Hacker News: argee

New comment by argee in "Why does Opus 5 feel worse to work with?"

argee — Sat, 15 Aug 2026 05:57:16 +0000

My bad, the English word I had in mind was "technology" (技術) but I misremembered due to feeling like "tech" should be its own token.

New comment by argee in "Why does Opus 5 feel worse to work with?"

argee — Fri, 14 Aug 2026 17:53:39 +0000

Read this.

https://openai.com/index/where-the-goblins-came-from/

> We retired the “Nerdy” personality in March after launching GPT‑5.4. In training, we removed the goblin-affine reward signal and filtered training data containing creature-words, making goblins less likely to over-appear or show up in inappropriate contexts. Unfortunately, GPT‑5.5 started training before we found the root cause of the goblins. When we began testing GPT‑5.5 in Codex, OpenAI employees immediately noticed the strange affinity for goblins, and we added a developer-prompt instruction (opens in a new window) to mitigate. Codex is, after all, quite nerdy.

Note that the permanent solution was not just adjusting the prompt, and in fact being perfectly aware of that option they decided on a different course of action. That means either you are wrong or they are wrong.

New comment by argee in "Why does Opus 5 feel worse to work with?"

argee — Fri, 14 Aug 2026 17:12:39 +0000

> Anyway, you might have more luck just writing to it in your native language.

This is potentially expensive advice (at least for many mainstream options). Where an English word like "literature" is one token, a couple of Chinese characters that spell a word can be 4 tokens. You'll pay more for input/output and get less of a context window (per word) too.

New comment by argee in "That time when I failed the Microsoft interview"

argee — Tue, 04 Aug 2026 08:21:40 +0000

Back when I interviewed at MS as a college senior, I had no clue about these "formats". So when someone asked me how I would make (I think they meant design) a web crawler, I started writing one in Python. They didn’t stop me, and it seemed to be awkward to them that I was doing this, whereas I felt awkward that they were acting that way when I was doing what they asked. It was just stupid all around. These systems deserve to be gamed, and the interviewers can't handle anything else, like in TFA.

New comment by argee in "Why Book Corners won't sync contributions back to OpenStreetMap"

argee — Mon, 03 Aug 2026 03:15:44 +0000

I feel like having such a map at all is kind of antithetical to building community. Shows you don't care enough to take a stroll in your 'hood to find such a one by happenstance (or to find neighbors to populate yours) and are hunting them down to share your self-published slop or something. Besides, it makes things a hell of a lot easier for griefers and trolls.

New comment by argee in "Dario Amodei's stance on open weights is self-serving and short-sighted"

argee — Fri, 31 Jul 2026 22:30:28 +0000

The only widely publicized AI orchestrated cyberattack in recent memory came from a closed-weight model, and the only widely publicized defense from said attack came from an open-weight Chinese model. What do you make of that?

> I don’t think that an AI monopoly or duopoly should be allowed to exist either

Monopolies have always been allowed to exist, even welcomed. People have gotten confused, but there is a big difference between a monopoly and an illegal monopoly. Deciding what should and shouldn’t be allowed to exist as determined by regulatory bodies and not consumers goes against democratic principles (though it is sometimes necessary, such as with drugs).

I don’t think I want to live in a world where police decides what brand of chips I can eat on the sole basis of the chips having gotten too popular. Do you?

New comment by argee in "Bad Apple but It's Traceroute"

argee — Fri, 31 Jul 2026 01:03:58 +0000

Yes, anything that was statically lit was fine. The game had absurdly low graphics requirements for modern standards, but even so I didn't have a GPU, just integrated graphics. So hacking the game to not render dynamically lit items was enough to make it playable (also needed some tweaks that would let it even launch in the first place). I obviously can't say exactly what I did though, this was a very, very long time ago (24 years).

New comment by argee in "Bad Apple but It's Traceroute"

argee — Thu, 30 Jul 2026 21:43:37 +0000

I had similar situations in my youth, I have fond memories of hacking games up so they'd work without the 3D models. You haven't lived until you've played Mafia only being able to see the player's shadow beating up the target's shadow with a bat.

