. CharlesW 这里有点被埋没了,但凯尔西是做这件事最快的人。她比男性纪录保持者的时间快了6天。 QuadRF can spot drones and see WiFi through my wal… # https://news.ycombinator.com/item?id=48864276 QuadRF creator here. Happy to answer questions! We have a quick demo video as well: https://m.youtube.com/watch?v=QvniJk3uNyA Along with a deeper dive video: https://m.youtube.com/watch?v=zdJ9Tbm8ALg We didn’t give Jeff great direction on camera alignment calibration or setting the radio gain but he seemed to mostly figure it out. We’re improving the UI based on his suggestions (it’s open source so you can customize it too) The RF augmented reality is just one of many applications of this brand new 4x4 MIMO software-defined radio built from the ground up. The AR uses a web app to stream RF points that your phone/laptop browser then live-merges with your local camera in the browser. I’ve been obsessed with low latency and high frame rate to make it a truly AR experience. More technical details at https://QuadRF.com/ mrtnmcc 我是QuadRF的开发者。欢迎提问! 我们还有一个快速演示视频:https://m.youtube.com/watch?v=QvniJk3uNyA 以及一个更深入的讲解视频:https://m.youtube.com/watch?v=zdJ9Tbm8ALg 在摄像头对齐校准或无线增益设置方面,我们没能给Jeff提供明确的指导,但他似乎基本自己摸索出来了。我们正根据他的建议优化用户界面(该软件是开源的,你也可以自定义)。 射频增强现实只是这款全新自主研发的4x4 MIMO软件无线电的众多应用之一。该AR功能通过一个网页应用流式传输射频数据点,你的手机或电脑浏览器会将这些数据点与本地摄像头画面实时融合。我一直痴迷于低延迟和高帧率,力求打造真正的AR体验。更多技术细节请访问https://QuadRF.com/ Why American ambulance rides are so expensive # https://news.ycombinator.com/item?id=48856363 Society should find a better way to pay for ambulances Society has this figured out, at least a decent solution that works until we find a perfect one. Only the US society seems to be unable to find a solution. otherme123 社会应该找到更好的方式来支付救护车费用 社会已经找到了解决办法,至少是一个可行的、不错的方案,直到我们找到完美的方案。只有美国社会似乎还没能找到解决方案。 Apple sues OpenAI, accuses ex-employees of stealin… # https://news.ycombinator.com/item?id=48868482 This is basically the end of OpenAI hardware. This is by far worst than the Waymo vs. Uber lawsuit which killed the Uber self driving project. Also if you are a business using OpenAI models, I would highly suggest you do not because they are most likely looking at your code and IP. impulser_ 这基本上标志着OpenAI硬件的终结。这比导致Uber自动驾驶项目夭折的Waymo诉Uber案还要糟糕得多。另外,如果你是一家使用OpenAI模型的企业,我强烈建议你不要这么做,因为他们很可能在窥探你的代码和知识产权。 Modern decor may be straining people’s brains # https://news.ycombinator.com/item?id=48874022 I had a discussion regarding this some time ago with my grandchild who has an ADHD diagnosis. She has troubles being in noisy (especially visually) environments, yet she finds my home (relatively large home full of books, music always playing etc) comforting. She explained that all this stuff in my home is interesting for her and speaks with her - “It’s you and grandma, it’s full of stories”. But the very modern and “must be comforting” environment in school full of patterns and pictures drawn on walls etc is just irritating – “There is no stories, just noise”. obscurette 不久前,我和确诊ADHD的孙女讨论过这个问题。她在嘈杂(尤其是视觉杂乱)的环境中会感到不适,但我的家(相对宽敞,满是书籍,音乐常伴)却让她觉得舒适。她解释说,家里的一切对她而言都很有趣,仿佛在跟她对话——“这是你和奶奶的家,充满了故事”。