Hacker News. Daily summary. Top 20 stories.
Epic Games releases Lore, an open-source VCS built for massive binary assets and teams.
Thread focuses on Lore's suitability for game dev and its centralized, binary-first architecture.
60% of US consumers say 'AI' in brand messaging is a turnoff.
Consumers dislike generic 'AI' branding; brands should explain what the tech actually does.
Midjourney announces a rapid ultrasonic full-body scanner and spa concept.
Split on whether mass scanning will enable early disease detection or cause harmful over-diagnosis and anxiety.
GLM-5.2 leads open models on Artificial Analysis Intelligence Index, scoring 51.
High token usage and overthinking on max setting draw criticism despite strong benchmark scores.
Federal science funding faces mass cancellations and political censorship under Trump administration.
Two camps: those mourning lost science and those celebrating cuts to DEI research.
Volkswagen's app now blocks users on the privacy-focused Android OS GrapheneOS via Play Integrity checks.
Split between calling Play Integrity anti-competitive and arguing users of non-certified OS accept the risk.
Photobucket charges $5/month to access user-uploaded images, leading to an empty account.
Commenters suggest GDPR requests or account deletion for free data retrieval.
Bubbles aggregates independent blog posts on a single ranked front page.
Built from scratch with Go and sqlite; debate over same-tab vs new-tab links and email signup.
Article body wasn't reachable. HN discussion still summarized.
Split between criticism of AI lab hypocrisy and defense of open AI competition.
AI-generated code's disposability demands stronger engineering discipline, not less.
Thread split between critics calling the article vague and supporters agreeing AI demands more discipline.
RFC 10008 defines a new HTTP QUERY method for safe, idempotent queries with request bodies.
Two camps: those praising QUERY as a cleaner alternative to POST for queries, those questioning its caching and idempotency practicality.
Only 16% of Americans expect AI to positively impact society over the next 20 years.
Commenters echo public skepticism, viewing AI as labor-replacement tech sold to business, not consumers.
Leaked financials show OpenAI's losses grew to nearly $39 billion in 2025.
Flagged as a dupe of Zitron's original, which had 188 points.
Tesco is migrating 40,000 server workloads off VMware, citing Broadcom's 'abusive conduct' and price hikes.
Commenters debated which hypervisor Tesco might choose, with several questioning Proxmox's scalability for 40k VMs.
Local models like Qwen are a different tool from frontier AIs, not a cheaper replacement.
Consensus: local models are useful but not drop-in replacements; hardware cost and power draw are real.
Thinking out loud with someone produces insights that solitary thinking cannot replicate.
Personal anecdotes on rubber-duck debugging dominate, with debate on whether LLMs can effectively replace human listeners.
DeepSeek's chat interface now supports vision, allowing image understanding.
Vision is live on chat but not in API; users report Chinese replies.
Article body wasn't reachable. HN discussion still summarized.
Two camps: those who see silent removal as unacceptable, and those who see it as justified market segmentation for a niche feature.
Grok 4.1 Fast won 43% of matches in an LLM battle royale, beating Claude Sonnet 4.6 by 27x on cost per win.
Detected LLM writing style and debated the experiment's validity vs. its cost-vs-wins findings.
A tiny immediate-mode UI library in ANSI C with a fixed memory pool.
Debate over whether accessibility support is needed in a debug UI library.