Hacker Newsnew | past | comments | ask | show | jobs | submit | dag11's commentslogin

It's really interesting despite others pointing out that it's a little wrong, that can be fixed.

But I also almost left immediately because this claudism at the top really hurt to read:

> Every dowel, cam lock and nail where the page puts it.

I always feel so manipulated when I read it regurgitating the prompt you gave it in such an advertisey way, like it's trying to assure you it met _every_ word of the prompt, regardless of how it actually did.


I honestly don’t know why people get worked up over this. I assume that either the person doing it isn’t a native speaker, or they are a techie that just doesn’t have good English language writing skills. And having those skills is rare for technical people. It’s much easier to deal with Claude than what those people would have written without assistance.

Over the last few decades I’ve read many technical docs that were obviously mechanically translated from some other language. This is much better.


Technology doing code is really cool because you give it an input, a functional purpose, and your intended function output - and it can do a pretty darn good job at producing the code that fulfills the requirements.

But when you want me to read something AI written, without the human refining the UI I see, it almost seems like I'm a test case and all the author cares about is my output in terms of "I like it" or "I don't like it". There's no information about _why_ the author wanted me to see this. Can you tell me your inspiration for making this? Some sort of personal connection I should have with the thing I'm seeing?

Vibe coded apps can still be useful in this state as tools, but don't expect me to use my creative brainpower to provide feedback on your product in a constructive manner if you can't be asked to ensure the text is your intention for the code, not how the AI interpreted it.


I don't really get worked up over it, but I do have a preference for original writing these days. If it's someone's pet project or something creative, I prefer the authenticity of the creative's actual writing. AI writing is just so boring and sterile out of the box. You lose nuance and personality.

It is a bit different with translation, in that case AI translation might get closer to the original intent than manual translation by the author. But otherwise I want the words from their brain meats, not layers removed and cleansed by system prompts and training.


I just know if the first spoonful tastes like dung then I'm not going to keep eating the soup.


> Over the last few decades I’ve read many technical docs that were obviously mechanically translated from some other language. This is much better.

I don’t think it’s better. Any frontier LLM is capable of competent translation without significantly re-wording the text.

I use Whisper to transcribe foreign language videos and then run them through a couple of models to translate into English. It works really well, and much better than mechanical translation because they can figure out likely mistakes based on how whisper might have misheard (or surrounding context from the script).


I dont find it better. I fimd it significantly worst.


Thanks for the feedback! Fair point about the claudism. The copy could definitely use a more human touch. I just don't have the capacity to go through every line myself lol


Just don’t use Claude Code for copy and run your “Claudisms” thru another model like Gemini for better results.


The worst part for me is, with a little effort, you can make a final product that doesn't use claude-isms and is 100% correct.

But the creator doesn't care because their goal isn't to make a tool people want to use. They make a minimum viable product, that looks super impressive but lacks that final polish and move on.

It infuriates me enough, that when I get the time I'm going to recreate the 'lighthouses of the world' example I saw the other day. Only I'm going to do it right...


> But the creator doesn't care because

I’ve always found it annoying when people make declarations like this. You can speak criticize a work without attacking someone’s personality.


A person's intent leads to the quality of a final product.

They should care if it actually has errors.

They should care if it has claude-isms.

But it certaintly seems like they don't care, at this point in time, about those things.

I'll take your feedback on board. But I'm not calling them stupid, or a bad person. I'm saying their goals are not aligned with my idea of a finished product.


or some people don't care about claudeisms as long as it is to the point and correct.


A person's intent can be the problem though, we can criticise that. We're not factory bots here (yet), the community aspect is human in nature not just output focused.


I've noticed a usage shift in the last few years. In 2020, "doomscrolling" originally meant to be stuck scrolling at all the doom-y posts we were bombarded with as the world fell apart around us. As things "stabilized" in the last few years, I've noticed that everyone and myself still tends to use "doomscrolling" even just for when we're stuck, locked in place, scrolling low quality content.

