Hacker News Daily — 2026-04-13 Trending
Scraped at 21:41, April 13, 2026 (PDT)
(1) DaVinci Resolve releases Photo Editor
Blackmagic Design rolls out Photo Editor inside DaVinci Resolve, blurring the line between video and photo editing. The feature leverages Resolve's color pipeline and non-destructive editing, appealing to professionals who juggle photos and video in a single project.
Blackmagic Design 将 Photo Editor 纳入 DaVinci Resolve，使视频与照片编辑的边界更模糊。该功能利用 Resolve 的调色与非破坏性编辑能力，方便在同一工作流中处理照片与视频。
Stacked PRs bundle several dependent pull requests into a single review flow, letting teams ship large features incrementally. The gh-stack tool provides commands to create, rebase, and track the stack, helping reviewers stay in sync across related changes.
堆叠 PR 将若干相互依赖的 PR 打包到一个评审流程中，便于分阶段交付大型功能。gh-stack 提供创建、变基和跟踪等命令，使评审人能够在相关改动之间保持同步。
(3) Someone bought 30 WordPress plugins and planted a backdoor in all of them
A security researcher demonstrates the depth of supply-chain risk in plugin ecosystems by buying 30 WordPress plugins and inserting a backdoor into each one. The stunt highlights how third-party extensions can compromise many sites if integrity checks and monitoring are lax.
安全研究者展示了通过购买 30 个 WordPress 插件并在每一个插件中植入后门，能够大范围入侵网站。这暴露了第三方插件的供应链和信任机制漏洞，强调加强插件审计、完整性校验和运行时监控的重要性。
(4) A new spam policy for "back button hijacking"
Google introduced a new policy to curb spammy practices that exploit back button hijacking. The rule penalizes sites that mislead users through overlays or deceptive navigation, affecting search rankings and user experience.
谷歌宣布对回退按钮劫持的垃圾邮件行为更新政策，打击通过覆盖层和误导性导航等方式强迫用户返回的做法。新规将影响搜索排名和用户体验，促使开发者改进 UX。
(5) Lean proved this program correct; then I found a bug
Lean proved the program correct, but a subsequent bug exposed a mismatch between the formal model and the real implementation. The post underscores that formal proofs are not a substitute for testing and real-world validation. It also highlights how small modeling gaps can lead to surprising failures.
Lean 证明了该程序正确；随后作者发现实现与模型之间的不匹配导致了漏洞。该文强调形式化证明不能替代测试和现实世界的验证。还指出微小的建模差异也会导致意想不到的失败。
(6) Hacker compromises A16Z-backed phone farm, calling them the 'antichrist'
A hacker compromised a16z-backed phone farm, using access to post memes and publicly insult the firm, calling it the antichrist. The breach underscores the security risks of operator-controlled hardware farms and the need for robust incident response and access controls in high-profile startups.
黑客入侵了由 a16z 支持的手机农场，利用访问权限发表迷因并公开侮辱该公司，称其为反基督。此事件凸显了运营型硬件农场的安全风险，以及在高知名度初创企业中加强 incident response 与访问控制的必要性。
(7) WiiFin – Jellyfin Client for Nintendo Wii
WiiFin is a Jellyfin client for the Nintendo Wii, enabling media streaming on retro hardware. It demonstrates how open-source media projects extend to older consoles, though the experience is constrained by the platform's limits.
WiiFin 是面向 Nintendo Wii 的 Jellyfin 客户端，允许在这台老旧主机上从 Jellyfin 服务器流媒体。该项目展示了开源媒体生态向旧硬件的扩展潜力，但受限于硬件性能与控制体验。
(8) Nothing Ever Happens: Polymarket bot that always buys No on non-sports markets
A Polymarket bot consistently places 'No' bets on non-sports markets, revealing a recurring pattern in market dynamics that automated strategies can exploit. This raises questions about prediction-market design and how such bots can skew odds or drain liquidity.
一个 Polymarket 机器人在非体育市场持续以‘否’下注，暴露了市场机制中的漏洞和自动化策略的潜在影响。此现象引发对预测市场设计、赔率偏差以及流动性消耗的讨论。
(9) The Future of Everything Is Lies, I Guess: Safety
The piece argues that safety is not a simple checkbox but a design and governance problem, requiring practical, measurable safety engineering. It emphasizes building robust safety practices into systems rather than chasing vague guarantees.
作者质疑把“安全”当作一个简单勾选项的做法，认为安全是设计与治理的问题，需要落地的、可度量的安全工程实践。文章强调在系统设计、实施与监管中持续投入以实现稳健性。
(10) Make tmux pretty and usable (2024)
A practical guide to customizing tmux, covering themes, status bars, fonts, and key bindings to improve aesthetics and workflow. It shares concrete configurations and tips for a more pleasant, productive terminal multiplexer experience.
这篇指南给出实用的 tmux 自定义方法，涵盖主题、状态栏、字体和快捷键等，提升外观与工作流。通过具体的配置建议，帮助开发者把终端多路复用器变得更好用。
(11) Servo is now available on crates.io
Servo, the Rust-based browser engine, is now published as a crates.io crate. This makes it easier for developers to depend on Servo directly from their Rust projects and experiment with its components like WebRender. The release could accelerate integration efforts and broader experimentation within the Rust ecosystem.
Rust 生态中的浏览器引擎 Servo 现已以 crates.io 包形式发布。开发者可以直接在 Rust 项目中依赖 Servo，并尝试其组件如 WebRender。此举可能加速在 Rust 生态内对渲染模块的集成和实验。
(12) Stanford report highlights growing disconnect between AI insiders and everyone
The Stanford report underscores a widening gap between AI insiders—labs, VCs, policy-makers—and everyday tech workers and the public. It argues that elite narratives outpace broad understanding, risking misaligned expectations, policy blind spots, and unaddressed workforce impacts. The takeaway: better translation of AI risks and more inclusive dialogue is essential.
