<rant> I really dislike that term.

It’s antithetical to everything programming means to me. I don’t even like the word "coding"... Programming is an art of intellectual precision, and I prefer the language used to talk about it to reflect that.

How do you know you’re a top “vibe coder”? You really dislike what I wrote here. You can feel it in your gut. The "vibe" just isn’t right. </rant>

No, just using AI to solve specific tasks in my usual software dev process. Sort of like pairing with a junior dev who is genius in some ways and retarded in others.

Absolutely not. I enjoy my craft, why let something do it for me, and do it worse?

While I agree with the rather negative sentiment towards the term itself, I find it at least interesting to see how some "engineers" have their pride and feelings hurt here. Well, turns out software engineering isn't gonna be amongst the "sexiest jobs" rather sooner than later and the 2021 "all the companies want me" arrogance is erupting in real time. Jokes on who?

No but that fact that kids seem to like it and people here hate it so much makes me think that it is a legitimately novel thing in the world. I would like to try it out but I bet it will be hard to get a satisfying feel for it since I've come up in the era of rigor and understanding.

No - all the projects I'm working on at the moment benefit from high code quality and maintainability. Either I'm being paid for them and want to deliver a good product, or if I'm giving them away I want them to be useful to other people long term and not a black mark on my github.

However, I've tried it a few times after getting stuck on a bug or approach to implementing a new feature. In all cases it was much faster than me at completely failing to solve the problem. (models: QWQ, Gemini 2.5 Pro, Claude Sonnet 3.7) For now I still find LLMs more useful as a sub-line autocomplete.

One vibe-y thing I have found to be handy is translating between languages. I had a project that needed some functions moved from Python to C and JS and Sonnet did a pretty good job of that. They only needed a few manual tweaks to iron out floating point issues.

I use AI to write 95% of my code, but I wouldn't say I 'vibe code'. Even if I have it generate an entire script (rather than a few functions), I'll give it a quick 30 second read-over before running it. I want to make sure that A) I agree with what the AI produced, that it's a reasonable take on how to solve the problem, and B) sometimes I learn something new from the generated code.

It seems like a waste to not read code in your project, even if quickly. What's the harm? And think of the upside.

No. "Vibe Coding" is a scam, and disregards quality and encourages more slopware.

Clicking "accept all" without checking what the code does is destructive and is not engineering.

No. At least not in the way this term is defined - which seems to just take whatever the LLM spits out and push / check-in the code.

I use LLMs otherwise and careful to test (if it is code), validate and generally not take things they say for granted.

I picture ‘vibe coding’ as coding under the influence or not following proper engineering principles. No one is talking about ‘vibe building’ bridges or buildings - yet anyway - and I hope we don’t get there.

Yes, I put music on and do regular coding while vabing to the music.

Yes, and before it got the name, even.

I do like @techpineapple's phrasing of "vibe tripping and falling on my face" (https://news.ycombinator.com/item?id=43536079), this also matches my experience.

Only for personal projects because of the tripping hazard, but yes.

Yes!

https://tendayweekcalendar.com/

Vibing in progress...

lol, no. I've used LLMs to spit out example code, or to give a first-draft explanation of something that I can't find good documentation on. I don't even end up with snippets that I'd be comfortable copy-pasting into either my personal or professional code.

The closest I've gotten was using ChatGPT to generate the initial code for a kind of toy program. It got stuck in a loop though, and the code was a mess anyhow. I refactored and then finished adding the features I wanted in there myself.

Yes? With the more than occasional vibe tripping and falling on my face, but I’m trying.

Nope, it just feels sloppy + a weird mix of zero control and neon pink.

Yes! My role is not programming. But often I write code in Google Apps Scripts or Python to automate processes.

An example would be like when one of our salespeople closes a deal they complete a task in our sales software that triggers a bunch of Asana tasks for the onboarding team and builds the Google Drive folders and documents we'll eventually need for that client.

I used to have to figure out the code for things like that "by hand" and I was occasionally limited by my technical ability.

But what I've found now is that I'm limited by my knowledge of our business. If I can explain it to ChatGPT, I can get working code in minutes for things that used to take hours.

And I'm trying things now that end up taking hours that I wouldn't have even tried before because I saw them as "too hard" for me.

I honestly don't even read the code it generates. I have ChatGPT generate own log files that I feed back into it for debugging or analysis when anything breaks.

Hell no.

No.

I don't even know what is supposed to be different about it ... sounds like a meme.