而学校那种现代且“理应让人安心”的环境,墙上尽是图案和绘画,反而令她烦躁——“那里没有故事,只有噪音。” AI-generated videos to maximally drive a target br… # https://news.ycombinator.com/item?id=48858387 This is the absolutely horrific next stage for social media platforms: They’re already well able to surface the most addictive short video for a specific user out of millions of real videos. But these millions of real videos are just darts thrown into the space of “videos that could hook the user”, in the end even the best-selected of them is not perfect. Now, behold! AI allows to generate the perfect video to surgically hit all the switches in the viewer’s brain and turn it into a zombie hooked for days on end. Let’s hope our regulations hit these “social networks” hard enough so that never dare deploy this kind of technology. aubanel 这是社交媒体平台绝对可怕的下一阶段: 它们早已能够从数百万真实视频中,为特定用户精准推荐最令人上瘾的短视频。 但这些数百万真实视频,不过是扔向“可能让用户上瘾的视频空间”的飞镖,最终即使筛选出的最佳视频也并非完美。 而现在,请注意!人工智能能够生成完美视频,精准触发观众大脑中的每一个开关,将其变成连续数日无法自拔的僵尸。 但愿我们的监管能够狠狠打击这些“社交网络”,让它们永远不敢部署这种技术。 Einstein’s relativity rules chemical bonds in heav… # https://news.ycombinator.com/item?id=48866500 I had incredible difficulties with Chemistry, more than any other subject, because most everything was hand waved away, requiring mostly rote memorization. I could never get an intuitive understanding, partly because my profs seemingly refusing to think about things from a physics perspective. My physics prof was able to help with some of it. It was very odd. If I would have stuck with it, would things have improved? nomel 我在化学上遇到了极大的困难,比其他任何科目都大,因为几乎所有内容都被含糊带过,主要靠死记硬背。我始终无法获得直观的理解,部分原因在于我的教授们似乎拒绝从物理学的角度思考问题。我的物理教授倒是在某些方面帮上了忙。这非常奇怪。 如果当初我坚持学下去,情况会不会有所改善呢? Apple sues OpenAI, accuses ex-employees of stealin… # https://news.ycombinator.com/item?id=48866293 A company that behaves like this in one area, cannot be trusted in any area. Any enterprise that endorses/allows OpenAI products to be used is taking a big risk. xnx 在某个领域如此行事的公司,在任何领域都不可信。任何支持或允许使用OpenAI产品的企业都在承担巨大风险。 Interview with Mitchell Hashimoto about Ghostty an… # https://news.ycombinator.com/item?id=48854737 I remember watching HN and seeing every time there was something Rust related trending, there was ALWAYS a post made shortly after trying to hype Zig and this went on for like 4 years. You just got a tiny taste of what Rust enthusiasts have been doing to every C++ related submission here on HN for years. spacechild1 我记得在HN上看到,每次有Rust相关的内容上热门,之后不久总会出现一个试图吹捧Zig的帖子,这种情况持续了大约四年。你刚刚只是尝到了一小部分Rust爱好者多年来在HN上对每一个与C++相关的提交所做的事情。 Apple sues OpenAI, accuses ex-employees of stealin… # https://news.ycombinator.com/item?id=48867931 An acquaintance of mine was accidentally wired about $100k when it was supposed to be $5k. Before it could be reversed, they moved accounts and immediately bought a one way flight out of country. They then changed all socials and handles. They are now ignoring all court documents and are on track to get a default judgement against them. Their rationale? “It’s mine, they owed me this”. They are 100% convinced that they are in the right, not just that they can keep it but that they actually intended to send them this to begin with. I get it $100k isn’t nothing but they’re also throwing their life away for less than what they used to make a year in salary. People do weird things when given sudden access to money or power. appplication 我一个熟人原本应该收到5000美元,结果意外被汇了大约10万美元。