斯坦福报告揭示 AI 内部人士与公众之间的认知差距正在扩大，精英叙事往往超出大众理解，风险被低估。需要更透明的风险沟通与包容性的对话来避免错配与政策盲点。
(13) Building a CLI for all of Cloudflare
Cloudflare demonstrates a local explorer CLI that lets developers probe and test Cloudflare features offline, speeding iteration and debugging.
Cloudflare 展示了一个本地 CLI 探索工具，便于开发者在离线状态下试用和查看 Cloudflare 的功能，从而提升开发效率与离线调试能力。文章讨论了如何设计一个安全且易用、覆盖广泛功能的命令行界面。
(14) US appeals court declares 158-year-old home distilling ban unconstitutional
An appeals court struck down a long-standing home distilling ban as unconstitutional, signaling a notable shift in how aging alcohol regulations are interpreted. The decision could influence hobbyist distillers and spark broader debate about regulatory modernization in the U.S.
联邦上诉法院裁定一项历时158年的家庭蒸馏禁令违宪，标志着对过时酒精法规的重大挑战。此判决可能影响爱好者自行酿酒的合规性，并推动监管现代化的讨论。
(15) Microsoft isn't removing Copilot from Windows 11, it's just renaming it
Microsoft is keeping Copilot in Windows 11 but rebranding it, signaling a broader strategy to unify AI-assisted features across Windows. The rename could affect user expectations and how privacy and data usage are communicated.
微软保留 Windows 11 版 Copilot，仅重新命名，显示出将 AI 助力功能统一到 Windows 生态的策略。改名可能影响用户预期，并对隐私与数据使用的传达产生影响。
(16) This year’s insane timeline of hacks
A Substack timeline recounts major hacks and breaches this year, highlighting attack trends, supply-chain risk, and defense lessons for developers and security teams. It provides quick takeaways on how to improve resilience.
汇总本年度重大黑客事件与安全漏洞时间线，梳理攻击趋势、供应链风险与防御要点，帮助开发者与安全团队提升韧性。
(17) All elementary functions from a single binary operator
Proves that a single binary operator can generate all elementary functions, revealing a surprisingly small algebraic basis for calculus. The result has implications for symbolic computation, functional programming, and universal function representations.
研究表明，通过一个二元运算即可构造所有初等函数，揭示了微积分的极简代数基础。这一发现可能影响符号计算、函数式编程以及统一表示函数的方法。
(18) How to make Firefox builds 17% faster
Firefox builds get 17% faster thanks to caching WebIDL codegen, which eliminates repeated work across incremental builds. The takeaway is to identify expensive, deterministic codegen steps and cache them to accelerate iteration in large codebases.
通过缓存 WebIDL 代码生成步骤，增量构建中的重复工作被消除，使 Firefox 构建快了约 17%。对于大型代码库，抓住耗时且可重复的代码生成步骤并对其进行缓存，可以显著提升迭代速度。
(19) Android now stops you sharing your location in photos
Android now strips location data from photos before sharing, helping protect user privacy when posting to social apps. The change reduces inadvertent location leaks and sets a baseline for privacy controls across mobile platforms, though users can still re-enable when needed.
Android 新功能在分享照片前移除位置信息元数据，帮助用户避免地理位置信息泄露。该改动提升隐私保护，但仍允许需要时开启位置标记的选项。
(20) Show HN: Ithihāsas – a character explorer for Hindu epics, built in a few hours
Ithihāsas is a lightweight, quickly-built explorer of characters in Hindu epics. It visualizes relationships and story connections, showcasing what can be built in a few hours and serving as a prototype for narrative tools.
Ithihāsas 是一个用于探索印度史诗人物的轻量级工具，快速搭建完成。它可可视化人物关系与故事线索，展示短时间原型可以实现的功能，适合作为叙事工具的雏形。
(21) The AI revolution in math has arrived
Discusses how large-language models and AI systems are transforming mathematical practice, from conjecture generation to automated proof assistance.
讨论大型语言模型和 AI 系统如何改变数学研究实践，从猜想生成到自动证明辅助。既展示突破，也提出关于严格性、验证以及数学家如何与 AI 工具协作的核心问题。
(22) GAIA – Open-source framework for building AI agents that run on local hardware
GAIA is an open-source framework for building AI agents that run offline on local hardware, enabling privacy-preserving and low-latency AI workflows. The project emphasizes modular agent composition, local computation, and interoperability with popular AI models, reducing reliance on cloud infrastructure.
GAIA 是一个在本地硬件上运行 AI 代理的开源框架，支持离线隐私保护和低延迟的 AI 工作流。项目强调模块化代理组合、本地计算，以及与主流 AI 模型的互操作性，降低对云端基础设施的依赖。
(23) Tell HN: Docker pull fails in Spain due to football Cloudflare block
A Docker pull failure in Spain is linked to Cloudflare blocking football-related requests, illustrating how regional filtering can disrupt developer workflows. It underscores the need for resilient mirrors and alternative routing.
西班牙的 Docker 拉取失败被追踪至 Cloudflare 对足球相关流量的拦截，反映出区域性过滤可能干扰开发工作流。需要更稳健的镜像源与路由策略以缓解。
(24) The looming college-enrollment death spiral
The Atlantic piece analyzes demographic shifts and higher education enrollment declines, discussing implications for universities, student debt, and regional economies; it warns about long-term sustainability and the need for policy adaptation.
