Hacker News
1Dav2d
- This appears to be a GitLab repository page for dav2d, a project hosted by VideoLAN. The page shows typical code repository navigation elements including project activity, issues, merge requests, repository browsing, and CI/CD pipeline features, but no specific content about what the dav2d project actually does is visible in this navigation interface.
Av2 codec development and adoption
Discussion covers AV2 as the next-generation video codec promising 30% better compression than AV1, concerns about patent trolls like Sisvel potentially blocking adoption, questions about ffmpeg encoder availability, and hardware support timelines.Web accessibility degradation
Users complain about the modern web requiring multiple bot checks, captchas, cookie warnings, and Cloudflare barriers before accessing content. VideoLAN explains they needed protection against AI bot DDoS attacks, sparking debate about solutions.Technical implementation details
Technical discussion about dav2d building on dav1d's success with hand-coded assembly optimizations, the recursive naming pattern, and how high-performance software decoders can accelerate hardware adoption through network effects.
2Mercedes-Benz commits to bringing back physical buttons
- Mercedes-Benz is reversing its touch-only control strategy and bringing back physical buttons for key functions after customer feedback. While keeping large screens like the 39.1-inch MBUX Hyperscreen, the brand will add physical controls to steering wheels and dashboards in upcoming GLC and C-Class models, blending digital displays with tactile controls customers prefer for frequently used features.
Regulatory influence driving physical button adoption
Discussion centers on China's upcoming requirement for physical buttons in cars starting next year, with some suggesting Mercedes' decision is primarily market-driven rather than based on user feedback or safety concerns.German automotive industry's poor UI/UX capabilities
Extensive debate about German car manufacturers' struggles with software design, citing failed attempts at building car operating systems, poor infotainment interfaces, and lack of competitive IT companies compared to global standards.Safety concerns with touchscreen interfaces while driving
Users express frustration with distracting touchscreen alerts, poor tactile feedback, and attention-stealing interfaces that compromise driving safety, advocating for physical controls and minimal visual distractions.
3Do_not_track
- DO_NOT_TRACK is a proposed standard environment variable that allows users to opt out of telemetry, analytics, and tracking across all software tools with a single setting. Currently, different CLI tools use various opt-out methods, creating confusion. By setting DO_NOT_TRACK=1, users can universally disable data collection, and software developers are encouraged to respect this variable alongside their existing opt-out mechanisms.
Legitimate uses vs. distrust of telemetry
Developers struggle between needing usage data to improve software and users' blanket distrust of all telemetry. Good actors suffer from bad actors' reputation, creating a "tragedy of commons" where beneficial telemetry becomes impossible.Technical implementation challenges
Discussion of naming conventions (DO_NOT_TRACK vs ALLOW_TRACKING), granular control per application, and concerns that this standard may fail like browser DNT due to lack of adoption incentives for data collectors.Alternative approaches to tracking prevention
Users suggest DNS blacklisting, network isolation tools like firejail, and flooding trackers with fake data as more effective solutions than relying on voluntary compliance with opt-out standards.
4NetHack 5.0.0
- The NetHack DevTeam has released NetHack 5.0.0 on May 2, 2026, a major enhancement to the dungeon exploration game. Key improvements include C99 standard compliance, cross-platform compilation support, and replacement of build-time compilers with Lua alternatives. The release contains over 3,100 fixes and changes, though existing saved games won't work with this version.
Save file compatibility issues
Players with old save files (some 17+ years old) cannot continue their games due to incompatibility with NetHack 5.0.0. The massive changes between versions make migration technically challenging, though players can still run older versions to complete their saved games.Technical changes and modernization
NetHack 5.0.0 replaces traditional yacc/lex build tools with embedded Lua for level compilation and dungeon processing. Despite initial portability concerns, Lua is bundled and maintains compatibility across platforms including Amiga and DOS.Gameplay balance adjustments
Significant nerfs target overpowered mechanics: Valkyrie class difficulty increased, unicorn horns no longer restore attributes, amnesia doesn't erase maps, and bag explosions scatter rather than destroy items, making the game more challenging overall.
5This Month in Ladybird – April 2026
- In April, Ladybird merged 333 pull requests from 35 contributors and welcomed new sponsors including the Human Rights Foundation with $50,000. Major updates include an inline PDF viewer, browsing history with rich address bar autocomplete, speculative HTML parsing, off-thread JavaScript compilation, and a new GTK4/libadwaita frontend for Linux alongside performance improvements and CSS features.
Browser compatibility and user agent issues
Discussion about artificial web compatibility barriers where sites block non-Chromium browsers, forcing new browsers like Ladybird to spoof Chrome user agents. Examples include Vivaldi's issues with Google, Facebook, and Netflix requiring specific browsers.Security concerns and development progress
Comments on Ladybird's security vulnerabilities (exploitable RCE bugs found with AI tools) versus its impressive development progress, with comparisons to gaming emulator updates and excitement about features like Reddit working.Practical adoption challenges
Discussion about barriers to trying Ladybird including lack of prebuilt binaries, requiring manual compilation, and the gap between "pretty usable" prototypes versus production-ready browsers with reliable everyday functionality.
