Hackers' Pub is a place for software engineers to share their knowledge and experience with each other. It's also an ActivityPub-enabled social network, so you can follow your favorite hackers in the fediverse and get their latest posts in your feed.
졸려서 이만 자러 갑니다 안녕히 주무시고 좋은 꿈 꾸세요 고양이 사진 투척
Philosophical/computing question: How much (CPU) power is needed to satisfy a human vs. an AI running on that CPU? And is it inversely proportional to the intelligence of the human?
Palantir Gets Millions of Dollars from New York City's Public Hospitals
Link: https://theintercept.com/2026/02/15/palantir-contract-new-york-city-health-hospitals/
Discussion: https://news.ycombinator.com/item?id=47025624
RynnBrain
Link: https://github.com/alibaba-damo-academy/RynnBrain
Discussion: https://news.ycombinator.com/item?id=46968964
EDIT: alt text added
Reintroduction:
Yeah. I'm a board certified practicing psychiatrist.
I enjoy overturning plutocracies, iconoclasm, egalitarian aims, and surviving my enemies.
Yeah. I had a larger account but needed a reboot.
No, I dont care about football, pro wrestling, or hockey in a material way, but they are nurturing distractions that prevent depression.
I think RFK is an ignorant loser but that the US Healthcare "system" is also a joke re: meeting needs of patients.
I'd rather read Malcolm X & Angela Davis than anything by anyone of influence in modern mainstream circles.
I *do* have ADHD.
I *do not* have a family as I disowned their Nazi asses.
Adieu.
게임 클라이언트용 SDK 만드느라 SDK 테스트에 사용할 견본 게임을 만드는데, Codex(gpt-5.3 high)든 Claude Code(opus 4.6)든 일 잘 못하긴 마찬가지구나.
특히 이미 2D 애셋이 있어서 애셋에 맞춰 구현하는 경우에, "눈"과 "미감"이 없다는 점이 치명적으로 약점으로 작용한다. 드로우콜을 줄이려는 최적화 기법, 가령 스프라이트 아틀라스는 AI 코딩 에이전트에게 악몽이다. 어느 정도로 악몽이냐면,
Claude Code는 포기했는지 집요하게 애셋을 쓰라고 지시하는데도 계속 회피하고 거짓말로 외면하며 애셋을 안 쓰고 지 멋대로 벡터 이미지를 그린다. 기존 애셋을 분석해서 네가 작업하기 좋게 재구성하라고 했더니 눈가리고 아웅식으로 구성하는 척하고선 실제로는 사용을 안 했다. 거짓말하는 솜씨가 가히 Gemini급이다.
Codex는 쬐.끔. 나은데, 기존 아틀라스를 분석하여 꽤 유의미하게 애셋을 구성했다. 그 와중에 많은 애니메이션이 유실되거나 정체를 파악 못한 애셋은 무시하거나 사용하지도 않으면서 복사하긴 했지만.
캐릭터 관절로 스프라이프가 구성된 경우, 부위별 애니메이션은 포기하는 게 낫다. 이 작업에서 Codex는 재밌는 모습을 보였는데, 각 부위 위치를 잡고 충돌 처리를 파악하기 위해 부위에 박스를 그렸다는 점이다. 다시말해 숫자와 공식으로 처리하는 게 아니라 사람이 눈으로 보며 캐릭터 부위를 파악해 위치를 조정하는 것처럼 작업했다는 것이다. 비유가 아니라 실제로 그렇게 출력하고 화면 캡쳐하며 아틀라스에 있는 메타데이터의 숫자 의미를 파악하려고 했다. 흥미로웠다. 흥미롭긴 했는데 결과물은 역시 별로였다.