文章分析人口结构变化导致的高等教育入学率下降及其对大学、学生债务和地区经济的冲击，呼吁政策层面的对策和长期可持续性思考。
(25) The economics of software teams: Why most engineering orgs are flying blind
Most engineering organizations lack visibility into the true cost of software delivery, often optimizing for velocity rather than value. The piece argues that without product-aligned incentives and cost-aware metrics, teams accumulate technical debt and misallocate resources. It offers practical levers like measuring feature economics, improving cross-functional collaboration, and building dashboards that surface cycle time, cost per feature, and delivered impact.
工程组织往往对软件交付的真实成本缺乏清晰度，常以追求速度为目标而忽视价值。若缺乏以产品为导向的激励与成本感知的指标，团队容易积累技术债务、资源错配。文中提出通过衡量功能经济学、提高跨职能协作，以及建立能反映交付成本、周期时间和实际影响的看板等来纠偏。
(26) Michigan 'digital age' bills pulled after privacy concerns raised
Michigan pulled its 'digital age' bills after concerns over privacy and data handling surfaced, halting partisan push for new tech rules. The setback highlights the fragility of privacy safeguards in proposed tech legislation and how public input can sway policy.
密歇根州在暴露的隐私与数据处理担忧后撤回了《数字时代》相关法案，暂停了对新技术监管的推进。此举凸显在拟议技术法案中隐私保护的脆弱性，以及公众意见对政策的影响。
(27) Tax Wrapped 2025
Tax Wrapped 2025 provides guidance and tooling for tax planning in 2025.
Tax Wrapped 2025 提供 2025 年税务规划的指南与工具。
(28) DIY Soft Drinks
A guide to making your own carbonated drinks, including syrups, carbonation methods, and flavoring ideas, with tips on cost, customization, and scaling. It shows how small-batch experimentation can rival commercial beverages.
一份自制软饮指南，涵盖糖浆、碳化方法与口味搭配，强调成本、定制化与小批量制作的可行性。通过动手尝试，可以在家实现与商业饮料相近的口感与乐趣。
(29) Apple's accidental moat: How the "AI Loser" may end up winning
Apple's cautious stance on AI could become a durable moat, built on privacy, on-device AI, and a tightly integrated ecosystem. As competitors chase scale, Apple's trust advantage and user experience may prove sturdier over the long term.
苹果对 AI 热潮的谨慎态度可能成为长期护城河，源自隐私保护、设备端 AI 与紧密的生态系统。随着竞争对手追逐规模，苹果在信任与用户体验方面的优势或在长期内更为稳固。
(30) I went to America's worst national parks so you don't have to
A personal travel piece visiting some of the most poorly reviewed US national parks, sharing candid impressions, practical tips, and why some parks underperform.
这篇个人游记走访了美国评价最差的国家公园，直陈体验、实用建议与为何有些公园表现不佳。作者用亲身经历为读者提供省事的参观要点与改进的思考。
(31) N-Day-Bench – Can LLMs find real vulnerabilities in real codebases?
N-Day-Bench introduces a benchmark to evaluate whether LLMs can discover real vulnerabilities in actual codebases. Results demonstrate potential and current limits of AI-assisted security analysis, guiding prudent deployment in code review tasks.
N-Day-Bench 引入一个基准测试，用以评估大型语言模型在真实代码库中发现漏洞的能力。结果展示了 AI 辅助安全分析的潜力与现阶段的局限，为将来在代码审查中的应用提供参考。
(32) New Orleans's Car-Crash Conspiracy
The New Yorker piece investigates a car crash conspiracy in New Orleans, tracing misinformation patterns, narrative construction, and potential misreporting. It sheds light on how digital media can shape public perception around a tragedy.
本文深入调查新奥尔良的一起车祸阴谋论，追踪信息传播、叙事建构及报道中的偏差，揭示数字媒体如何左右公众对事件的看法。
(33) B-trees and database indexes (2024)
The article explains B-trees and indexing strategies, likely comparing traditional B-trees vs newer structures in 2024; it discusses why proper indexing matters for query performance and how to choose between B-trees, B+ trees, and other indexes; maybe covers caching and statistics.
文章讲解 B 树与数据库索引的原理与实现，比较 2024 年的更新与优化，强调正确的索引选择对查询性能的影响，并介绍在实际场景中如何权衡不同实现。
(34) Pro Max 5x quota exhausted in 1.5 hours despite moderate usage
Quotas can throttle heavy workloads far quicker than one might expect, even with moderate activity. This case underscores the importance of understanding rate limits and having fallback or multi-account strategies for critical tasks.
在中等使用量下，配额在1.5小时内就被耗尽，暴露了按用户设定的限额在持续工作中的约束。此事提醒开发者在规划工作流时要考虑配额上限、突发流量以及备选方案。
(35) Google removes "Doki Doki Literature Club" from Google Play
Google removed the game from Play Store, triggering questions about platform moderation. The move highlights how policy decisions affect indie games' visibility and monetization. It also sparks debate over criteria, consistency, and censorship on major app stores.
谷歌已将该游戏从应用商店下架，引发关于平台审核与独立游戏曝光之间平衡的讨论。此举凸显了政策决策对独立开发者的可见度与盈利的影响，并引发关于标准、执行一致性以及平台审查尺度的辩论。
(36) I run multiple $10K MRR companies on a $20/month tech stack
Running several SaaS-like businesses on a $20/month stack is feasible with strong automation and lean operations. The post outlines strategies to maintain roughly $10K MRR per product at tiny ongoing costs, highlighting how a low-cost, automated stack can scale.
在极低成本和高度自动化的条件下，仍然能经营多家以订阅为核心的产品并实现可观的月经常性收入。作者分享了将每个项目的月耗控在约20美元、个别产品实现约1万美元MRR的做法，挑战了对高成本技术栈的普遍认知。
(37) Seven countries now generate nearly all their electricity from renewables (2024)
Across several nations, electricity generation is approaching 100% renewables, driven primarily by hydro, wind, and solar plus supportive policy frameworks. While intermittency remains a challenge, these cases show that with grid investments and capacity planning, a transition to low-carbon electricity is increasingly feasible at national scales.
