jacobian1d ago

Look this isn’t at all a defense of slop code, but it has me thinking — how much does code quality matter, and why?

It’s maintenance, right? We care about readability because we know we’ll have to make changes, fix bugs, etc.

But so … imagine a codebase that’s magically bug-free and feature-complete. (I’m aware this is a strawman - that’s the point, it’s a thought experiment.) Does it matter if this codebase is well-written? I’m not sure it does! (1/5)

Show threadJan Lehnardt 1d ago
@jacob et tu brute?
Show threadjacobian1d ago
@janl I’m not sure what you mean?
Show threadJan Lehnardt 
@jacob Every week someone I look up to falls for the genai scam.
Apr 2 at 3:37pmWeb
Show threadjacobian1d ago

@janl *sigh*

If having anything positive to say or any curiosity at all about LLMs means I’ve fallen for a scam than ok I guess. I wish there was room for more nuance here.

Show threadJan Lehnardt 1d ago
@jacob yeah, I know you understand there are some topics where asking for nuance is a dogwhistle and I consider this technology one of them. To me the only sensible position is to stay the eff away from it.
Show threadjacobian1d ago
@janl I don’t know where to start with that. I’m sorry you see me that way. I still think you’re great.
Show threadjacobian1d ago
@janl To be clear - my position is that genAI is on balance probably a net negative and if I could unring the bell I would. But also I would say that about the internet — and “there are good things about the internet” is hardly a controversial option (I think?). I get that The Discourse demands extreme opinions but I don’t think it’s “falling for a scam” to find positive things about AI, and I think it’s dead wrong to approach ANYTHING without a sense of curiosity.
Show threadJan Lehnardt 1d ago
@jacob hence my original post. I’m so very sorry they got you.
Show threadjacobian1d ago
@janl That’s certainly one narrative. It’s not mine.
Show threadJan Lehnardt 1d ago
@jacob I also want to state that I am a person and not A Discourse.
Show threadFrank Wiles1d ago

@janl @jacob can I ask the last time you used genai for some code? Maybe not even your main/work codebase just anything even a toy script?

I’ve found that often the people who hold this opinion tried it for a couple hours a year or more ago and formed an opinion.

Show threadJan Lehnardt 1d ago

@frank you can ask and the answer is never.

I do not think evaluating these things based on their merits is ethical.

The tool you are using has been built by stealing* from me** and my friends and it offers to rent their theft back to me, so excuse me when I am not interested.

*and that’s only one of the many issues
**I’d be entitled to thr Anthropic lawsuit money if I were a US citizen

Show threadAdriano1d ago

@frank @janl @jacob

On the other hand, we've just seen what LLMs did rewriting sqlite, a monstrous botch that resulted in a monstrous decrease in performance. And we've *just* seen the quality of claude code, which was, purportedly, vibe-coded. So.

Show threadAdriano1d ago

@frank @janl @jacob I'm obviously not Jan, but maybe Jacob gets pushback like that because stuff like "imagine a codebase that’s magically bug-free and feature-complete. " and going on from there to talk about LLMs is practically indistinguishable from the propaganda and grandstanding Altman and Amodei and Musk peddle. Just imagine what would be possible if. Fascinating.

Yeah it's fascinating but so is imagining what we could do if we could photosynthesize. And...

Show threadFrank Wiles1d ago
@adriano @janl @jacob I recently got some blood work done and I’m very Vitamin D deficient so I’m personally quite glad I can’t photosynthesize as my vampire ass would be dead 🤣
Show threadjacobian1d ago
@adriano @frank @janl It’s frustrating because the thing I was trying to discuss about the question of why code quality matters, not yet another “LLMs: good or evil” argument. And for the most part replies have been focused on that — and have really teased out some super interesting points about what “quality” means and why it’s important. And in doing so I think makes a much better case against agentic coding ever working the way proponents claim it might!
Show threadDogiedog6415h ago

@janl @jacob @frank @adriano

"Imagine a codebase…" buddy I DON'T HAVE TO, Claude Code leaked! It's dogshit! It's all dogshit! There's little point in entertaining thoughts about how good LLMs """could be""" at some nebulous future when the best LLM """tools™️""" TODAY are nightmares of over-engineered spaghetti that somehow cost hundreds of billions of dollars to function.

We're 4 years into the LLM hype-train, 8 years into transformer models as a concept, and over a TRILLION dollars spent on development. If these things were gonna """get good""", IT SHOULD'VE HAPPENED BY NOW!!!