애셋이 대부분 상대 크기에 가깝다보니 실제 화면에 그려지는 오브젝트 크기를 AI 코딩 에이전트는 감을 잡지 못하는데, 사람이 편의상 스프라이트에 여백을 준 경우 이미지가 계속 엉뚱한 위치에 그려졌고, AI 코딩 에이전트는 원인도 모른 채 계속 크기와 위치를 산술 계산으로 잡으려 애썼다. 그러니까 오른쪽으로 조금 옮겨봐, 라고 하면 50px을 더하는 게 아니라 오른쪽으로 옮길 오브젝트의 실제 크기와 비율 조정한 크기 계산을 하고, 현재 화면의 크기와 대상 오브젝트 간 거리를 계산해서 위치를 조정하는 건데, 스프라이트 자체에 여백이 들어가있어서 AI 에이전트의 계산은 애초에 맞을 수가 없는 것이다.
Claude Code는 진작 포기해서 거짓말로 일관했고, Codex는 오브젝트 위치 조정하는 데 20분씩 써가며 고군분투하더라.
시차 스크롤링 (Parallax scrolling)도 AI 코딩 에이전트가 애셋만으로는 감조차 잡지 못해서 몹시 헤맸는데, 통으로 된 배경 스프라이트 몇 장이면 간단히 구현한다. 문제는 풍성한 미감을 주면서도 최적화를 위해 배경 오브젝트 요소가 다양하게 구성된 경우인데, 무엇이 산이고, 무엇이 바위이며, 무엇이 화면 맨 앞에(카메라쪽) 배치되는 나무인지 모르니 엉망진창으로 배치한다. 결국 Claude Code에겐 시차 스크롤링을 최소화하게 지시해서 캐릭터가 배경에 파묻히는 문제를 해결했고, Codex는 2시간 걸리고 애셋을 과도하게 늘리거나 해서 지저분해지긴 했지만 결국 애셋 의도에 거의 비슷하게 시차 스크롤링을 구성했다.
그나마 HTML 5 기반 웹 2D 게임인 경우 Playwright로 캡쳐하며 이것 저것 시도하는데, 유니티로 구현하니 카메라 위치 잡는 데에도 한참 걸렸다.
출근하면 회사 내 게임 클라이언트 개발자 옆에 앉아서 AI 코딩 에이전트를 어떻게 활용하는지 관찰 좀 해봐야겠다.
Ars Technica retracted an article about how AI is making the world worse because...
the Ars article itself contained AI-generated quotes in it.
Welp, we had a decent run, folks. But it's time to call it.
#Roblox, #Reddit and #Discord users must use biometric #ageverification tool backed by Palantir co-founder Peter Thiel
@JamesBaker says:
“People are worried, and angry, about having to hand over biometric scans of their face to overseas companies just to access online platforms. This includes children.
“Users deserve to know who is behind these systems, how their data is being used, and to have a real choice about who verifies their identity online.”
the CSS inspector
Permalink: https://wizardzines.com/comics/css-inspector
Our third release candidate for upcoming #GIMP 3.2 will feature a new splash image, made by astronomer 
@markmccaughrean (as were the first 2 RCs' splashes),
This #JWST image shows "a protostellar outflow complex called HH288, which we nicknamed The Dragon Jet some thirty years ago when we first discovered it using ground-based telescopes".
Enjoy!
🚨 Last call for all Pythonistas! 🚨 Europython 2026 CFP is closing soon. Don't miss your chance to be part of the event! Submit your proposals now! 🐍
https://programme.europython.eu/
A practical guide to observing the night sky for real skies and real equipment
Link: https://stargazingbuddy.com/
Discussion: https://news.ycombinator.com/item?id=46990888
这真的太可爱了——
I am using #deltachat 2.35.0. I have a problem with push notifications. They are not arriving anymore. #deltachat android is installed from the play store. The messages arrive when I open the app. Any thoughts?
Countering fediverse cringe on Bluesky
Feel free to join in!
https://bsky.app/profile/did:plc:lhck2qa73dovvfqdy7k3ffj6/post/3mew5n664bc2x
Surprising connections. #grickledoodle #fairytales #gingerbreadman #fox #friends #cartoon #art #drawing #funny #humor
"If the problem is that we’ve painted our development environments into a corner that requires tons of boilerplate, then that is the problem. We should have been chopping the cruft away and replacing it with deterministic abstractions like we’ve always done. That’s what that Larry Wall quote about good programmers being lazy was about. It did not mean that we would be okay with pulling a damn slot machine lever a couple times to generate the boilerplate."
