Of all of ai's crimes, the greatest is bringing down the em-dash, the king of punctuation. Commas, periods, even semicolons can never hold a candle to the flexibility of the em-dash. But we must fold in the eyes of inflexible, unloving philistines.
For me - I forgive an em-dash, but it does make me more skeptical of the other tells of AI-written text. If I start to see those other tells I do start to get a bit annoyed.
It's not the readme that gets me. It's the damn "chasing loc" KPI in the industry that continuously drives Claude to spout entire essays of docstrings and comments that make code unreadable that infuriates me!
If you're interested in HDMI audio, I have code for that lying around. The current best Rust source is in [1], but I also have a local branch with hand-tuned asm for the terc4 encoding. It's maybe 25% faster, which might be significant, as the per-scanline CPU load does go up a fair amount.
The decompilation of Emerald itself is human-authored. Then somebody vibeslopped a buggy WASM port, and now this is a vibeslopped port of the WASM port.
This. Project would look a lot more cooler if it was done by a human. It was vibecoded to hell and I can't find a note on the README that says it was vibecoded and/or assisted with LLM.
You could say this about 99% of articles posted here. One might even say VCs are in the business of wasting peoples' time.
But, my time is not wasted, and my cortisol managed not to spike at the presence of a tool. I suggest therapy or meditation or some way to mediate the stress of interacting with reality.
That must be the difference between me and the “skilled” prompt-brained “junior developers” (allegedly) that I work with that revert to a pre-historic ape brain if they don’t have copilot. They’re just on another plane of “skills” that’s so far beyond me my puny brain can’t comprehend putting some text in a box and running a loop.
I’ll try to remember that the next time I ask one of them to explain something the agent vomited out and they just start drooling and stare blankly into space. Must be some new language I’m too dumb to comprehend.
What is it about the code that suggests it is vibe-coded? Is it the commit messages? Is it the large comment blocks? I'm asking because I almost never read code written by SWEs.
Since this is clearly coded with an LLM and the README has not been written by a human, I'm curious if it actually does play end-to-end.
I have personally recently ported all of the mainline GBA Pokemon games to run on a modern PC. It took a lot of work, because there were many latent bugs left in by the developers which relied upon the quirks of the GBA hardware. Lots of null dereferences, things like that. These bugs were scattered throughout the game and so many features did not work until extensive testing was performed.
I also rewrote the engine to support 3D and wrote a bunch of modding and editing tools and quality-of-life upgrade. I have an editor that runs realtime in the games and lets me modify tiles and terrain. Releasing first alpha version this week to playtesters.
It likely will since it includes the game assets in the repo. Typically the way to avoid being taken down is you require people to extract them from a ROM as a prerequisite.
Copyright isn't an issue because these ports never contain the actual assets from the games. You usually have to provide your own ROM "ripped from your own copy of the cartridge" (wink wink nudge nudge)
Super cool!
I'm an AI-assisted coder myself...but one thing I can't stand is the way Claude writes. It's so frustrating and cringe. Tells from the README:
emdash, byte-exact, "The full engineering log", piggybacks, "none of this is possible without it", etc
A remedy I've found for this is in my Claude.md to tell it: write like a mix of simple english wikipedia, mr. rogers and ernest hemingway.
Of all of ai's crimes, the greatest is bringing down the em-dash, the king of punctuation. Commas, periods, even semicolons can never hold a candle to the flexibility of the em-dash. But we must fold in the eyes of inflexible, unloving philistines.
I think Em dash fans are quitters ⸻ there are much longer dashes in Unicode they can use.
I'm a big fan of semicolons; they don't get used enough.
For me - I forgive an em-dash, but it does make me more skeptical of the other tells of AI-written text. If I start to see those other tells I do start to get a bit annoyed.
I have a wild suggestion so bear with me: you could also write the readme prose yourself?
It's not the readme that gets me. It's the damn "chasing loc" KPI in the industry that continuously drives Claude to spout entire essays of docstrings and comments that make code unreadable that infuriates me!
If you're interested in HDMI audio, I have code for that lying around. The current best Rust source is in [1], but I also have a local branch with hand-tuned asm for the terc4 encoding. It's maybe 25% faster, which might be significant, as the per-scanline CPU load does go up a fair amount.
[1]: https://github.com/DusterTheFirst/pico-dvi-rs/blob/main/src/...
How much of the codebase was vibe coded?
The decompilation of Emerald itself is human-authored. Then somebody vibeslopped a buggy WASM port, and now this is a vibeslopped port of the WASM port.
At some point "slop" stopped being actually indicative of quality and I miss those days dearly
This. Project would look a lot more cooler if it was done by a human. It was vibecoded to hell and I can't find a note on the README that says it was vibecoded and/or assisted with LLM.
I feel like this question was not asked in good faith.
We should be supportive of creative endeavours and exploratory coding in the spirit of learning.
Burning fossil fuels to power the slop machine in times of climate catastrophe is neither creative nor in the spirit of learning.
What are you learning if you just let the AI do everything for you?
One'll learn what is possible, how things work, etc. you sound more like a bitter veteran than a curious child and this attitude is not helpful.
One can do that without posting it to a public GitHub repository, and then posting it to HN and wasting everyone else's time with it.
You could say this about 99% of articles posted here. One might even say VCs are in the business of wasting peoples' time.
But, my time is not wasted, and my cortisol managed not to spike at the presence of a tool. I suggest therapy or meditation or some way to mediate the stress of interacting with reality.
How to get things done with prompts, for one ; a skill that's looking increasingly useful as time goes on.
That must be the difference between me and the “skilled” prompt-brained “junior developers” (allegedly) that I work with that revert to a pre-historic ape brain if they don’t have copilot. They’re just on another plane of “skills” that’s so far beyond me my puny brain can’t comprehend putting some text in a box and running a loop.
I’ll try to remember that the next time I ask one of them to explain something the agent vomited out and they just start drooling and stare blankly into space. Must be some new language I’m too dumb to comprehend.
any vids? also please have at least some writing that isnt claude
What is it about the code that suggests it is vibe-coded? Is it the commit messages? Is it the large comment blocks? I'm asking because I almost never read code written by SWEs.
Since this is clearly coded with an LLM and the README has not been written by a human, I'm curious if it actually does play end-to-end.
I have personally recently ported all of the mainline GBA Pokemon games to run on a modern PC. It took a lot of work, because there were many latent bugs left in by the developers which relied upon the quirks of the GBA hardware. Lots of null dereferences, things like that. These bugs were scattered throughout the game and so many features did not work until extensive testing was performed.
I also rewrote the engine to support 3D and wrote a bunch of modding and editing tools and quality-of-life upgrade. I have an editor that runs realtime in the games and lets me modify tiles and terrain. Releasing first alpha version this week to playtesters.
Hasn't it received a C&D from Nintendo yet?
It likely will since it includes the game assets in the repo. Typically the way to avoid being taken down is you require people to extract them from a ROM as a prerequisite.
Recently it seems these old game port posts show up on the HN front page quite often. Very cool, but I'm not sure about the copyright issues.
Copyright isn't an issue because these ports never contain the actual assets from the games. You usually have to provide your own ROM "ripped from your own copy of the cartridge" (wink wink nudge nudge)
No, the pret decompilations it's based on do include assets.