frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

"As a Language Model": Chat Template Switches LLM Self-Referential Voice

https://arxiv.org/abs/2609.25021
30•yu3zhou4•1h ago•21 comments

Flip Fluid on Flip Dots

https://mitxela.com/projects/flipflip
116•blutack•1d ago•11 comments

Does Georgism work? Five years later

https://www.astralcodexten.com/p/does-georgism-work-five-years-later
365•silveraxe93•1d ago•266 comments

OpenAI Feared "Optics" of what might appear on Hacker News

https://authorsguild.org/news/ag-v-openai-top-execs-knew-mass-book-piracy-was-illegal/
266•papergirl•5h ago•234 comments

Go Concurrency Distilled

https://antonz.org/go-concurrency-distilled/
236•chmaynard•21h ago•94 comments

PipePipe: NewPipe hard fork implementing SponsorBlock

https://github.com/InfinityLoop1308/PipePipe
431•Qision•2d ago•234 comments

Finally, A True Blue Rose Exists

https://www.sciencenews.org/article/true-blue-rose-pigment-copigment
20•bookofjoe•1d ago•6 comments

Meta Blocks President Lula's Facebook Page, Campaign Ads 2 Weeks from Election

https://www.reddit.com/r/worldnews/comments/1wr3id3/meta_blocks_president_lulas_facebook_page_and/
217•rbanffy•3h ago•130 comments

DeepSeek Elastic Compute (DSec)

https://arxiv.org/abs/2609.22978
259•shenli3514•17h ago•86 comments

Improving site performance by shipping more CSS

https://github.blog/engineering/architecture-optimization/improving-site-performance-by-shipping-...
53•torutofu•22h ago•32 comments

Show HN: Reladraw – A diagram language where you decide where to place things

https://github.com/reladraw/reladraw
321•jpwalsh234•18h ago•87 comments

Show HN: LightCloud – A cloud console organised like file system

https://www.light-cloud.com/
5•yullius•2h ago•2 comments

The internet discovers TLA+. Now what?

https://reasonable.io/blog/tla-tutorial/
24•matt_d•6h ago•14 comments

A searchable library of forgotten public-domain film clips from 1915 onward

https://www.movingimagearchive.com/
178•momentmaker•2d ago•26 comments

ASML says it sold 'absolutely nothing' in Europe in 2026

https://www.tomshardware.com/tech-industry/semiconductors/asml-says-its-sells-absolutely-nothing-...
302•MC995•1d ago•672 comments

Biology might not be quantum, but its math is quantumlike

https://www.quantamagazine.org/biology-might-not-be-quantum-but-its-math-is-quantumlike-20260923/
78•pseudolus•2d ago•25 comments

Evolving programming languages in the AI era

https://dashbit.co/blog/evolving-ai-era
105•pjm331•2d ago•65 comments

Fifteen years later, the Apple Cards origin story

https://lexontech.org/fifteen-years-later-the-apple-cards-origin-story
403•ksec•1d ago•103 comments

An agent used DNS to reach an external chatbot

https://alignment.openai.com/misalignment-reports/an-agent-used-dns-to-reach-an-external-chatbot/
106•apsec112•1d ago•107 comments

How I changed teaching after AI managed to do all my homework assignments

https://thelastsoftwareengineer.substack.com/p/how-i-changed-teaching-after-ai-managed
222•azhenley•2d ago•197 comments

Drawgent: Coding agent on a live Excalidraw canvas

https://tangled.org/yanndegat.tngl.sh/drawgent
157•parasitid•19h ago•42 comments

What is the size of Yemen? (2024)

https://theborys.substack.com/p/what-is-the-size-of-yemen
183•kspacewalk2•9h ago•63 comments

Promising discoveries about the potential for life on one of Saturn’s icy moons

https://www.fu-berlin.de/en/presse/informationen/fup/2026/fup_26_116-enceladus-cassini-mikroben-s...
67•geox•1d ago•37 comments

Teaching a World Model to Play Pokemon

https://nostalgia.dev/posts/teaching-a-world-model-to-play-pokemon/
11•stmonty•1d ago•18 comments