在钱被追回之前,他们转移了账户,并立刻买了一张单程机票出国。随后他们更换了所有社交账号和用户名。现在他们无视所有法庭文件,即将面临缺席判决。 他们的理由是什么?“这是我的钱,是他们欠我的。”他们百分之百确信自己占理,不仅觉得能留下这笔钱,甚至认为对方本来就想汇这么多。我明白10万美元不是小数目,但他们为了不到一年薪水的钱,正在毁掉自己的人生。 人在突然获得金钱或权力时,确实会做出奇怪的事。 Apple sues OpenAI, accuses ex-employees of stealin… # https://news.ycombinator.com/item?id=48866488 OpenAI is about to get ROCKED on this. From this report, this looks open and shut. Apple has basically infinite money and incredible lawyers. Not sure what OpenAI can counter with unless they have clear, hard evidence this hasn’t been happening. Robdel12 OpenAI 在这件事上怕是要被重创了。从这份报告来看,这案子似乎一目了然。苹果基本上有花不完的钱和顶级的律师团队。不确定 OpenAI 能拿什么来应对,除非他们有确凿的硬证据证明此事从未发生过。 Focus # https://news.ycombinator.com/item?id=48855172 Takes some extraordinary lack of self awareness to write something like this while burning tens of billions of dollars a year spearheading the “Metaverse”, which is as big a digression from the company’s core competency as you could possibly get. And soon after publishing this he would go on to lay off a large chunk of the team and more than 20,000 employees total. paxys 每年烧掉数百亿美元带头搞“元宇宙”,这跟公司核心能力偏离得不能再远了,居然还能写出这种东西,真是极度缺乏自我认知。而且这篇东西发出去没多久,他就解雇了一大帮团队和总共超过两万名员工。 Apple sues OpenAI, accuses ex-employees of stealin… # https://news.ycombinator.com/item?id=48868383 Until the industry addresses the Original Sin of Generative AI (and the ascendance of Thievery Corporations), we should expect more and more of this. So far, theft has been rewarded. As long as you make enough money, people seem to be okay with ignoring long-lasting impacts of intellectual theft. As long as you become King of the Cannibals, it seems many are happy to remember you as King and not as the Cannibal. jtfrench 除非行业正视生成式AI的原罪(以及偷盗企业的崛起),否则这类现象只会愈演愈烈。迄今为止,盗窃行为反而得到了回报。只要赚得足够多,人们似乎就能容忍对知识产权的长期危害视而不见。只要你成为食人王,许多人便乐于记住你是王,而不是食人魔。 Postgres rewritten in Rust, now passing 100% of th… # https://news.ycombinator.com/item?id=48855256 That’s not true for Postgres however: due to its usage of a shared memory pool, whenever a subprocess is terminated unexpectedly, Postgres will kill all other processes and enter recovery mode, replaying the WAL, during which time it will not accept connection requests. It does this because it can’t possibly know whether the dying process did bad things to the shared memory pool. pilif 然而,对于Postgres来说情况并非如此:由于它使用共享内存池,当某个子进程意外终止时,Postgres会杀死所有其他进程并进入恢复模式,重放WAL日志,在此期间不会接受连接请求。 它这样做是因为无法知道正在终止的进程是否对共享内存池造成了破坏。 Apple sues OpenAI, accuses ex-employees of stealin… # https://news.ycombinator.com/item?id=48867809 This is how you behave when you think you’re so much smarter than everyone around you that consequences don’t apply you. Spot on perfect. I see this too often and not just in tech. grvdrm 这就是你自以为比周围所有人都聪明,以至于规则对你无效时的样子。 说得太对了。我经常看到这种情况,而且不仅仅是在科技领域。 Modern decor may be straining people’s brains # https://news.ycombinator.com/item?id=48873736 If you’ve ever been in an home owned for generations, filled with books and knickknacks and heirlooms and family photos, despite the clutter it all feels comforting in a way that modern decor doesn’t. The article doesn’t touch much on why modern decor emerged as it did. It’s a market response where everyone needs to (or feels the need to) pick up and move at a moment’s notice. Companies are either expanding or like to think they’ll be expanding soon. People move jobs so often that they have a hard time feeling settled where they are, so they design for that possibility. The modern aesthetic is one of planned impermanence. michaelchisari 若你曾踏足一座传承数代的老宅,里面堆满书籍、小摆件、传家宝和家庭照片,尽管杂乱,却处处透着一种现代装饰无法比拟的温馨感。 