多国的电力生产正在接近100%来自可再生能源，主要靠水电、风电和太阳能及配套政策。尽管间歇性仍是挑战，但这些案例表明通过电网投资和容量规划，国家层面的低碳发电转型正变得更加可行。
(38) Who's Been Impersonating This ProPublica Reporter?
ProPublica reports a ProPublica reporter's identity was spoofed by someone impersonating them to mislead sources or spread misinformation. The incident underscores the growing threat of media impersonation and the need for verification workflows and journalist identity protections.
ProPublica 警告称有人冒充该记者身份，以误导消息源或散布不实信息。事件凸显了媒体身份冒充的日益严重，需要加强身份验证流程和记者安全保护。
(39) Visualizing CPU Pipelining (2024)
This post visualizes CPU pipeline stages and hazards, helping readers grasp how instruction-level parallelism and stalls affect performance. It provides diagrams and explanations that are accessible to both students and professionals.
本文可视化 CPU 流水线的各阶段与冒险，帮助读者理解指令级并行性与停顿如何影响性能。配以图示与解释，适合学生与专业人士阅读。
(40) Bring Back Idiomatic Design (2023)
The piece argues for designing APIs and UIs that follow the host language's idioms rather than importing external design paradigms. It reinforces how idiomatic design improves readability and safety.
文章主张在软件设计中遵循目标语言的习惯用法，而非照搬外部风格。强调地道设计提升可读性、安全性和开发者体验。
(41) Show HN: boringBar – a taskbar-style dock replacement for macOS
boringBar is a macOS replacement for the Dock with a taskbar-style interface. It emphasizes persistent pinned apps, quick-launch, and consistent shortcuts across spaces, offering an alternative UX to Apple’s native Dock. If it gains traction, it signals demand for more modular, multitasking-oriented macOS UI.
boringBar 是一个为 macOS 设计的 Dock 替代，采用类似任务栏的界面，便于固定应用、快速启动并在各工作区保持一致。它为多任务工作提供了新的交互方式，若受欢迎，可能反映出对更模块化、以多任务为导向的 macOS UI 的需求。
(42) The peril of laziness lost
Removing laziness can carry hidden risks. The piece argues that disciplined efficiency can backfire in surprising ways, potentially stifling creativity and resilience.
文章指出失去懒惰所带来的风险，强调在克制与纪律之外可能出现的隐性成本。
(43) AI could be the end of the digital wave, not the next big thing
A provocative argument that AI may mark the culmination rather than the next leap of the digital revolution, suggesting diminishing returns and a potential need to rethink innovation incentives.
一个挑衅性观点提出：人工智能可能成为数字化浪潮的顶点而非引领下一个重大突破，这暗示收益趋于边际并促使重新审视创新激励。
(44) Missouri town fires half its city council over data center deal
A Missouri town ousts half of its city council after approving a data center deal, illustrating governance tensions around large tech incentives and transparency. The incident demonstrates how high-stakes deals can destabilize local leadership and ignite public backlash.
密苏里小镇在批准数据中心交易后解雇了半数市议会，反映了围绕大型科技激励措施和透明度的治理紧张。此事显示高额交易可能动摇地方领导层并引发公众反对。
(45) Ascending into the Realm of Japanese Charts
A data-visualization piece exploring Japanese charts, mapping trends, genres, and popularity trajectories; it reveals cultural and market dynamics through visual storytelling.
一份数据可视化作品，聚焦日本排行榜领域，呈现趋势、体裁和受欢迎程度的演变，以及由此揭示的文化与市场动态。
(46) Why it’s impossible to measure England’s coastline
Explains the coastline paradox and why measuring a coastline becomes longer as you measure with smaller rulers, using England's coast as a case study. The piece touches on geometry, scale, and practical implications for maps, climate models, and governance.
解释了海岸线悖论：随着测量单位变小，海岸线长度会变得更长。以英格兰海岸为例，讨论了尺度、几何与对地图、气候模型和治理的实际影响。
(47) Show HN: I built a social media management tool in 3 weeks with Claude and Codex
Demonstrates rapid AI-assisted development by building a social media management tool in three weeks using Claude and Codex. The project highlights how large language models can accelerate UX-focused tooling, with trade-offs around reliability and governance.
展示在三周内借助 Claude 与 Codex 构建的社媒管理工具，体现 AI 辅助开发的速度与潜力。项目强调在提升 UX 的同时需关注可靠性与治理等权衡。
(48) They See Your Photos
An examination or project about how photos are scanned or analyzed by services, devices, or apps, raising privacy concerns and practical steps to protect image data.
聚焦照片在云端、设备或应用中被扫描和分析的机制，提出隐私风险与保护影像数据的实际建议。
(49) Just Enough Chimera Linux
This guide shows a minimal Chimera Linux installation with ZFS as root, outlining practical steps to get a lean, reliable system up and running. The guide focuses on a small footprint and thoughtful configuration for experiments or hosting workloads.
该指南演示在 Chimera Linux 上以 ZFS 作为根分区的最小化安装方案，提供从引导到基本配置的实用步骤，帮助你快速搭建一个精简且可靠的系统。内容聚焦于低占用和理性配置，适合做实验或托管负载。
(50) The Rational Conclusion of Doomerism Is Violence
The piece argues that doomerism—pessimistic, fatalistic worldviews—can normalize violent behavior as a supposedly rational outcome. It advocates countermeasures like constructive narratives, community resilience, and policy efforts to reduce despair-driven harm.
文章认为悲观、宿命式的世界观会让暴力被视为“理性结果”，并呼吁通过建立积极叙事、提升社区韧性以及制定减少绝望驱动伤害的政策来对抗这一趋势。
(51) MEMS Array Chip Can Project Video the Size of a Grain of Sand
A MEMS-photonics chip demonstrates ultra-compact video projection at grain-of-sand scale, signaling advances in micro-display tech. The breakthrough points to potential applications in wearables, AR/VR, and mini projectors.
新型 MEMS 光子芯片实现了极小尺度的视频投射，尺寸可达到沙粒等级。该突破揭示了微型显示技术的进展及在可穿戴设备、AR/VR 和微型投影设备等领域的潜在应用。
(52) In Denmark, the spread of solar panels has become a divisive issue
The Guardian story examines how solar panel adoption in Denmark became politically divisive, balancing climate goals, energy independence, and local opposition ahead of elections.
报道丹麦太阳能扩散引发的政治分歧，平衡气候目标、能源独立与地方反对意见，凸显选举前的政策博弈。
(53) Filing the corners off my MacBooks
A playful hardware tweak: the author files the corners of their MacBooks to smooth edges, sharing steps, tools, and cautions about warranty and durability. It doubles as a meditation on DIY customization vs. risk of damaging laptops.
作者动手把 MacBook 的边角削圆，给出具体步骤与所用工具，并讨论此改造对保修与耐用性的影响。全文从极客自我改造的角度，兼具风险评估的思考。
(54) Taking on CUDA with ROCm: 'One Step After Another'
AMD's ROCm roadmap aims to compete with CUDA through incremental improvements, focusing on portability, kernel compilation, and performance optimizations. The approach underscores the difficulty of breaking NVIDIA's software moat and the need for strong open standards in GPU computing.
AMD 的 ROCm 计划通过渐进改进与 CUDA 竞争，聚焦可移植性、内核编译与性能优化。该策略凸显打破 NVIDIA 软件生态护城河的难度，同时也强调在 GPU 计算领域坚持开源标准的重要性。
(55) Anthropic downgraded cache TTL on March 6th
Anthropic reduced Claude's cache TTL on March 6, affecting how fresh responses are served to clients. The change increases cache churn, potentially improving freshness at the cost of latency and API throughput. It signals ongoing tuning of balance between speed, cost, and accuracy.
3月6日，Anthropic 下调了 Claude 的缓存 TTL，导致返回结果的时效性和新鲜度受到影响。更高的缓存替换率提升了新鲜度，但可能增加延迟与成本，反映出在速度、成本与准确性之间的持续调参。
(56) Initial mainline video capture and camera support for Rockchip RK3588
Collabora reports initial mainline Linux kernel support for video capture and cameras on the RK3588, enabling better Linux on RK3588-based devices. This milestone helps developers ship hardware with upstream drivers and smoother support for cameras and video pipelines.
Collabora 公告 RK3588 已实现主线 Linux 内核的视频捕获与摄像头支持的初步阶段，提升基于 RK3588 的设备在 Linux 下的兼容性。该里程碑为开发者提供更直接的上游驱动和更顺畅的视频/摄像头管线支持。
(57) Ask HN: What Are You Working On? (April 2026)
Developers share a mix of side projects, product experiments, and work-in-progress ideas, reflecting current tech interests. Topics span AI tooling, cloud infrastructure, and deployment hurdles, offering a barometer of what a tech audience is tinkering with in April 2026.
开发者分享本月正在做的项目、产品实验和在研点子，揭示当前的技术趋势。话题涵盖 AI 工具、云基础设施与部署挑战，为技术人群在 2026 年 4 月的热议方向提供风向标。
(58) We have a 99% email reputation, but Gmail disagrees
Despite claiming a 99% email reputation score from one service, Gmail disagrees, leading to confusion about deliverability. The incident highlights variability in reputation metrics and the importance of robust email authentication practices.
尽管某服务给出99%的邮件信誉分，但 Gmail 的判定并不认同，导致投递问题与误解。此事凸显邮件信誉评估之间的差异以及强化 DKIM/SPF 等认证、送达策略的重要性。
(59) I ran Gemma 4 as a local model in Codex CLI
I ran Gemma 4 as a local model using Codex CLI, demonstrating offline inference with a relatively small footprint.
将 Gemma 4 作为本地模型通过 Codex CLI 运行，展示了离线推理的可行性与相对低延迟。开源权重模型在本地实验的可行性为隐私和成本控制提供了机会，但需要权衡部署复杂度与硬件需求。
(60) Exploiting the most prominent AI agent benchmarks
Berkeley piece critiques how prominent AI benchmarks are exploited or gamed, questioning reliability of current benchmarks for real-world agent evaluation. It argues for more robust, trustworthy benchmarks to guide research and product decisions.
Berkeley 的文章质疑当前最显著的 AI 代理基准的健壮性，指出评测结果可能被操纵或不充分反映现实场景。它呼吁建立更可信的基准，以正确引导研究方向和产品决策。
(61) A perfectable programming language
Explores design principles for a programming language meant to be progressively improved over time, emphasizing a minimal core, extensibility, and verifiable correctness. The piece argues for a language with a small, stable core and a pathway for community-driven extensions, balancing practicality with long-term evolvability.
探讨一种可逐步完善的编程语言的设计原则，强调最小核心、可扩展性与可验证性。文章提出以小而稳定的核心和社区驱动的扩展为路径，兼顾实用性和长期演化。
(62) European AI. A playbook to own it
Europe's playbook outlines strategies for building sovereign, ethical, and competitive AI ecosystems, balancing regulation with innovation. It emphasizes data governance, responsible AI, and regional leadership to reduce dependence on non-European tech stacks.