New comment by argee in "Man seeks millions after being shot by police in game-related swatting incident"

argee — Thu, 30 Jul 2026 21:25:30 +0000

More sensibly, their insurance could just toss a clause in and charge higher premiums to cover for this, while adding some conditions to help prevent it.

New comment by argee in "Some thoughts about Anthropic's new cryptanalysis results"

argee — Wed, 29 Jul 2026 20:23:04 +0000

Sorry, this discussion is too incommensurable for me.

New comment by argee in "Some thoughts about Anthropic's new cryptanalysis results"

argee — Wed, 29 Jul 2026 20:00:52 +0000

Depends. "Most" implies majority, and the majority of people are using these tools not for programming but in contexts where ontology is more relevant than capability (not that capability is irrelevant, but most people care, or are tricked into caring, far more about the former).

New comment by argee in "Some thoughts about Anthropic's new cryptanalysis results"

argee — Wed, 29 Jul 2026 19:31:28 +0000

Exactly. They ARE "glorified autocomplete" in an ontological sense. That says nothing about capability or outcome. The people who come out swinging against that characterization usually ignore the whole ontological argument (which is...the entire point) and go after an outcome-based strawman.

New comment by argee in ""Uncensored" open LLMs are measurably more optimistic than their base models"

argee — Wed, 29 Jul 2026 16:25:07 +0000

Good thing we have you here to be a stellar role model based on how you responded.

New comment by argee in "Starling: The first real desktop written by AI"

argee — Wed, 29 Jul 2026 15:59:48 +0000

Great products are made through interaction with humans. Products like this one have barely had interaction with their creator, let alone early customers, and get foisted upon us with the expectation that we will happily download and execute god-knows-what made by god-knows-who, practically guaranteed to have an awful and thoughtless experience.

Then there’s the HNers who love that, nevermind the rules or intellectual curiosity, because they hope to likewise foist slop upon others and their agents and subagents are churning away even as they post their chatGPT "cleaned" comments about how awesome these products are.

These products need to go through a lot more before they come to HN. It’s not even AI-generation that’s the problem. Show at least a modicum of care, or earn trust by telling us who you are and why you made this (no, I don’t mean putting that in a Claude prompt and posting the result).

New comment by argee in ""Uncensored" open LLMs are measurably more optimistic than their base models"

argee — Wed, 29 Jul 2026 04:20:31 +0000

Don't be on HN then?

https://news.ycombinator.com/newsguidelines.html#generated

> Don't post generated text or AI-edited text. HN is for conversation between humans.

New comment by argee in ""Uncensored" open LLMs are measurably more optimistic than their base models"

argee — Wed, 29 Jul 2026 03:19:10 +0000

There's a difference between getting a tool to translate what you've written (a la Google translate), and getting an LLM to generate content. If you don't see that difference, we'll have to agree to disagree.

New comment by argee in ""Uncensored" open LLMs are measurably more optimistic than their base models"

argee — Tue, 28 Jul 2026 19:57:10 +0000

What would you have done pre-LLM? Do that. It's better.

Because "English isn't my first language so i used AI to clean up the writing" translates to "I don't respect an English speaking audience." And if you know enough English to disagree with that translation, it just proves the point further.

New comment by argee in "Why I'm building a note taking app without AI"

argee — Sun, 26 Jul 2026 15:56:58 +0000

Hate beating a dead horse but I just found this again. [0]

Ctrl/Cmd + F for "worst mistake".

[0] https://news.ycombinator.com/yli.html

New comment by argee in "How to write a quine"

argee — Sat, 25 Jul 2026 15:23:53 +0000

How foolish, explaining succinctly. I don’t see a Substack writing career in your future.

New comment by argee in "Claude Opus 5"

argee — Fri, 24 Jul 2026 17:16:38 +0000

Hackers and people of color.