这篇文章并未深入探讨现代装饰为何会以这种方式兴起。它本质上是市场的回应——每个人都需要(或感觉自己需要)随时准备打包搬走。企业要么在扩张,要么自认为即将扩张。人们频繁更换工作,很难在某个地方真正安定下来,因此他们便按照这种可能性来设计住所。现代美学的本质,是一种预谋好的无常。 Why American ambulance rides are so expensive # https://news.ycombinator.com/item?id=48854106 The average ambulance transport costs $2,673 to provide I think this ignores the 400 pound gorilla in the room. Why does an ambulance transport cost thousands for the operator? This is a short trip in an automobile, essentially a fancy uber ride. At first one might say that’s flippant - obviously ambulances are specialized vehicles, and you have paramedics, and they need to get to locations quickly, and so forth, but let’s consider those costs. A new, fully equipped ambulance is about $150k. Of course this is more than a regular car, but by a factor of 5, not 50. Let’s be generous and presume the ambulance fully depreciates in 2 years. Typically an ems crew will be two paramedics. Average paramedic wage is about $23/hr. Again, not orders of magnitude more expensive. Then you have liability, both for the vehicle and for the medical treatment; that’s about $12k per year. Throw in money for gas and wear and tear, which should be quite comparable to other automobiles, and it costs about $1600 to own and operate an ambulance for 24 hours. Now the other side of the equation is utilization. Taking the arbitrary example of Philadelphia Fire Department, they have 60 ambulances that handle on average 700 ems calls per day, and approximately 70% of ems calls lead to transport, so that’s about 8 transports per ambulance per day. So distributing this all out, the actual cost to the ambulance operator, ignoring overhead, ought to be somewhere around $200. I’m sure there are some additional costs I haven’t included in this back of the envelope calculation, and maybe some of the numbers I pulled off google are off a bit, this should be taken as a very rough estimate. But even if you significantly increase the cost, the medicare payment amount seems quite reasonable to cover the expenses with a healthy profit margin. Unless you want to claim that operating an ambulance is less than 10% of the cost of ambulance transport, and that the estimators with Medicare are absurdly out of touch with reality, whence cometh $2,673? jjk166 平均每次救护车运输费用为2673美元。 我认为这忽略了房间里的大象:为什么一次救护车运输对运营方来说要花费数千美元?这本质上就是一次短途汽车行程,不过是辆高级优步。乍一看可能觉得这话轻率——显然救护车是专用车辆,配有医护人员,需要快速抵达现场等等,但让我们来算算这些成本。 一辆全新的、装备齐全的救护车大约15万美元。当然这比普通汽车贵,但也就是5倍,不是50倍。就算宽裕点,假设救护车两年内完全折旧。通常一组急救人员是两名医护人员。平均时薪约23美元。同样,这并非数量级的昂贵。然后是责任险,包括车辆和医疗两方面,每年约1.2万美元。再加上燃油和磨损费用,这与其他汽车相当,那么24小时内拥有一辆并运营一辆救护车的成本约为1600美元。 现在来看等式另一侧:利用率。以费城消防局为例,他们拥有60辆救护车,每天平均处理700个急救电话,其中约70%需要运输,即每辆救护车每天约8次运输。分摊下来,忽略管理成本,一次救护车运输对运营方而言的实际成本应该在200美元左右。 我确信这个粗略估算中遗漏了一些额外成本,也许我从谷歌搜到的数字有些出入,这只能当作非常粗略的估计。但即便大幅提高成本,医保支付金额似乎也足以覆盖开支并保持健康利润率。除非你想声称运营救护车的成本不到运输费用的10%,并且医保的估价者荒谬地脱离现实,那么这2673美元从何而来?