这份行动指南提出在监管与创新之间取得平衡，打造主权、道德且具竞争力的欧洲 AI 生态系统。文中强调数据治理、负责任的 AI，以及提升区域领导力，以减少对非欧洲技术体系的依赖。
(63) Show HN: Oberon System 3 runs natively on Raspberry Pi 3 (with ready SD card)
A demo shows Oberon System 3 running natively on Raspberry Pi 3, accompanied by a ready-to-use SD card image. It demonstrates how a compact, single-tasking OS can still be useful for education and experimentation on modern SBCs.
演示展示 Oberon System 3 在 Raspberry Pi 3 上原生运行，并提供现成的 SD 卡镜像。此举显示了简约、单任务操作系统在现代单板计算机上的教育与试验潜力。
(64) The Closing of the Frontier
The piece argues that the era of easy, open-ended frontier exploration is ending as regulatory, geopolitical, and funding constraints tighten the boundaries. It discusses how this shift will reshape risk-taking, investment, and innovation strategies.
文章认为，随着监管、地缘政治和资金约束加剧，曾经容易探索的前沿正在收缩。此变动将改变风险偏好、投资方向及创新策略。
(65) Phyphox – Physical Experiments Using a Smartphone
Phyphox enables physical experiments using smartphone sensors, turning everyday devices into lab instruments. It lowers barriers for STEM education and fieldwork by offering ready-made experiments and data collection tools.
Phyphox 让智能手机内置传感器成为物理实验工具，降低教育与现场实验的门槛。通过现成的实验和数据采集工具，方便教师与学生进行动手研究。
(66) Tech valuations are back to pre-AI boom levels
Tech valuations have fallen back to pre-AI-boom levels, signaling a normalization of investor sentiment after AI hype. The reset may separate hype from fundamentals, presenting risks and opportunities for investors.
科技公司的估值已回落至AI繁荣前的水平，显示在AI 热潮之后投资者情绪回归理性。估值的回落可能将热度与基本面区分开来，给投资者带来潜在的机会与风险。
(67) JVM Options Explorer
A web-based explorer for Java Virtual Machine options helps developers discover and compare flags, from GC tuning to JIT and memory settings. It reduces guesswork during JVM tuning and speeds up performance experimentation.
这是一个在线的 JVM 参数浏览器，方便开发者查找、对比各类参数及其用途，包括垃圾回收、JIT 和内存设置。能够降低调优中的试错成本，加速性能试验。
(68) Caffeine, cocaine, and painkillers detected in sharks from The Bahamas
A study finds illicit drugs and stimulants in sharks off the Bahamas, highlighting pollution and bioaccumulation concerns in marine ecosystems. The finding raises questions about ocean health and drug traceability in wildlife.
研究在巴哈马近海的鲨鱼体内检测到咖啡因、可卡因和止痛药，揭示了海洋污染和生物累积的问题，对海洋健康及野生动物的药物暴露带来关注。
(69) Claude Mythos: The System Card
Explores Claude Mythos' System Card, a design artifact that defines the model's allowed behavior and safety constraints.
解析 Claude Mythos 的系统卡这一设计要素，阐明模型允许的行为、安全约束与运行边界。文章讨论系统卡如何影响提示工程、风险控制和 AI 系统的可控性。
A survey of home labs in 2026 highlights evolving hobbyist setups, with popular virtualization stacks, compact hardware, and a push toward automation and energy efficiency. It offers practical guidance for enthusiasts building compact, affordable labs.
对2026年家庭实验室的现状进行调查，揭示了流行的虚拟化架构、紧凑型硬件，以及对自动化与节能的重视。为希望以有限空间和预算搭建个人实验室的爱好者提供实用建议。
(71) The End of Eleventy
The piece argues Eleventy may be losing traction as modern front-end stacks demand more dynamic, integrated tooling. It highlights tradeoffs between simplicity and capability in static-site workflows, urging teams to re-evaluate their tooling choices as ecosystems evolve. The author suggests exploring hybrid approaches or newer frameworks that offer better scalability.
本文认为 Eleventy 在当今需要动态特性和集成工具的前端生态中可能逐渐失去吸引力。作者讨论静态站点工作流的取舍，建议团队在生态演进中重新评估工具选择，并考虑更具扩展性的混合方案或新框架。
(72) How long-distance couples use digital games to facilitate intimacy (2025)
The study documents how couples in long-distance relationships use digital games to maintain closeness, shaping interactions, intimacy, and relationship satisfaction. It offers takeaways for designers of social games and remote communication tools.
研究记录了远距离情侣如何通过数字游戏维持情感亲密，影响互动模式、情感表达与关系满意度。研究结果对社交游戏设计和远程沟通工具的发展提供了可转化的洞察。
(73) Tracking down a 25% Regression on LLVM RISC-V
A developer tracks a substantial performance regression (~25%) in LLVM's RISC-V backend, outlining steps like narrowing optimization passes, inspecting code generation, and validating with benchmarks. The work illustrates the fragility of compiler backends and the value of regression testing.
开发者追踪了 LLVM RISC-V 后端的显著性能回归（约 25%），描述了收缩优化阶段、检查代码生成以及回归基准测试等定位步骤。这一工作凸显了编译器后端的脆弱性以及回归测试的重要性。
(74) X Randomly Banning Users for "Inauthentic Behavior"
X has reportedly banned users without transparent justification, labeling actions as inauthentic behavior. The policy ambiguity and apparent automation lead to unpredictable moderation, underscoring the tension between platform safety rules and user trust. It spotlights the risk of overreach in large-scale content moderation.
