Hacker News. Daily summary. Top 20 stories.
UEFA and its 55 member associations will boycott FIFA competitions over a plan to sell World Cup equity to private investors.
Broad support for UEFA's boycott, with many citing FIFA's greed and the Kushner family's involvement in the sell-off plan.
Cheap TV streaming sticks secretly run ad fraud and proxy scams, says new analysis.
Split over whether ad fraud is victimless; some fear IP reputation damage from proxy abuse.
GitHub launched stacked pull requests in public preview, enabling multi-PR stacks.
Excitement for native stacking, with tips for alternatives and reports of broken squash-merge.
Google DeepMind's Gemini Robotics 2 enables whole-body control, dexterity, and multi-robot collaboration.
Skepticism dominates: demos are cherry-picked, hardware limits persist, and household use is years off.
OpenAI cuts GPT-5.6 Luna prices 80% and Terra 20%, citing model-driven efficiency gains.
Price cuts welcomed, attributed to Chinese model competition and model-driven efficiency gains.
Inference APIs are moving toward non-portable session state, locking users into single providers.
Broad agreement on lock-in concerns, with pushback on whether all opaque features are unjustified.
DeepSeek releases V4-Flash public beta with major agent benchmark gains.
Mixed sentiment: excitement over benchmark gains tempered by harness comparability concerns.
EU Court rules VPNs are lawful, reducing copyright liability for providers.
Thread split: VPNs as essential business tools vs. redundant due to modern encryption.
Google expands the Play Age Signals API globally to share children's ages with apps.
Skeptics see the age API as a slippery slope toward mandatory verification and surveillance.
The AI era is creating new design patterns like streaming text and tiny icons.
Two camps: those seeing new norms in AI design, those attributing homogeneity to LLM code.
The GCC steering committee has banned AI-generated code in legally significant contributions.
Commenters split on enforceability, with one camp calling it unenforceable and another defending quality control.
Article body wasn't reachable. HN discussion still summarized.
Skill is useful for some, dismissed as a trend by others.
Refactoring a 17,000-line Rust file cut AI agent token consumption by 83% per change.
Consensus: refactoring helps AI agents, but LLMs cannot do it well without human guidance.
Two reviewers found 68% of 22 submissions had fabricated citations or LLM-generated content.
Multiple commenters agree the review system is broken, with some calling for payment or deposit fees to filter slop.
Ron Gilbert announced production on Thimbleweed Park 2, due early 2028.
Enthusiastic reception, tempered by curiosity about the sequel's ending after the first game's meta conclusion.
Google used AI agents to find and fix over 1,000 Chrome security bugs in mid-2026.
Skepticism about the blog's one-sided success metrics vs. belief in AI's real security benefits.
New lattice calculations suggest the muon wobble matches the Standard Model, casting doubt on prior data.
Debate on whether science's self-correction is a strength or a weakness, with sympathy for affected PhDs.
Speed has become a moral position that protects bad work by mistaking motion for progress.
Widespread agreement, with racing and military mantras on smoothness, and debate on whether first-movers ever win.
DeepSeek V4 Flash 0731 offers strong performance at roughly one-third the cost of OpenAI Luna Max effort.
DeepSeek V4 Flash is called 2-3x cheaper than OpenAI Luna Max, but slower.