I think the Vibecoding reddit has accidentally stumbled on the best description of vibecoding:

It's "roleplay for guys [it is always guys] who want to feel like hackers without doing the hard part".

(Source: https://www.reddit.com/r/vibecoding/comments/1mu6t8z/whats_the_point_of_vibe_coding_if_i_still_have_to/ )

#ai #vibecoding

@tante
Associations:
1. Today it felt like I fix stuff someone vibe-deployed. I know, the plagiarism-machine cannot and maybe will never be able to do stuff like that, but people can produce output that superficially looks like machine-generated.

2. Do we know for what kind of programs and languages the helpfulness of LLMs is mostly claimed? Is it maybe websites and other graphical output that always needs repetetive, logic-poor, non-complex but syntactically voluminous code?

@tante been thinking about something similar for a while. Essentially we've been on this trend for decades of "dumbing down" technology in the sense that you completely lower the learning curve ti be able to use it. For a lot of things, I think it's a good idea. However it does seem like we have almost gotten entitled to it. "Oh, I can make this supercomputer in my pocket edit pictures easily. I should be able to code out videogames with a few sentences!"

I had this mindset when I was 8 lol

@tante

What's sort of funny is one comment takes someone to task, and mentions the guy who came up with the phrase. Meanwhile every comment seems to be based on a varying definition of the phrase very different than the original meaning.

I get that this is just how culture works but it's sort of hard to discuss a topic when no one first defines what it means to them.

@tante I always thought programming was fun and easy, and that anyone could do it. Until vibe coding became a thing lol. Not sure why you’d ever want to skip the “hard” stuff, that’s the fun part. Making products is only enjoyable because you built something from nothing and it was useful. I can’t relate to people who only want an end product right now (and I’m a visual learner)!
@tante From what I get outside of the dev world, vibecoding and llms are just glorified Translation Memories with added mistakes. Which is a shame because translation memories are much more efficient at what they do (predictive text based on previous translations).

Correct me if Im wrong, i never use chatgpt but it seems thats what it does.

@tante A comparison that I’m a bit hesitant to make, since it actually involves doing the work yourself, but has some similarities, so I’ll make it anyway: I know how do some plumbing, but if it fails spectacularly there is a high probability that I wouldn’t know how to fix it. So I hire a plumber to do it.

Vibe-plumbers not only wouldn’t know how to seal a joint, they wouldn’t know which joints need sealing. Wait, I’m giving them too much credit. They wouldn’t even know what a joint was.

@tante The reality is they *will* have to do the hard part later (vibe coding is horrible at generating maintainable code and good abstractions; unless that gets magically fixed, those prototypes need extensive rework).
I found a podcast with the Anthropic Claude Code team on that quite "insightful": they saw "less dependency on existing libraries, just generate local specific code" as a *plus* (b/c velocity).