Hacker News. Daily summary. Top 20 stories.
AI's mathematical edge may come from vastly larger working memory, not superior reasoning.
Two camps: those fearing incomprehensible AI proofs versus those arguing abstraction and study preserve human understanding.
Semaglutide linked to lower predicted dementia risk based on proteomics signature.
Two camps: effects from weight loss vs. direct anti-inflammatory benefits.
Article body wasn't reachable. HN discussion still summarized.
LLMs excel at autonomous optimization loops with verification and profiling.
A man is falsely matched to a non-existent person on a U.S. sanctions list.
Commenters share costly false-positive experiences and criticize fuzzy name-only matching as a systemic failure.
Super El Niño strengthening rapidly, forecasts project a record event for winter 2026/2027.
Split between alarm over record strength and skepticism of catastrophic predictions.
Working with AI feels more like leadership through conversation than issuing commands to a machine.
Split between those who see leadership parallels and those insisting deep technical skills remain essential.
Waist circumference predicts heart disease risk more accurately than BMI alone, study finds.
Many commenters say this finding is well-known; BMI remains popular for its simplicity despite known flaws.
Article body wasn't reachable. HN discussion still summarized.
Personal accounts of Lyme disease in UK and US, plus climate change driving tick spread.
LLMs excel at narrow coding tasks but cannot replace deep engineering reasoning like maintainability and architecture.
Two camps: some report full agentic maintenance of 150k-LOC apps, others find generated code haphazard without human architecture oversight.
Solitude and inner attention, not coworking, are essential for generating new ideas.
Thread splits on whether solitude or collaboration better fosters original ideas.
LLMs trained only on K-5 material reveal post-training cannot exceed the pretraining data ceiling.
Skepticism about whether the filtered dataset truly excludes advanced concepts vs. poor model recall.
A controversial neck surgery for Alzheimer's may improve symptoms, but scientists are deeply divided.
Split between cautious hope for temporary relief and skepticism of unproven treatments preying on desperation.
Engineers who use AI agents are unknowingly reinventing program management and waterfall.
Commenters largely agree but caution against taking waterfall advice literally.
AI in drug discovery hasn't yet delivered real clinical impact, experts say.
Two camps: those cautioning that AI drug discovery hasn't yet proven clinical impact, and those arguing it just needs time.
Anthropic finds AI agents lack coordination, exhibit uniformity, and are vulnerable to deception.
Skepticism that the study is a promo for Anthropic's next model; the behaviors are blamed on training data quality.
In Loosdrecht, Netherlands, houses sit on skinny manmade islands within a lake.
Two camps: holiday vs primary residence, with some noting million-euro prices.
ThoughtDAG creates editable context graphs where edges control LLM context inclusion.
Interest in the concept but strong preference for plugin over standalone; security patch issued for shell injection.
LLMs remain unproven for non-trivial software, with thin evidence and poor output quality.
Sharp divide: skeptics cite thin evidence and poor code; proponents report real personal gains from careful use.
Satellite images show Europe's heatwave-ravaged landscapes: scorched fields, shrunken glaciers, low rivers.
Fatalism dominates: individual sacrifice vs. systemic inaction, with personal accounts of Europe's heat.