Hacker News: cma

New comment by cma in "Statement on US government directive to suspend access to Fable 5 and Mythos 5"

cma — Sat, 13 Jun 2026 01:35:58 +0000

They are saying that comparison to other models only about the problems it was jailbroken to complete in the government's example, not all vulnerabilities it could exploit unjailbroken.

New comment by cma in "Blue Origin's New Glenn blows up during static fire test"

cma — Fri, 29 May 2026 12:42:06 +0000

Or because it was a thing Elon pursued: https://arstechnica.com/space/2025/05/spacex-pushed-sniper-t...

New comment by cma in "The real cost of owning a home"

cma — Tue, 26 May 2026 19:47:08 +0000

That's pretty contingent. If you had Snapchat shares instead you'd be ahead.

New comment by cma in "Google facing court for retaliation against Gaza whistleblower"

cma — Mon, 25 May 2026 19:03:57 +0000

Ratified international treaties are the supreme law of the land in the US, and the whistleblower protection stuff being invoked is claimed to be from retaliation to a crime allegation.

New comment by cma in "Magnifica Humanitas"

cma — Mon, 25 May 2026 14:56:09 +0000

Yes, it starts slow with a lot of history of Cybernetics stuff through Leibniz and stuff (kind of prescient given chain rule -> back propagation, and control theory's relevance to optimizers). It is about twice as long as God & Golem and covers a lot of the same plus more examination of automation and human augmentation. I think either it or Cybernetics also goes into mass communication with some relevance to how social media played out.

God & Golem is the most succinct and up to date though, probably a 2hr read.

The book Cybernetics is a lot of math and ergodic theory stuff that went beyond me, but is the longest and you still get a lot out of it skimming over that stuff if you don't have the background for it. The last revision of it in the 50s added some of the same blackbox function copying/imitation learning/distillation stuff, reinforcement learning with reward hacking concerns, and superintelligence as genie/monkey's paw.

I would read the three in reverse order of publication.

He also foresaw another big area of potential existential danger, Wiener filter for guidance and control of missiles (later superceded with Kalman filter bringing the nuclear hard targets era with 15min retaliation windows) and refused to work on it or share prior work, and he also had bioweapons delivery concerns before the bioweapons treaties, publishing this open letter in The Atlantic in 1947:

https://archive.ph/D7BPt

New comment by cma in "Magnifica Humanitas"

cma — Mon, 25 May 2026 12:16:11 +0000

I would also recommend God and Golem, Inc., A Comment on Certain Points Where Cybernetics Impinges on Religion:

https://monoskop.org/images/1/1f/Wiener_Norbert_God_and_Gole...

Norbert Weiner was an atheist but he talks about three areas religion is the only thing to have really examined that relate to capable AI: omniscience, omnipotence, and worship (gadget worship). It has very prescient stuff on blackbox learning/distillation, reinforcement learning/reward hacking, alignment through human feedback.

His The Human Use of Human beings and Cybernetics are extremely good too and have more of a mash of the themes between Rerum Novarum and Magnifca Humanitas, and more near-term automation.

New comment by cma in "SpaceX launches Starship v3 rocket"

cma — Mon, 25 May 2026 11:10:51 +0000

I thought Starship has many unique tiles, they ended up needing to vary thickness to match heat patterns to save weight and have some complicated geometry near the fins.

Shuttle had more unique perimeter shapes, but starship still has a lot of variation due to tickness. I don't know if that is easier to vary in production or still needs custom molds for the variation.

New comment by cma in "Greg Brockman interview [video]"

cma — Mon, 25 May 2026 00:45:00 +0000

Part of the counter was Microsoft was going to try and hire everyone individually, compensating for the lost stock appreciation, without the org itself, if they went through with maintaining the firing. I think that could have been much harder to pull off, but maybe they made them believe it was an inevitable outcome.

New comment by cma in "An OpenAI model has disproved a central conjecture in discrete geometry"

cma — Mon, 25 May 2026 00:34:17 +0000

No, problem #326, you didn't give a few days timeline.

Google released a paper about solving 9 more Erdos problems for an average of $100 each:

https://arxiv.org/html/2605.22763v1

In a year I think we'll probably have seen hundreds of open problems solved, even if there is some a low hanging fruit exhausiton bottleneck.

New comment by cma in "Greg Brockman: Inside the 72 Hours That Almost Killed OpenAI"

cma — Sun, 24 May 2026 11:24:04 +0000

Unsupervised

New comment by cma in "An OpenAI model has disproved a central conjecture in discrete geometry"

cma — Fri, 22 May 2026 16:40:23 +0000

#326 solved

https://github.com/teorth/erdosproblems/wiki/AI-contribution...

New comment by cma in "Sam Altman Won in Court Against Elon Musk. But, We All Lost"

cma — Fri, 22 May 2026 15:16:02 +0000

Leaked emails I think showed the open source part at the heart of the concept of OpenAI was never serious and was just to help recruit.

New comment by cma in "An OpenAI model has disproved a central conjecture in discrete geometry"

cma — Thu, 21 May 2026 09:40:12 +0000

Earlier of their systems have solve other Erdos problems that people had worked on, this one was more monumental and had had a lot more prior effort that didn't solve, but this isn't a one-off.

New comment by cma in "An OpenAI model has disproved a central conjecture in discrete geometry"

cma — Thu, 21 May 2026 09:29:37 +0000

Same could be said for human mathematicians that learn from tools like Lean.

New comment by cma in "Anthropic is expanding to Colossus2. Will use GB200"

cma — Thu, 21 May 2026 06:59:17 +0000

What's the blind rage, he's totally out in the open.

New comment by cma in "I’ve joined Anthropic"

cma — Tue, 19 May 2026 22:25:01 +0000

There's a wide variety of seriousness https://www.youtube.com/watch?v=FIg4zQKBpAs

New comment by cma in "Waymo updates 3,800 robotaxis after they 'drive into standing water'"

cma — Mon, 18 May 2026 00:02:33 +0000

They had an issue detecting them here. But my understanding was people were saying beyond seeing the flooded section there is a fundamental problem that they can never solve that if there is new construction under the flooded section lowering the level of the road they can never deal with that. But as long as it is not a sudden dip or sinkhole I think HD mapping can still help with that since they can monitor the cars altitude as it traverses it and detect divergence from what is mapped early.

New comment by cma in "Waymo updates 3,800 robotaxis after they 'drive into standing water'"

cma — Sat, 16 May 2026 23:51:44 +0000

I'm assuming there is no issue detecting water over the road, only depth below it that has changed since the HD map was made.

New comment by cma in "Waymo updates 3,800 robotaxis after they 'drive into standing water'"

cma — Sat, 16 May 2026 17:19:59 +0000

People usually drive slow too through flooded sections. They don't have the advantage of any descent of the car diverging from HD maps, or any water level visually below HD map surface serving as an immediate warning.

New comment by cma in "Waymo updates 3,800 robotaxis after they 'drive into standing water'"

cma — Sat, 16 May 2026 01:31:58 +0000

Some of the Tesla robotaxi deployments have relied on HD maps too, or at least they were spotted extensively lidar mapping Austin.