alt Hacker NewsMy eternal advice to anyone doing heavily LLM assisted projects is to do them a bit less LLM assisted. You should still use LLMs, in my opinion, they're wonderful tools. But I know reading this page that this text is mostly LLM-generated, and that leaves there to be little hope that much else of the project isn't.
It's one thing if your code is really truly "co-authored by" Claude. It's another thing if it isn't even really co-authored by you.
Evidently the majority of people see no problem with LLM prose, but me and many others think it is some of the most annoying crap possible. Please consider speaking in your own voice.
I know this gets repetitive, but it is important.
Replies
(concurring opinion)
LLM generated text lacks a certain "soul" to it, and it becomes more and more empty the more places one reads it.
It has a certain quality to it that, when read again and again, feels hollow. That hollowness comes not only from the text itself and the same sentence structures and turns of a phrase, but also (and yes, https://en.wikipedia.org/wiki/Wikipedia:Signs_of_AI_writing#... ) that it suggests that the author didn't do more than massage the text with their eyeballs (as my high school Latin teacher accused me of doing).
And so, why should I (a human mind running on wet stuff and fueled with caffeine and various sugars from my morning soda and breakfast roll) pay any significant amount of my limited attention during a day on something that shows no indication that it was ever had its human co-author think about a single word that was put into the page?
And sometimes, the text belies that it wasn't written by a human as it misses things that are part of the human experience (or adds things that are not). "Keymap, disk, partition sizes, package set — then it shows you what it is about to do. Real time, with the long waits taken out." No human would write that. No human proofread it.
There is a deluge of content out there. Far too much for me to read everything that might be interesting. Text that looks like it's written entirely by an AI is more likely to be uninteresting than something that was crafted by a human.
If the page itself hasn't shown indication of a few seconds or minutes of a human touch... how much time should I spend reading it to see if it might be interesting to read more or look into the project?
This is the 3rd time I am reading such a comment on HN within a day and I absolutely agree. It comes off as instantly disingenuous and a ‘turn off’ for someone to use an LLM for basic writing stuff. Not to mention it ends up being distracting in a meta sort of way.
Also, to congratulate the author feels like congratulating a tech savvy manager for a project rather than the skills and knowledge of a programmer
It’s cool they’ve managed various agents to get their end result but I’ve always been interested in coding, computer science and maths rather than tech savvy management
And perhaps tech savvy manager will replace programmers but if I wanted to get into management and business analysis then I would have done so
As I’ve also repetitively said, it’s simply dishonest and plagiarizing to offload most of the work you (explicitly or lying by omission) claim as "yours". I can’t understand people who think it’s just fine.
It's not important at all and I wish people would stop mentioning AI. If you have criticisms on the docs or code, by all means make them but they shouldn't concern "AI vs human" considerations. Humans have written plenty of bad and documentations, too, it really doesn't matter who wrote what to what extent, what matters is whether it's the right quality.
I don't think its important. I think these comments are more annoying than ai prose. You are fighting a losing battle and annoying everyone in the process. Adapt.
FWIW by this point, many of us are in deep symbiosis.
Useful or not, another Amiga / Unix fan probably had fun and some learnings too.
At the very least, the real-world impact is that tokens have been spent, the outputs open sourced.
It would be nice to hear their voice survive the loops of agentic abstraction of course but I don't know if that is even possible at that higha level.
Maybe it's that, maybe it's "just another issue of alignment"? Potentially serious in aggregate and over time..
I personally don't mind the imo little "neuralese" (is that what we call it now?) here specifically. In any case it's good for you that you can (still?) tell?
Why not applaud the playground time/energy/token budget directed? Especially as a decades long NetBSD (and Amiga:) fan I do appreciate multitudes; maybe here not exactly the case in tone of writing but in relatively cheap, fun experiments / projects during the agentic take-off of 2026. The whole trend reminds me a bit of remix/DJ culture; the "producing DJ"?