据称，X 会在缺乏透明原因的情况下封禁用户，并以非真实行为为由。此类执法的随意性与自动化倾向会带来不可预测的结果，削弱用户信任，也暴露出大型社交平台在大规模内容治理上的风险与挑战。
(75) Uncharted island soon to appear on nautical charts
A previously uncharted island has been confirmed and will soon appear on nautical charts, marking progress in remote-area mapping. This matters for navigation safety, logistics, and scientific operations in the region.
一座先前未被绘制的岛屿已被确认，即将出现在航海图上，标志着对偏远水域的制图工作取得进展。这对航海安全、补给与科学考察的后勤具有重要意义。
(76) Point Cloud Allemansrätten
Explores how Sweden's Allemansrätten natural rights law intersects with publishing and sharing point cloud data, highlighting tensions between open data, privacy, and land access norms.
探讨瑞典的 Allemansrätten（全民可享用的自由权）如何与点云数据的发布与共享产生冲突，凸显开放数据、隐私和土地使用规范之间的张力。
(77) Show HN: Claudraband – Claude Code for the Power User
Show HN about Claudraband, a project enabling advanced users to leverage Claude's coding capabilities; features may include faster prompts, templates, or CLI workflows.
Claudraband 是一个让高级用户更高效利用 Claude 的编码工具集，提供改进的界面或工作流以加速编码、测试与部署。面向 power users 的能力扩展体现了对大语言模型在专业场景中可用性的追求。
(78) The Physics of GPS
Explains how GPS works, including relativistic corrections and signal timing, and shows how geometry and timing enable precise positioning.
详细解释 GPS 的工作原理，涵盖相对论修正、信号时序，以及几何与时钟同步如何实现高精度定位。
(79) Haunt, the 70s text adventure game, is now playable on a website
A classic 1970s text adventure, Haunt, is now playable on the web, letting modern players experience parser-driven storytelling without any setup. The project preserves vintage interactive fiction using browser technology, highlighting how retro games can reach new audiences.
经典的1970年代文本冒险游戏 Haunt 现已在网页上可玩，现代玩家无需额外设置即可体验解析器驱动的叙事。这一项目通过浏览器技术实现对老式交互式小说的保存与传播，展示了复古游戏的跨世代可访问性。
(80) I gave every train in New York an instrument
TrainJazz outfits NYC trains with a different instrument, creating an evolving city soundtrack; it explores data sonification and public space as music. It highlights how creative data projects can repurpose transit systems into interactive art while inviting public participation.
TrainJazz 给纽约地铁列车配上乐器，形成不断变化的城市声景。此项目把数据声音化与公共空间艺术结合，体现了将交通系统转化为互动音乐的潜力，同时鼓励公众参与。
A playful project runs Doom using curl-based tooling, demonstrating creative use of everyday CLI utilities as a game-like experience. It showcases how far developers can push tooling boundaries and what counts as interactive programming.
一个有趣的项目用 curl 构建的工作流来运行 Doom，展示了用常见 CLI 工具打造游戏体验的创造力与边界扩展。
(82) Optimization of 32-bit Unsigned Division by Constants on 64-bit Targets
Proposes an optimization strategy for performing 32-bit unsigned division by constants on 64-bit targets, using precomputed multipliers and shifts to replace costly division instructions. The technique can help compilers generate faster code for numeric-heavy workloads.
提出在64位目标上将32位无符号常数除法替换为乘法因子和位移的优化方法，避免昂贵的除法指令。该技巧有助于编译器生成对数值密集型负载更高效的代码。
(83) Most people can't juggle one ball
The piece argues cognitive limits cap how many things we can track or manage; multitasking often reduces performance even with simple tasks. This finding has implications for UX and product design, reminding designers not to promise effortless multitasking.
作者主张人类在同时处理多项任务时存在认知极限，哪怕是处理一个简单任务也会有困难。这一发现对用户体验和产品设计有启示：避免把多任务承诺得过于美好，需更真实地考虑人类注意力与工作记忆的边界。
(84) Why AI Sucks at Front End
AI struggles with frontend tasks due to nuanced UI/UX, accessibility, and component composition that require human judgment. The piece argues AI is best used as an assistant rather than a replacement for frontend engineers.
AI 在前端开发中难以处理细粒度的用户体验、无障碍与组件组合等问题，易出错。文章认为应将 AI 作为辅助工具，而非试图替代有经验的前端工程师。
Obsidian is a Markdown-based note app with bidirectional links and a knowledge graph. This intro covers core concepts, organizing notes, and why it shines for personal knowledge management and research workflows.
Obsidian 是一个基于 Markdown 的笔记应用，支持双向链接和知识图谱。本文介绍核心概念、笔记组织方式，以及它在个人知识管理与研究工作流中的优势。
A talk that challenges whether mathematical ideas live at a 'big' scale or can be built from small, composable units. It sketches how mathematical abstractions scale in theory and computation, with implications for teaching, formalization, and tool-building.
讨论数学到底是由少数“大”概念驱动，还是由大量可组合的小概念构成。演讲主张将数学视作模块化领域，强调简单且易理解的基本单元如何组合出复杂结构，并探讨对教学、形式化和工具开发的影响。
(87) Zed, A sans for the needs of 21st century (2024)
Zed is Typotheque's sans-serif designed for modern reading needs, with features like high legibility on screens, versatile weights, and multilingual support. It showcases how typography aims to balance readability, aesthetic versatility, and web-friendliness in a global context.
Zed 是 Typotheque 设计的现代无衬线字体，强调在屏幕上提高可读性、提供多样的粗细与广泛的语言支持。它体现了在全球化语境下追求可读性、美学灵活性与网页友好性的字体设计思路。
(88) If you started a company two years ago, many assumptions are no longer true
The piece argues that many startup assumptions made two years ago no longer hold due to shifting funding climate, consumer behavior, and competitive pressure. Founders should re-evaluate markets, unit economics, and strategies, pivoting or sunsetting unviable plans if needed.