I figure in ten years when someone asks what "doomscrolling" means, we'll probably just shrug and say it means scrolling away to our doom.


I use the phrase "brain rot" for the low quality stuff that I scroll through. I do really hope that we retain the original meaning of "doom scrolling", as I think it's still as relevant today as the COVID days.


> runs as a 1 kHz leaky-integrate-and-fire (LIF) simulation

That hyphenated thing is always such a give. Clearly a thing discussed in the Claude conversation, made its way into the plan, then the readme. It always creates hyphenated phrases out of things it feels are core requirements.

I'm actually over caring about Claude's obsession with the concepts of "real", "honest", and "rather than". What annoys me the most now is how every single sentence, bullet, etc that even makes its way into the products it builds are advertisements for itself. Product settings pages now read like ads, or like something written by a student begging for extra credit from the teacher. You can't unsee it!


To be fair, as a term it has always been written as "Integrate-and-Fire" neuron model. In this case correct would be "Leaky Integrate-and-Fire" but that is nitpicking.

Resources that outdate ChatGPT: - W. Gerstner "Neuronal Dynamics" (2014): https://neuronaldynamics.epfl.ch/online/Ch1.S3.html - E. Izhikevich "Which model to use for cortical spiking neurons?" (2004): https://ieeexplore.ieee.org/abstract/document/1333071


You could say it landed cleanly.


Good instinct. Fair challenge. This corrects my framing. It makes your point sharp. This is a significant finding. Positively confirmed. That settles it, and it flips the picture. Honest verdict - this is not small. Let me wire it in.


There's got to be a Weird Al song like this coming soon.



This is one of my favorite things to have ever read on HN


Bless the lord that it's not been presented to us yet, because you just know that it exists already.


"Bet ya a signed dollar it's true." - Musk's follow-up

It went so, so far beyond a schoolyard taunt. All of my friends and I remember the debacle clearly, and it's the moment we all went from childhood idolizers of Musk to feeling pretty sick about the dude.


You can achieve that with a background-image, because you can stack multiple gradients.

Since this problem bothers me every time I see it on the web, I put together a quick example:

https://jsfiddle.net/ng7hmuxd


Now you need to publish an NPM package with this. Or you know what, I'll do it—and take credit—muahah! Agent, steal this guy's code!


> you could go to a rave and be you without fear someone would take pictures or people outside the rave would know who you are

These still exist! Look for events promoted as such, or look for smaller local events for the genres you're interested in - the latter might not ban phones, sure, but the vibe is still what you're seeking. Nobody's recording you.


I feel the same thing.

I think part of it also is that, games with the same scope of flash games are still being made, but they're being made for phones which is where the customers. Flash games were the perfect mobile game before mobile games existed.

But the magic was that flash games were created on the same machines they were made on, so curious players (often kids!) had a natural funnel in to dabbling with the creation side, so whole communities of creatives formed naturally.

I don't know how we can solve this disconnect between creation and consumption :( Sure there's many apps that let you build content from phones (swift playgrounds, other game-making apps, and now a whole gold rush of agentic prompting app-building apps...) but a phone is inherently non-immersive so I don't know how a creator can ever get into a flow state of building content on a phone itself.

But also we possibly just miss being teens on computers.


The soul of this game requires that your cursor can go "behind" things (like trees, or partially submerged in water), can have subtle nudges to keep you on paths and add friction when in water, and also to be able to take full control of your cursor for the lazy river etc!


How do you proceed? I've tried clicking and interacting with everything I can find but I just see the spinning cassette model. Looks cool though!


Check your extensions, might be blocking the cookie banner. For me uBlock blocked the cookie banner. Afterwards it worked just fine.


nor me. tried space bar. is it a firefox problem?


or an ad blocker, here i had to disable to load the cookies consent window. On Firefox worked better for me here, Chrome had some lag.


Try holding spacebar or tapping it to continue.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: