given the claude code leak i'm starting to suspect that when boosters say they don't look at the code it's an act of fear, not confidence
@hynek Care to provide evidence? Have read quite a few articles, much about interesting architecture decisions. Nothing about poor quality.
@osmosisch @hynek Thanks, I‘ve seen that. Have seen plenty of production code in pre-AI times that has stuff like this. Wouldn‘t also say that the huge main file is a sign for bad AI coding… So, not convinced this reeks of poor AI code. Seems also this is what most comments say.
@Mastokarl @osmosisch I don’t agree but I won’t argue the point with you because my main theory for the broad acceptance of slop is that the majority of people lack both taste and engineering skills

@hynek @osmosisch I get your point, even if much production code looks like that, tech debt will always bite you.

But what I noticed with AI code is that the choice between 100% beautiful hand-written code that takes t time and good machine-written code that takes t/2 time I tend to lean for the later, unexpectedly (I‘m the kind of person who writes their HTML by hand because I can‘t bear the shitty HTML that web site generators create). LLMs write - in my opinion - „good enough“ code…

@Mastokarl @hynek Personally, I find if the code does what it's supposed to, then it's good enough.

As long as you have enough tests then you don't really need to look at the implementation. Hynek is too precious. We're past that now.

@judy2k @Mastokarl and given how overall software quality managed to further deteriorate over the past ~6 months – from an already very low baseline – I don't think my preciousness is the problem here.
@hynek @Mastokarl It's fine as long as you give the tests a quick skim to make sure the LLM has written enough test code.
@hynek @Mastokarl People keep sending me this gif, I do not know why.
@judy2k @Mastokarl must be a people problem
@hynek Yes, I'm pretty certain it's a coincidence. It can't be me.