两年前创业者的假设如今多已不再成立，原因包括资金环境、市场需求与竞争格局的变化。创始人应重新评估市场、单位经济与策略，必要时调整方向或退出不再可行的计划。
(89) A Canonical Generalization of OBDD
OBDDs offer a canonical representation of Boolean functions used in verification and synthesis. The paper introduces a canonical generalization that preserves canonicity and efficient manipulation, enabling broader use in hardware verification, compiler optimizations, and formal methods. This could simplify reasoning about equivalence and reduce space in large-scale symbolic models.
有序二元决策图（OBDD）提供布尔函数的规范化表示，广泛用于验证与综合。该研究提出一种规范化的广义推广，保持可判定性与高效操作，便于在硬件验证、编译优化和形式方法中获得更通用、规模化的表示。此举可能简化等价性判定并在大规模符号模型中节省空间。
The post argues for a simpler S3 experience or proposes a minimal S3 API layer to reduce boilerplate, with practical tips or a simpler alternative; It may propose a design for a friendlier S3-like storage.
文章主张让 S3 使用更简单，提出简化 API 层或替代方案的要点，降低实现成本并提高开发效率。
(91) Bouncer: Block "crypto", "rage politics", and more from your X feed using AI
An open-source AI tool that blocks categories like crypto and rage politics from your X feed, giving users customizable moderation. It illustrates the potential and risks of automated content filtering on social feeds.
一个开源 AI 工具，可按自定义策略从 X 动态中屏蔽加密货币、愤怒政治等类别，提升浏览体验。它展示了自动化内容过滤的潜力与局限性，以及对偏见和言论自由的权衡。
(92) Textbooks and Methods of Note-Taking in Early Modern Europe (2008)
The paper analyzes how early modern students used textbooks and note-taking methods, revealing practices that influenced knowledge transmission and pedagogy in that era.
论文探讨早期现代欧洲的教科书使用与笔记方法，揭示当时知识传播、教学实践与学习方式的历史脉络。对教育史和信息组织的研究提供了有价值的比较视角。
(93) A Tour of Oodi
A tour of Oodi, Helsinki's flagship library, highlighting architecture, flexible spaces, and public services enabled by modern design. The piece offers takeaways on how libraries combine culture with digital infrastructure.
文章带你走访赫尔辛基 Oodi 图书馆，聚焦建筑风格、灵活公共空间以及现代设计支持的数字化服务。展示图书馆如何在文化与数字基础设施之间取得平衡。
(94) Air Powered Segment Display? [video]
Shows a novel air-powered segmented display prototype, illustrating a playful approach to actuation and typography. The video demonstrates low-power, ambient motion as an alternative to motorized displays and invites discussion on hardware prototyping tradeoffs.
展示了一种新颖的空气驱动分段显示原型，展示了驱动与排版的有趣结合。该视频展示了低功耗、环境友好的驱动方式，探讨了与电机驱动显示相比的权衡。
(95) An Interview with Pat Gelsinger
Pat Gelsinger shares Intel's strategic outlook on AI workloads, manufacturing, and IDM 2.0, providing candid perspectives on the road ahead for semiconductors. The interview sheds light on how Intel plans to compete in a rapidly evolving chip ecosystem.
帕特·盖尔辛格在访谈中谈及英特尔在 AI 工作负载、制造业和 IDM 2.0 的战略展望，提供对半导体行业快速演变格局的直观洞见。
(96) Design and implementation of DuckDB internals
Deep dive into DuckDB's internal architecture, including storage, query execution, and vectorized processing. The write-up explains how DuckDB achieves high performance as an embedded analytics DB and the tradeoffs behind its design. Worth reading for developers building in-process analytics or database systems.
本文深入讲解 DuckDB 的内部架构，涵盖存储结构、查询执行与向量化处理等核心设计。介绍了 DuckDB 作为嵌入式分析型数据库在性能上的取舍与实现细节，如向量化执行与快速启动的设计。对打算构建本地数据库或分析工具的开发者有借鉴意义。
(97) Write less code, be more responsible
Promotes minimalism in code without sacrificing correctness, arguing that fewer lines can reduce bugs and maintenance cost. The piece offers guidelines for safer APIs, clearer abstractions, and responsible design choices that improve long-term reliability.
强调在保持正确性的前提下追求代码简洁，认为更少的代码有助于降低错误和维护成本。给出在设计 API、抽象和实现方面的实用建议，助力提升长期可靠性与可维护性。
(98) What we learned building a Rust runtime for TypeScript
This article shares lessons from building a Rust runtime for executing TypeScript, covering interoperability, performance tradeoffs, and mapping TS semantics to a native runtime. The piece offers practical design choices and cautions for similar language-runtime projects.
总结了用 Rust 构建 TypeScript 运行时的经验教训，涉及互操作性、性能取舍以及将 TypeScript 语义映射到原生运行时的挑战。文章给出实用的设计选项与对类似语言运行时项目的警示。
A playful note on a neat mental-math trick that speeds up multiplication. It showcases how a simple shortcut can make math feel almost magical.
一篇轻松的短文，介绍一个神奇的心算乘法技巧，让乘法变得更快、像魔法一样有趣。
(100) Cooperative Vectors Introduction
An introductory piece on cooperative vectors, outlining how vector representations can encode collaboration and coordination in multi-agent settings. The article sketches potential uses in optimization and AI systems, suggesting a framework for studying cooperative behavior in computational contexts.
本文对合作向量进行了入门介绍，阐释如何通过向量表示编码多智能体环境中的协作与协同。文章勾勒出在优化与 AI 系统中的潜在应用，并提出用于研究计算场景中合作行为的框架。