Hacker News: blazespin

New comment by blazespin in "Muse Spark: Scaling towards personal superintelligence"

blazespin — Wed, 08 Apr 2026 21:26:05 +0000

Because bots and trillion dollar ipos and even bigger stakes. People need to better appreciate the level of manipulation going on. Social media has an outsized impact. Bots and even people are getting paid to post and upvote/downvote narratives.

New comment by blazespin in "GLM-5.1: Towards Long-Horizon Tasks"

blazespin — Tue, 07 Apr 2026 22:00:50 +0000

Anthropic's reply? A model you can't use.

New comment by blazespin in "Project Glasswing: Securing critical software for the AI era"

blazespin — Tue, 07 Apr 2026 21:19:55 +0000

Dario is big on beating china, and no doubt he believes cyber security is how to do that. You can tell, but anthropic is sht at everything else. Nobody uses it for real research.

New comment by blazespin in "System Card: Claude Mythos Preview [pdf]"

blazespin — Tue, 07 Apr 2026 20:51:10 +0000

Anthropic needs money like the 112B OpenAI got. They could be hyping and this is good hype. Who knows how benchmaxxed they are.

If they provide access to 3rd party benchmarking (not just one) than maybe I'll believe it. Until then...

New comment by blazespin in "System Card: Claude Mythos Preview [pdf]"

blazespin — Tue, 07 Apr 2026 20:48:53 +0000

Yeah, need some good RE benchmarks for the LLMs. :)

RE is very interesting problem. A lot more that SWE can be RE'd. I've found the LLMs are reluctant to assist, though you can workaround.

New comment by blazespin in "A new Polymarket account made over $500k betting on the U.S. strike against Iran"

blazespin — Sun, 01 Mar 2026 20:00:22 +0000

All I've ever seen is it encouraging people and not discouraging them. The thrill of the easy money, right?

Forget it jake, it's Polymarket.

New comment by blazespin in "A new Polymarket account made over $500k betting on the U.S. strike against Iran"

blazespin — Sun, 01 Mar 2026 19:56:55 +0000

Yeah, trying to beat the market on actually predicting will get you pretty lame returns. Probably do better in non zero sum games like the stock market. At least there you get the benefit of the market always going up eventually.

No, the best way to win on Polymarket is purely by insider trading. Which is why it's a useful thing to watch. Insider news..

That said, the definition of 'insider trading' is always tricky. At what point does it become insider? Some things people call insider others just call clever detective work.

New comment by blazespin in "A new Polymarket account made over $500k betting on the U.S. strike against Iran"

blazespin — Sun, 01 Mar 2026 19:55:17 +0000

"People don't play in corrupt markets for very long." ... ahhh, it's not a bug with Polymarket - it's a feature.

New comment by blazespin in ""Cancel ChatGPT" movement goes mainstream after OpenAI closes deal with U.S. Dow"

blazespin — Sun, 01 Mar 2026 19:39:42 +0000

I'd say it's more a kowtow to his voters and democracy. Whatever you think of him, he was democratically elected. Do remember to show up this November, though, and remind all your friends..

New comment by blazespin in "US blocks all offshore wind construction, says reason is classified"

blazespin — Tue, 23 Dec 2025 19:41:37 +0000

trump literally said he wants venez to return the oil it 'stole' when it nationalized.

Advanced fusion control breakthrough brings clean, reliable energy closer

blazespin — Wed, 17 Dec 2025 16:55:31 +0000

Article URL: https://www.jpost.com/science/article-880512

Comments URL: https://news.ycombinator.com/item?id=46302026

Points: 2

# Comments: 0

New comment by blazespin in "AI agents are starting to eat SaaS"

blazespin — Mon, 15 Dec 2025 00:47:39 +0000

There is a significant risk of uncertainty in all of this, the most damaging aspect really. If AI improves, and it is threatening to, then growth in SaaS may decline to a point where investing in it needs to be reconsidered.

The problem is, nobody knows how much and how fast AI will improve or how much it will cost if it does.

That uncertainty alone is very problematic and I think is being underestimated in terms of its impact on everything it can potentially touch.

For now though, I've seen a wall form in benchmarks like swe-rebench and swebench pro. Greenfield is expanding, but maintenance is still a problem.

I think AI needs to get much better at maintenance before serious companies can choose build over buy for anything but the most trivial apps.

New comment by blazespin in "The Gorman Paradox: Where Are All the AI-Generated Apps?"

blazespin — Sun, 14 Dec 2025 23:16:55 +0000

Nobody collapses, everything just shrinks.

And we're seeing that in the labor numbers.

Sometimes things are harder to see because it's chipping away and everywhere at the margins.

New comment by blazespin in "DeepSeekMath-V2: Towards Self-Verifiable Mathematical Reasoning [pdf]"

blazespin — Fri, 28 Nov 2025 20:13:06 +0000

who likely wins, fify

New comment by blazespin in "DeepSeekMath-V2: Towards Self-Verifiable Mathematical Reasoning [pdf]"

blazespin — Fri, 28 Nov 2025 20:11:00 +0000

if you read the paper that is the intention, to guide stuff like lean.

i don't think llm is a great pure rlvr

New comment by blazespin in "DeepSeekMath-V2: Towards Self-Verifiable Mathematical Reasoning [pdf]"

blazespin — Thu, 27 Nov 2025 23:08:22 +0000

Advanced math solving, as the results indicate. Informal proof reasoning is advancing faster than formal proof reasoning because the latter is slow and compute intensive.

I suspect it's also because there isn't a lot of data to train on.

New comment by blazespin in "DeepSeekMath-V2: Towards Self-Verifiable Mathematical Reasoning [pdf]"

blazespin — Thu, 27 Nov 2025 23:07:06 +0000

Verifying math requires something like Lean which is a huge bottleneck, as the paper explains.

Plus there isn't a lot of training data in lean.

Most gains come from training on stuff already out there, not really the RLVR part which just amps it up a bit.

New comment by blazespin in "DeepSeekMath-V2: Towards Self-Verifiable Mathematical Reasoning [pdf]"

blazespin — Thu, 27 Nov 2025 23:05:43 +0000

More training data on advanced math. Lean is cool, but it's mostly about formalizing stuff we already know.

New comment by blazespin in "Google Antigravity exfiltrates data via indirect prompt injection attack"

blazespin — Wed, 26 Nov 2025 01:19:35 +0000

You can't process untrustworthy data, period. There are so many things that can go wrong with that.

New comment by blazespin in "Correction: Anthropic attack did not have 1000/s requests"

blazespin — Sat, 15 Nov 2025 07:04:57 +0000

Given they made a geopolitical accusations and dozens of mainstream publications repeated the "thousands of requests per second", this seems like a grossly negligent flub that should not be dismissed as a mere typo.

I frequently, alone, do 1000s of requests over a period of time, especially ones that are mostly cache hits, which can be $10-$50 in API costs.

This was not a "large scale" attack by any means.