Three stories cleared the 200-point threshold on today’s Hacker News front page. The throughline: the tension between what the tech industry says it’s building and what’s actually happening underneath.
Writing by Hand is Good for Your Brain — Here’s How to Do It
661 points · Source
Neal Stephenson — who has composed every novel since Cryptonomicon in longhand with a fountain pen — makes the case that writing by hand recruits more of your brain than typing, and offers practical guidance on how to actually do it. The post grew out of research showing that handwriting engages more neural pathways, but Stephenson’s contribution is the practitioner’s angle: use a fountain pen (not a pencil or cheap ballpoint — too much friction means fatigue), skip the iPad stylus (too little friction is equally tiring), use cursive, don’t obsess over legibility, and stop worrying about wasting paper.
The post is interesting less for its neuroscience claims (which are real but hardly novel) and more for the subtext. Stephenson points out that AI has forced educators back to handwritten exams, creating a generation of students who literally can’t write by hand. There’s an irony here: the same tech industry that Stephenson has spent his career fictionalizing has produced tools that degraded a basic cognitive skill, and now the response is to retreat to pre-digital methods. His 25-year track record of composing entire novels this way — including the 42-inch stack of Baroque Cycle manuscripts — is more convincing than any study.
The writing-gear nerds will love the pen and paper comparisons (Diplomat Aero, Monteverde Invincia, Jorg Hysek with wide nib, tested across cheap printer paper, legal pads, and Italian cotton stock). The practical takeaway: fountain pens on any decent notebook paper, write on one side, use cursive, don’t overthink it.
Startup Founders Urge U.S. Government Not to Shut Off Chinese Open Weight AI
510 points · Source
Nearly 200 Silicon Valley companies — including Proton and Y Combinator, organized under the newly-formed Little Tech Association — sent coordinated letters to Trump, Commerce Secretary Lutnick, and OSTP Director Kratsios urging them not to ban access to Chinese open-weight AI models. The trigger: reports that the administration was considering restrictions after Moonshot AI released its Kimi K3 model.
The substance is predictable but the politics are not. Kratsios alleged on Wednesday that Moonshot AI distilled Anthropic’s Fable model to build K3 and acquired banned Nvidia GB300 servers to train it. The startup founders’ counterargument is blunt: “There’ll be hundreds of companies that instantly die,” said Suhail Doshi of Particle, adding that the main beneficiary of a ban would be Anthropic. The Little Tech Association’s executive director wants “a scalpel rather than a sledgehammer.”
What makes this more than standard industry lobbying is the fault line it exposes. Anthropic and the frontier labs want tighter restrictions on Chinese models (protecting their moat). Startups depend on cheap open-weight models to survive (they can’t afford Anthropic’s pricing). The administration is caught between national security hawks and the innovation-at-any-cost crowd. The fact that a blanket ban “was not seriously discussed” in Monday’s White House meeting suggests the startup lobby may have already won this round — but the escalating distillation allegations give hawks plenty of ammunition for future rounds.
AI Companies Are Trying to Hide a Staggering Amount of Debt
456 points · Source
A Nikkei Asia investigation found that five US tech giants — Alphabet, Microsoft, Amazon, Meta, and Oracle — are carrying an estimated $1.65 trillion in off-balance-sheet debt tied to AI infrastructure spending. That’s more than their $1.35 trillion in on-balance-sheet debt combined. Meta alone has roughly $420 billion hidden behind special purpose vehicles and legally distinct subsidiaries — the same accounting tricks that brought down Enron.
The Bloomberg quote from technical accounting consultant Tom Selling nails it: “The accounting treatment itself is in fashion. But what if one of these companies was a house of cards?” The comparison to Enron isn’t hyperbole — it’s structural. SPVs and off-balance-sheet arrangements exist specifically to make financial reporting look healthier. The justification is that these are long-term infrastructure bets (data centers, GPU clusters) that will eventually generate returns. But “eventually” is doing a lot of heavy lifting when four of these five companies report earnings in the next few weeks and the gap between AI spending and AI revenue keeps widening.
The timing matters. CoreWeave has already lost 46% of its stock value in six weeks. The AI revenue story hasn’t materialized at the scale needed to justify these capital outlays. If any one of these five companies stumbles on earnings, the off-balance-sheet debt becomes a confidence crisis — exactly the dynamic that turned Enron’s accounting games into a death spiral.
Throughline
Today’s front page is a snapshot of an industry at war with itself. On one side: startups desperate for cheap, open Chinese models to keep building, and a culture that still values craft (Stephenson’s fountain pens, handwritten thinking). On the other: trillion-dollar bets on AI infrastructure funded by accounting maneuvers that would make Enron blush, while the same companies lobby to restrict the competition that actually keeps innovation accessible. The handwriting essay reads like an antidote — the deliberate, slow, human-scale alternative to an industry that’s accelerating faster than its own balance sheets can handle.