> But bringing on a CEO from a corporate journalism background into public media carried risk.
I think this is a major risk that we often fail to catch. Say it in media, governments, schools, companies, anywhere. Even if one really seems to resonate with the org value and mission, they may only know corporate ways to actually pursue that. That can really degrade things.
https://www.thewesternedge.media/p/i-had-a-dream-journalism-job-heres
is dance dance revolution sports or esports
"Inventing the Future" by Nick #Srnicek (2015) is a political and philosophical journey leading to 3 mutually-reinforcing demands:
1. Full automation
2. Universal basic income
3. Reduced work-time
Automation is great when its rewards are shared: the Luddites protested not the machines, but the concentration of capital.
It seems there's not been a better occasion to demand full automation - many are doing that already.
Now is the time to agitate for #UBI and reduced work-time (again.)
chatgpt가 나를 Precision-driven aesthete, 라고 평했는데, 음... 마음에 드는 결과 내놓을 때까지 이걸로는 부족하다며 너무? 갈궜나. 😅😇
Prince (Bristol, UK) likes to get in the shower first, to make sure it is safe in there. He also seems to enjoy waiting for it to be turned on so he can race the water. Normally he can make it out of the shower faster than gravity and pressure can move the water droplets.
I love a lot of things about #BSD but high on the list is the community itself, it is a wonderful bunch of misfits, which is probably why i fit.
Palantir vs. the "Republik": US analytics firm takes magazine to court
Link: https://www.heise.de/en/news/Palantir-vs-the-Republik-US-analytics-firm-takes-magazine-to-court-11176508.html
Discussion: https://news.ycombinator.com/item?id=47025188
the birth of the internet?
I've been honestly adding Co-Authored-By: Claude <noreply@anthropic.com> to every commit where I used an LLM even slightly—whether it's generating test scaffolding, drafting docs, or just bouncing ideas. I thought transparency was the right thing to do. Turns out, people see that trailer and immediately assume the whole thing is “vibe coded” AI slop, no further questions asked. The irony is that being honest about my process is what's getting my work dismissed.
Now I'm genuinely torn. Do I keep the trailer and accept that some people will write off my work at a glance? Or do I drop it and lose something I actually believe in? It's frustrating that there's no widely understood distinction between “I prompted an LLM to write my entire app” and “I used an LLM as a tool while writing my own code.” I don't have an answer yet—just sitting with the discomfort for now.
Adding to this: much of my LLM usage actually goes into writing documentation and communicating in English—a language that isn't mine but that open source essentially demands. For non-native English speakers, LLMs are a genuine equalizer. They let me write docs that don't get ignored for sounding awkward, and participate in discussions without spending twice the effort a native speaker would. But when English-native developers dismiss LLM-assisted writing wholesale, they're not even aware of the privilege baked into that judgment. It's like someone who's never had to do housework scoffing at a washing machine for making people “lazy”—easy to say when the burden was never yours to begin with.
Hereby resolving to stop fucking nerd-sniping myself while making these important security-related changes to Paramiko.
It's one thing to "make thing better while you're touching it" & opportunistically edit code I wrote most of myself, like Fabric and Invoke.
It's another entirely for a ~25-year-old codebase (ie: many historical layers, API inconsistencies across eras, etc), the majority of whose LOC are (still!) not written by me or my core contributors (ie: style/layout differences constantly causing an eye-twitch).
In light of all the discourse, I found this a remarkably clear statement.
> Our efforts are prioritised on our target markets which currently does not include being a Discord replacement
>
> https://github.com/element-hq/element-web/issues/5993#issuecomment-3884340391
Yeah, yeah, I know... Y'all think you need AI to be better engineers. But have you tried estrogen and longer socks?