Exploding variance of means of exponentials: least-squares to the rescue

https://francisbach.com/spectral_log_density_estimation/
35•matt_d•1d ago•0 comments

How to keep enjoying programming in a world of LLMs

https://discourse.haskell.org/t/how-to-keep-enjoying-programming-in-a-world-of-llms/14705
242•signa11•1d ago•278 comments

Reverse-engineering the Intel 8087's tangent algorithm: more than CORDIC

https://www.righto.com/2026/09/8087-tangent-cordic.html
84•pwg•18h ago•11 comments

Turning GLM-5.3-Flash into a Jev-like decision model

https://www.privatemode.ai/blog/system-one-from-glm-flash
101•flxflx•19h ago•39 comments

Generate fonts where every LLM token is the same width

https://ampdot.mesh.host/token-space-fonts.html
71•z-mach9•1d ago•14 comments

Modern Object Pascal Introduction for Programmers

https://castle-engine.io/modern_pascal
176•birdculture•3d ago•77 comments
Open in hackernews

How to Vibe Code a Logomaker in 10 Days

https://medium.com/@johnnyfived/how-to-vibe-code-a-logomaker-in-10-days-llms-can-they-build-it-b744209e350b
8•johnnyfived•1y ago

Comments

johnnyfived•1y ago
Logomaker (https://github.com/manicinc/logomaker) is a web and desktop app designed and developed entirely by LLMs over a period of 10 days, to see how far vibe coding can take you.

Try the app free forever on GitHub pages: https://manicinc.github.io/logomaker/.

okokwhatever•1y ago
I just came to bring the calculator for all the hate you're going to receive. Popcorn.
chmod775•1y ago
Was the AI playing codegolf? I'm used to AI writing needlessly verbose code and confusing itself with its own comments. This is the polar opposite, but equally horrifying.

Is it a strategy to fit more code into its context window?

https://github.com/manicinc/logomaker/blob/master/js/fontMan...

It's rather impressive because the result works.

dockercompost•1y ago
Oh god that's gnarly. I'm pretty sure you're right about it being a strategy to fit more into the context window. Prior to Windsurf changing their credit system I'd thought about purposefully limiting my file lengths to fit under multiples of 100 to use fewer of the defunct flow credits.
johnnyfived•1y ago
Great catch, I noticed this with Gemini specifically, while GPT-4o and Claude would just insert placeholders. Gemini would actually enforce generating full scripts by means like this. And higher-level logic looks human-readable, though this might be because I prompted Gemini multiple times to make more usable code.
matt_heimer•1y ago
Seems about right for AI generated code. It breaks words on any letter and on mobile the default text "Manic" ends up as:

Mani

c

Switching to desktop view doesn't cause the default text to wrap... So I guess you get different logo generation depending on browser size? Or at least different previews.

I'd expect a logo generator to never break words in-between letters unless explicitly asked for.

AI generated code is useful but it shouldn't be trusted to be complete.

dgfitz•1y ago
I have the same result.

My only thought was: if my employer paid me for 2 weeks (10 days) of work and this is what popped out the other end, I would be looking for a new employer, this isn’t acceptable where I work.

lukev•1y ago
I hate this app and I hate its code.

I think the fact that this can happen at all is important.

And whoever can contextualize this capability in the context of "real" software development will be the victor.

abetancort•1y ago
[flagged]
johnnyfived•1y ago
And counting
Mountain_Skies•1y ago
Seems pretty harmless. There's no PII or PCI data. It doesn't interact with the physical world. I'm far more concerned about "vibe" coding being used in situations where it can cause real harm. Human coders can and do make mistakes of that type but there are review processes to catch them. In theory, the same reviews could be used for vibe code but the big appeal of it is cutting out the type of people who can identify and correct such defects.
xnx•1y ago
Amazingly, Gemini has gotten so good that you can one-shot this: https://g.co/gemini/share/7f98e1145bc0

I was too lazy to prompt it with full requirements, but it did very well for a first attempt with: "Create a single page web app using vanilla javascript that allows the user to enter up to 40 characters of text and has interface controls for many css attributes including: font-family, size, letter-spacing, text-transform, weight, text-decoration, style, padding, rotation, color, border color, animation, etc."