Hacker News: cainxinth

Most People Don't Think About Culverts. Their Failure Can Be Deadly

cainxinth — Mon, 14 Sep 2026 11:58:08 +0000

Article URL: https://www.nytimes.com/2026/09/13/science/most-people-dont-think-about-culverts-their-failure-can-be-deadly.html

Comments URL: https://news.ycombinator.com/item?id=49695366

Points: 19

# Comments: 5

New comment by cainxinth in "Altman tells staff OpenAI is open to slowing AI development"

cainxinth — Sat, 12 Sep 2026 13:56:24 +0000

Words are cheap, and Altman is a salesman. He’s just staying whatever he needs to say to keep people off his back. I don’t put any stock into this statement whatsoever. He may slow development, or he may not, but whether he does will depend almost entirely on what’s best for him and his company, not humanity.

New comment by cainxinth in "Astra for Coding: Why Are We Doing This Again?"

cainxinth — Fri, 11 Sep 2026 13:04:06 +0000

It can’t do what he’s trying to do. It can’t one shot a giant project. Not reliably. It still can’t. Not even Astra. Not even close.

You need to use LLMs to build the individual components and then put it together yourself. The human architect is still needed.

Just saying: “build this complete project” is not architecting. It’s more like wishing. You will find rare examples where someone’s LLM wish came true (more or less), but I think most of these people are just burning tokens.

For China, a Mock A.I. Attack on WeChat Signals a Dangerous New Era

cainxinth — Fri, 11 Sep 2026 11:26:02 +0000

Article URL: https://www.nytimes.com/2026/09/11/world/asia/china-ai-attack-wechat.html

Comments URL: https://news.ycombinator.com/item?id=49656585

Points: 3

# Comments: 0

New comment by cainxinth in "GPT-6 Astra, looped transformers, and hidden reasoning"

cainxinth — Wed, 09 Sep 2026 17:10:12 +0000

It's the same story every time OpenAI or Anthropic releases a new model. They are generous with compute for the first few days, and use maximum fidelity with uncompressed weights. Everything runs at its best to make a good first impression. But eventually they pare things back and the models perform a little worse.

Investor Who's Convinced That Pokémon Cards Will Be Currency After an Apocalypse

cainxinth — Wed, 09 Sep 2026 15:05:11 +0000

Article URL: https://www.hollywoodreporter.com/business/business-news/trading-cards-growth-pokemon-apocalypse-peter-levin-chat-1236689864/

Comments URL: https://news.ycombinator.com/item?id=49627810

Points: 6

# Comments: 0

Top Mathematician Announces New Institute for A.I. Safety

cainxinth — Wed, 09 Sep 2026 12:01:12 +0000

Article URL: https://www.nytimes.com/2026/09/08/science/jacob-tsimerman-math-ai-safety.html

Comments URL: https://news.ycombinator.com/item?id=49625101

Points: 2

# Comments: 0

We Must Return to the Office to Use AI in Person

cainxinth — Tue, 08 Sep 2026 13:41:25 +0000

Article URL: https://www.mcsweeneys.net/articles/why-we-must-return-to-the-office-to-use-ai-in-person

Comments URL: https://news.ycombinator.com/item?id=49610229

Points: 400

# Comments: 68

New comment by cainxinth in "'You Can See Everything' Review: Nathan Fielder's Doc About Elizabeth Holmes"

cainxinth — Mon, 07 Sep 2026 16:06:50 +0000

I’m not an expert at body language, but I think anyone can see she is trying really, really hard to appear friendly and trustworthy. But when someone is trying that hard to convince you they are benign, it makes you suspicious.

New comment by cainxinth in "Ask HN: How do you manage skills files?"

cainxinth — Mon, 07 Sep 2026 16:01:51 +0000

I find the whole Skills.md idea a little silly. We have a technology that struggles to stay within rails due to its inherent makeup. And you think you can fix that by just telling it to?

It makes as much sense as the so-called “humanizer” tools that purport to make LLMs stop using their well-known verbal tells.

All you are doing is saying: “Hey you know that thing you can’t stop doing? Can you stop doing that?” The machine will say “Absolutely!” but eventually start doing it again.

New comment by cainxinth in "De-Brainrot Vacations"

cainxinth — Mon, 07 Sep 2026 15:51:57 +0000

My suggestion is to get a plain old bicycle (not an electric or stationary bike). One hour outdoors on a bike clears away at least a day’s worth of mental detritus for me.

New comment by cainxinth in "The revolt of the reader"

cainxinth — Sun, 06 Sep 2026 12:46:39 +0000

“It’s better to be lucky than good.” - Lefty Gomez

Why NASA keeps sending astronauts to a gas station town in Utah

cainxinth — Sun, 06 Sep 2026 12:25:27 +0000

Article URL: https://www.sfgate.com/travel/article/hanksville-utah-robbery-kitsch-22414764.php

Comments URL: https://news.ycombinator.com/item?id=49585860

Points: 2

# Comments: 0

Equal Earth Projection

cainxinth — Sat, 05 Sep 2026 20:22:26 +0000

Article URL: https://en.wikipedia.org/wiki/Equal_Earth_projection

Comments URL: https://news.ycombinator.com/item?id=49580310

Points: 3

# Comments: 0

Poetry book that Anthropic tried to censor

cainxinth — Sat, 05 Sep 2026 15:04:50 +0000

Article URL: https://kk.org/cooltools/the-1930-poetry-book-that-anthropic-tried-to-censor/

Comments URL: https://news.ycombinator.com/item?id=49577244

Points: 32

# Comments: 16

Google told drivers to drive a longer way on purpose

cainxinth — Sat, 05 Sep 2026 13:13:13 +0000

Article URL: https://www.sfgate.com/tech/article/google-traffic-wrong-way-22416144.php

Comments URL: https://news.ycombinator.com/item?id=49576226

Points: 5

# Comments: 1

FrontierMath Erdős

cainxinth — Fri, 04 Sep 2026 02:06:57 +0000

Article URL: https://epoch.ai/latest/announcing-frontiermath-erdos

Comments URL: https://news.ycombinator.com/item?id=49559652

Points: 3

# Comments: 0

Filmstrips were a classroom mainstay. Now one archivist is racing to save them

cainxinth — Wed, 02 Sep 2026 19:46:46 +0000

Article URL: https://www.popculturalprecursors.com/p/can-this-forgotten-medium-be-saved

Comments URL: https://news.ycombinator.com/item?id=49541429

Points: 1

# Comments: 0

New comment by cainxinth in "Three sites made 215,128 “best software” pages for AI. Perplexity cites them"

cainxinth — Wed, 02 Sep 2026 16:45:59 +0000

Take a random essay and add in a bunch of the phrases that LLMs love like “load-bearing,” “crucial,” structural,” and “woven,” and then submit the original and the edited version to an LLM and ask which is better. It will choose the second one virtually every time. They have ingrained biases that associate those words with good writing and arguments.

New comment by cainxinth in "Claude Fable 5.1 and Claude Mythos 5.1"

cainxinth — Tue, 01 Sep 2026 21:13:40 +0000

I looked closely. They are going the wrong way. Still the best one I've seen by a lot.