みんなは
LT6502: A 6502-based homebrew laptop
Link: https://github.com/TechPaula/LT6502
Discussion: https://news.ycombinator.com/item?id=47025399
DjVu and its connection to Deep Learning (2023)
Link: https://scottlocklin.wordpress.com/2023/05/31/djvu-and-its-connection-to-deep-learning/
Discussion: https://news.ycombinator.com/item?id=47022213
When you pledge to our KUMARI LOVES A MONSTER Kickstarter reprint drive, you can donate a copy to THE FREE LIBRARIES NETWORK, a fantastic initiative serving communities across India.
Kids need third spaces... and books about wholesome consensual monsterlove!
https://www.kickstarter.com/projects/blaft/kumari-loves-a-monster
If you have a CD-ROM drive in your machine, you can make the CD-ROM that is
presently inserted available by typing 'mount /cdrom' as root. The CD-ROM
will be available under /cdrom/. Remember to do 'umount /cdrom' before
removing the CD-ROM (it will usually not be possible to remove the CD-ROM
without doing this.)
Note: This tip may not work in all configurations.
Air to Ground Message:
MIRROR MIRROR ON THE WALL... WILL LGA BE THE BIGGEST SHTSHOW OF THEM ALL WE FLY ALL DAY TODAY LOL YALL EXPECTING IROPS MG
Area: Philadelphia, PA, USA
Type: Embraer 175 (Enhanced Wing)
A: #a3f91bb3363
F: #f96ab740ee0
If you have a CD-ROM drive in your machine, you can make the CD-ROM that is
presently inserted available by typing 'mount /cdrom' as root. The CD-ROM
will be available under /cdrom/. Remember to do 'umount /cdrom' before
removing the CD-ROM (it will usually not be possible to remove the CD-ROM
without doing this.)
Note: This tip may not work in all configurations.
🕐 2026-02-15 18:00 UTC
📰 技術発信を後回しにした私が、転職で詰んだ話。 (👍 255)
🇬🇧 Developer shares hard lesson: delaying tech blogging meant lacking proof of skills during job hunt. Advocates documenting learning publicly.
🇰🇷 기술 블로깅을 미루다 이직에서 실력 증명 자료 부족으로 고생한 개발자의 교훈. 불완전해도 공개하고 기록하라.
우익 우경화 그거 별거 아니다.
미 연방대법원 트럼프 관세 사건은 판결이 7월까지도 늦춰질 수 있다고. 친트럼프 대법관들이 계속 논점을 늘리고 있는 듯 한데 그런다고 결과가 바뀔지는 모르겠다.
From Flock cameras to fusion centers, “free” surveillance equipment funded by tech vendors, federal grants, and private donors locks cities into recurring costs and long-term data‑sharing, including with ICE. The end result is an erosion of our civil liberties. https://www.eff.org/deeplinks/2026/02/free-surveillance-tech-still-comes-high-and-dangerous-cost
P.Macinkovi vadí, že EU má demokratický deficit a nemůžeme si volit členy Evropské komise. Možná si nevšiml, že v ČR si také nevolíme členy vlády. Členem české vlády se může stát nepolitik, člověk co ve volbách propadl či ani nekandidoval, a dokonce Slovák.
For custom domains, #Holos currently works with CNAME, which redirects your entire domain to the relay. That's fine for a subdomain, but if you want your identity on your root domain, you probably already use it for something else.
We will add support for that. The relay will handle everything on a subdomain you dedicate to Holos, and automatically generate a webfinger file to host on your root domain. Your identity will be @you@yourdomain.com while the setup runs on a subdomain.
소니가 작곡 AI의 학습 (원곡) 데이터를 특정할 수 있는 기술을 개발했습니다. 따라서 원곡의 권리 보유자가 AI 서비스측에 설명, 대가를 요구할 수 있게 되었습니다. www.nikkei.com/article/DGXZ... (유료 기사)
ソニーグループ、作曲AIの学習データを特定 創作者への対価算...
Hideki Sato, designer of all Sega's consoles, has died
Link: https://www.videogameschronicle.com/news/hideki-sato-designer-of-segas-consoles-dies-age-75/
Discussion: https://news.ycombinator.com/item?id=47024907
Paolo Melchiorre
Ael Yokan
jnkrtech
wakest ⁂
Xe 
Misty
Stefano Marinelli
