0:00 / 4:38generalfeature Neal Stephenson makes the case for writing by hand#
Neal Stephenson argues that writing by hand recruits more of the brain than typing and says he has composed most of his books for the last 25 years with a fountain pen on paper. The piece is part personal workflow, part practical guide: avoid pencils and cheap ballpoints if they require too much pressure, choose pen-and-paper combinations that feel right, don’t be overly frugal with paper, and don’t assume left-handedness rules out fountain pens. It also lands in a timely context: Stephenson notes that AI has pushed some educators back toward in-person, longhand exams, exposing how many students have little practice writing by hand.
Discussion: Mixed — HN was broadly receptive to handwriting as a useful habit, with many commenters saying it helps memory, focus, studying, brainstorming, or daily work logs. But the thread was not uncritical: several pushed back on the idea that “more brain activity” automatically means better learning, defended iPads and digital-note workflows, or raised fatigue, left-handedness, and dysgraphia as real barriers. (personal note-taking and study habits, skepticism about brain-activity claims, fountain pens, gel pens, pencils, and paper preferences)
0:00 / 3:38sciencedeep dive Astronomers spot a possible first exomoon — but it’s not like ours#
ESO says observations from the Very Large Telescope show strong evidence for a Jupiter-mass object orbiting a brown dwarf in the CD-35 2722 system. Because that brown dwarf itself orbits a star, the team calls the object an exosatellite, but the discovery blurs the usual line between planet and moon. If confirmed, it would be the first plausible detection of a natural satellite outside the Solar System, a goal astronomers have pursued despite thousands of known exoplanets.
Discussion: Mixed — HN was broadly intrigued by the possible exomoon detection, but much of the discussion quickly turned pedantic and cautious about terminology. Commenters debated whether an object orbiting a brown dwarf should be called a moon, an exosatellite, or something closer to a planet, while others focused on brown dwarf physics, the ESO image, and the Atacama observing site. (excitement about a possible first exomoon, debate over moon versus planet terminology, brown dwarf and gas giant physics)
0:00 / 3:17aideep dive DARPA puts an AI agent in control of a modified F-16#
DARPA and the U.S. Air Force say a VENOM-modified F-16 is undergoing in-air testing with an AI agent autonomously controlling flight. The aircraft uses a VENOM Autonomy Kit that interfaces with flight controls and mission systems while allowing a pilot to switch between human and AI control. The program builds on earlier ACE work with the X-62A VISTA and is meant to support DARPA’s AIR program, including testing multiple AI agents in live-flight scenarios and future human command of autonomous aircraft teams.
Discussion: Mixed — HN reacted with a heavy dose of Skynet and apocalypse jokes, but the substantive thread was more cautious than celebratory. Commenters questioned the safety model for handing control back to a pilot, debated whether modified F-16s are meaningful in modern air combat, and compared the work to existing drone and target-aircraft programs. A smaller set pushed back that this is experimental flight testing with trained pilots and that military aerospace validation is not casual software development. (Skynet and autonomous-weapons jokes, concern about human handoff from AI control, skepticism about hype and military strategy)
0:00 / 0:02ai Startups push back on a possible U.S. ban on Chinese open-weight AI#
Politico reports that nearly 200 Silicon Valley companies, including Proton and Y Combinator, signed a Little Tech Association letter urging the Trump administration not to cut off U.S. access to Chinese open-weight AI models from companies such as Moonshot AI and Alibaba. The founders argue that broad bans would not stop global model proliferation, but would raise costs for startups and tilt the market toward large U.S. AI providers. The article also notes that while administration officials have escalated rhetoric around alleged Chinese distillation of U.S. models, a blanket ban was reportedly not seriously discussed and Commerce had not drafted plans to add Chinese AI companies to its Entities List as of Wednesday.
Discussion: Negative — HN commenters were overwhelmingly hostile to the idea of banning Chinese open-weight models, seeing it as protectionism or regulatory capture for U.S. AI incumbents rather than an effective security measure. A minority raised strategic and security arguments, including the need to preserve an independent U.S. AI stack or worry about adversarial models, but those views were mostly drowned out by skepticism about enforceability and fairness. (regulatory capture concerns, skepticism that open weights can be effectively banned, fear of advantaging OpenAI and Anthropic)
0:00 / 1:59security The Numbers went dark, and AI bot traffic is the prime suspect#
Stephen Follows reports that film-data site TheNumbers.com disappeared on March 5, 2026, then returned about a week later in a stripped-down form after what founder Bruce Nash described as a collapse amid heavy automated traffic and suspicious probing of old infrastructure. The site, which tracks tens of thousands of films and is used by journalists, academics, filmmakers, and prediction markets, had been running on decades-old code with roughly 160,000 source files. The broader warning is that AI crawlers and agentic scraping can impose real costs on small public-interest websites, while any security weakness in those sites can become financially valuable when their data resolves markets or business decisions.
Discussion: Mixed — HN was broadly worried about AI crawlers overwhelming small sites, but split on whether this was mainly a new internet-wide failure mode or a solvable architecture problem. Several commenters shared similar scraper-cost stories, while others argued static publishing, caching, BigQuery-style public datasets, CDNs, or robots.txt could mitigate the damage. A second thread focused on whether mass scraping is making people less willing to publish free work on the open web. (AI crawlers as de facto DDoS traffic, fragility of old but valuable public websites, static generation, caching, CDNs, and public datasets as mitigations)