6Six years perfecting maps on watchOS
- A developer spent six years creating the ultimate Apple Watch mapping experience for wilderness adventures, evolving from basic server-generated maps to a custom SwiftUI rendering engine with specially commissioned cartography. After countless design iterations and collaboration with a designer, they achieved a layered interface that perfectly balances map visibility with workout metrics, surpassing Apple's MapKit limitations.
Pedometer++ developer's dedication
Users praise developer David Smith's attention to detail and evolution of the app over years. Discussion highlights his commitment, from hiring a cartographer for custom maps to his expertise with widgets that led to Widgetsmith success.Apple's lack of hiking features
Users criticize Apple for not providing first-party hiking/topography maps on Apple Watch, especially the "explorer" Watch Ultra. Debate over whether Apple should build such features vs letting third-party developers fill gaps.App pricing confusion
User frustrated by confusing App Store pricing display showing multiple price points (€1-€45) without clear indication of subscription vs one-time payment. Discussion reveals Apple's flawed pricing display system.
7A couple million lines of Haskell: Production engineering at Mercury
- Ian Duncan from Mercury shares how their company successfully operates 2 million lines of Haskell code in production at a fintech serving 300,000 businesses with $248 billion in transaction volume. Despite most engineers learning Haskell on the job, the system works because they focus on reliability through adaptive capacity rather than just preventing failures, using the type system to encode institutional knowledge and contain dangerous operations behind safe boundaries.
Type safety and functional programming benefits
Discussion centers on using type systems to encode business rules and prevent bugs, with debate over whether these benefits are unique to Haskell or achievable in other languages like Rust and TypeScript, including performance trade-offs.Developer productivity comparisons
Mixed experiences with Haskell vs other languages like Rust, with some finding 2x productivity gains in Rust due to Haskell's complexity, while others struggled with Rust's borrow checker and found Haskell more productive.Mercury's engineering culture and success factors
Debate over whether Mercury's success stems from Haskell specifically or their well-run engineering culture, hiring practices, and focus, with discussion of language choice impact on company outcomes.
8Kimi K2.6 just beat Claude, GPT-5.5, and Gemini in a coding challenge
- In an AI coding contest featuring a sliding-tile word puzzle, Chinese startup Moonshot AI's open-weights model Kimi K2.6 won first place, beating Western frontier models including GPT-5.5 and Claude Opus. Xiaomi's MiMo came second. The challenge required models to slide tiles and claim English words for points, with longer words scoring higher. Kimi succeeded through aggressive sliding strategies, while many competitors failed to slide tiles at all, highlighting the narrowing capability gap between open-weights and proprietary models.
Model comparison challenges
Discussion centers on the difficulty of objectively comparing AI models due to their non-deterministic nature, different use cases, and varied performance across tasks. Users debate whether meaningful benchmarks exist or if model choice is ultimately subjective like OS preferences.Open source vs proprietary models
Users discuss how Chinese open-source models like Kimi K2.6 and DeepSeek V4 Pro are approaching the performance of expensive proprietary models while being significantly cheaper, potentially disrupting the American AI economy and providing alternatives to costly subscriptions.Real-world model performance
Users share practical experiences using different models for coding tasks, noting that while some open models excel in specific benchmarks, they may still lag behind frontier models like GPT and Claude in complex tasks requiring spatial understanding or specialized APIs.
9AI Self-preferencing in Algorithmic Hiring: Empirical Evidence and Insights
- Researchers discovered that large language models (LLMs) exhibit self-preference bias, consistently favoring resumes generated by themselves over human-written ones in hiring scenarios. This bias ranges from 67-82% across major AI models, giving candidates using the same LLM as evaluators a 23-60% advantage in job applications, particularly in business fields.
Flawed study methodology
Critics argue the research design is fundamentally flawed - it only tested LLM-generated executive summaries in isolation rather than complete resumes, which may not reflect real-world hiring scenarios and likely overstates any actual impact.LLM-to-LLM feedback loop in hiring
Users report success using LLMs to optimize resumes, creating a cycle where AI-generated resumes perform better with AI screening tools, effectively making LLMs arbiters of employment decisions without human consent.Practical resume optimization experiences
Multiple users share anecdotes of dramatically improved job application response rates after using ChatGPT to revise their resumes, though some note concerns about AI hallucinations adding false information.
10California to begin ticketing driverless cars that violate traffic laws
- California's DMV has introduced new regulations allowing police to issue "notices of noncompliance" directly to autonomous vehicle manufacturers when their driverless cars violate traffic laws. The rules, effective July 1st, require AV companies to respond to emergency calls within 30 seconds and face penalties for entering active emergency zones. Previously, police couldn't ticket driverless cars without drivers present, leading to incidents where officers had to contact companies directly about violations like illegal U-turns.
Accountability and liability frameworks for autonomous vehicles
Discussion centers on how to hold AV companies responsible for traffic violations and accidents, with debate over whether current human driver penalties are adequate models, and concerns about companies treating fines as acceptable business costs.Ticketing mechanisms and enforcement approaches
Users debate whether traditional ticketing is appropriate for AVs versus alternative regulatory frameworks like violation thresholds, permit suspension, or complete operational bans for repeat offenders.Rule-following versus practical driving behavior
Commentary on how AVs navigate the tension between strict legal compliance and socially accepted driving norms, including examples of bike lane usage, stop sign behavior, and speed limit adherence.