Hacker News: daishi55

New comment by daishi55 in "Israel creates fake think tank in likely attempt to dupe AI chatbots"

daishi55 — Tue, 18 Aug 2026 14:00:37 +0000

Absolutely! If Pakistan started committing a genocide and nobody talked about it but continued talking about Israel’s genocide, I would start to ask some questions. That is not the case though, no country on the planet is currently doing anything nearly as awful as Israel. So the coverage is well-calibrated.

New comment by daishi55 in "Anthropic's ‘watermark’ text adulteration in Claude is a perversion of writing"

daishi55 — Tue, 18 Aug 2026 02:50:42 +0000

Your original statement “LLMs use randomness in a fundamental way” is incorrect. LLMs have these knobs and randomness is not an inherent property of LLMs.

New comment by daishi55 in "Israel creates fake think tank in likely attempt to dupe AI chatbots"

daishi55 — Tue, 18 Aug 2026 02:48:00 +0000

It’s not “the Jews” it’s Israel, and I suspect the greater attention is due to the greater crimes being perpetrated by the Israeli government compared to those other states.

New comment by daishi55 in "Israel creates fake think tank in likely attempt to dupe AI chatbots"

daishi55 — Tue, 18 Aug 2026 02:45:10 +0000

I think the difference is probably related to the ends being pursued by each of those states at the moment. Only one is engaged in what the UN, HRW, and most legal experts call a genocide. So influence operations in support of an ongoing genocide are just more noteworthy than your run-of-the-mill authoritarianism/corruption.

New comment by daishi55 in "Anthropic's ‘watermark’ text adulteration in Claude is a perversion of writing"

daishi55 — Mon, 17 Aug 2026 12:47:22 +0000

Claude has those knobs, they are just not exposed to the user. They could make Claude nearly completely deterministic if they wanted to (of course it would be a far inferior product then. But they could).

New comment by daishi55 in "Cloudflare's AI Psychosis"

daishi55 — Sat, 15 Aug 2026 22:34:00 +0000

Every cloudflare product I've ever used has been a legit delight to use. I use it for lots of stuff on my homelab. Everything works well and is simply pleasant. Maybe if you use it heavily for enterprise that changes, but it just hasn't been my experience.

New comment by daishi55 in "Cloudflare's AI Psychosis"

daishi55 — Sat, 15 Aug 2026 22:29:35 +0000

I think that as soon as you start viewing certain groups/categories/etc as bad and blaming them for everything, as many engineers do to the non-engineering staff at tech companies, you have stopped thinking logically and are just reacting emotionally.

"PMs bad, engineers good" is such a common idea in these spaces and it really makes me think less of the thinking skills of anyone who espouses it.

New comment by daishi55 in "RISC-V: They Should Have Known Better"

daishi55 — Sat, 15 Aug 2026 19:15:19 +0000

We are using RISC-V for AI accelerators to great success

https://ai.meta.com/blog/meta-mtia-scale-ai-chips-for-billio...

RISC-V was a great choice due to being so customizable and extensible.

New comment by daishi55 in "Cloudflare's AI Psychosis"

daishi55 — Sat, 15 Aug 2026 15:08:12 +0000

Not directing this at you personally. But I've heard a lot of mediocre engineers whine and complain about PMs, MBAs, C-suite, etc. The best engineers I don't hear complaining much at all. Just a thought.

And sorry but I don't agree that launching AI products in 2026 is not related to the market. I think that's the whole basis for the decision.

New comment by daishi55 in "Cloudflare's AI Psychosis"

daishi55 — Sat, 15 Aug 2026 15:07:05 +0000

What values and principles are supposed to be preventing cloudflare, the internet and technology company, from exploring hot new internet markets and technologies?

New comment by daishi55 in "As AI eats the web, the internet’s collective memory is disappearing"

daishi55 — Sat, 15 Aug 2026 14:58:47 +0000

No actually it made a few specific historical analogies that I’ve never heard before and that absolutely could be responded to. The responder just chose not to.

New comment by daishi55 in "Cloudflare's AI Psychosis"

daishi55 — Sat, 15 Aug 2026 14:55:38 +0000

I don’t understand the impulse to whine and complain about the aesthetics of a big company.

> a little more cringey and clique

Why on earth do I care if someone thinks Cloudflare is cringe? What is interesting about that? They are following the market like everyone else.

> Too many ways to do the same thing, none of them great…They got D1 (SQLite serverless), Durable Objects with their own SQLite, KV, R2, Queues, and Hyperdrive to speed up external Postgres or MySQL.

What does this have to do with AI psychosis? I thought that was the thesis of this article? As I said, it’s just directionless complaining.

New comment by daishi55 in "As AI eats the web, the internet’s collective memory is disappearing"

daishi55 — Tue, 11 Aug 2026 14:51:48 +0000

Nothing they said demonstrated any understanding of what they replied to. They simply restated their opinion with no reference to any of the points made in the comment they were ostensibly replying to.

New comment by daishi55 in "As AI eats the web, the internet’s collective memory is disappearing"

daishi55 — Tue, 11 Aug 2026 12:42:53 +0000

You don’t seem to have read the comment you are replying to.

New comment by daishi55 in "How I use LLMs to learn complex topics"

daishi55 — Mon, 10 Aug 2026 16:38:02 +0000

This is really a theoretical problem and not a practical one though. Practically, for everything I use these tools for today (complex low-level systems programming), hallucinations don’t even exist anymore.

When they don’t know something, they figure it out empirically. For things they already know, they are consistently correct.

New comment by daishi55 in "How I use LLMs to learn complex topics"

daishi55 — Mon, 10 Aug 2026 16:34:29 +0000

I mean they have been solved practically I think. I use these things all day every day and haven’t seen a hallucination in a long time.

New comment by daishi55 in "New Amazon Data Center Is Set to Have the Most Polluting Power Plant in the U.S."

daishi55 — Sun, 09 Aug 2026 23:50:50 +0000

> I'm just linking charts and figures

It’s adorable that you think charts and figures can’t be biased or malicious.

> handwaving about future robot armies

Yeah and you would’ve said the same thing about China’s EVs, solar panels, or any number of other areas in which the Chinese state demonstrated both accurate foresight and ability to execute on long-term strategic economic/technological plans. And you would’ve been wrong. As you are now.

New comment by daishi55 in "How I use LLMs to learn complex topics"

daishi55 — Sun, 09 Aug 2026 23:47:13 +0000

Are there any reproducible hallucinations on any of the currently available OAI/Anthropic models? I’m not aware of any.

And even if they are related - if Opus 4.8 always has a 1:100 chance of a specific hallucination - then running the same model twice does indeed dramatically reduce the odds of an error in the final output.

New comment by daishi55 in "How I use LLMs to learn complex topics"

daishi55 — Sun, 09 Aug 2026 22:53:49 +0000

Not really. Take hallucinations for example. If they are 1 in 100 (actually they are much rarer, but for the sake of argument), then the chances that 2 LLMs or even just 2 runs of the same LLM have the same hallucination is, well, a lot less than 1 in 100.

New comment by daishi55 in "2027 memory capacity is reportedly sold out"

daishi55 — Sat, 08 Aug 2026 13:29:04 +0000

> In the real world, a supplier can contract to supply more units than they can actually produce.

Did this actually happen though? Is there any evidence or you are just imagining a scenario which if it did happen could lead to the scenario you outlined?