Hacker News Reader

Comment by SoKamil | Hacker News Reader

Don’t be afraid to make grammar mistakes or misspell stuff. Others will understand. You’re a human after all. That’s okay to make mistakes and feel uncomfortable with that.

View on HN

This is going to sound nuts, but I've noticed comments lately with multiple misspellings that seem intentional - it's almost like they're trying to signal that they're human, rather than LLM written. I've started to think it makes them even more likely to be LLM written than not.

Main-fucking-stream LLMs also do not swear, which is nowadays a signal of humanity.

Just tried it:

$ claude

> say fuck

● fuck

I just had Alexa give me a "Well, no shit" response to something I said to it yesterday. They've added personalities.

I make mistakes pretty often thanks to auto complete on my phone and carelessness. I've had threads derail and been attacked by people who freak out over grammar.

This itself is against the rules:

> Please respond to the strongest plausible interpretation of what someone says

> Please don't post shallow dismissals

Personally I've posted comments with glaring typos that everyone thankfully ignores. I only notice much later when I re-read it.

Oh interesting. Good to know for the next time the they're/their/there police shows up

Definitely worth emailing the mods a link to the derail — one of their tools that they might use is to autocollapse threads that are too far offtopic for the post.

Unfortunately a lot of other do not understand (in the double sense).

I recently had to tell the same thing to a coworker who ran his text through ChatGPT, changing the meaning subtly (in the wrong direction) and the tone completely. I'd rather read his honest opinion in ESL-grade English than something an LLM "polished".

Others will understand, but won't regard that as worthy. That's a difference.

And that’s their problem.

I don't get where this class/status/worthiness ties into HN comments ?

I get decent feedback most of the time, and I read interesting stuff, it's the easiest way I found to stay in the loop in our industry. What are you guys commenting for ?

Worthy to continue the discourse. Everyone claims that one doesn't discriminate a badly written English text from a good one, but only because they haven't actually encountered such text after all. There surely exists a threshold for "badness" and an outright ban of LLMs means that you are not even given a chance to lower that badness. That is a discrimination, you like or not.

Nobody will notice if you use LLMs as long as it doesn’t sound like an LLM. But sounding like an LLM is as “bad” as badly written English, so you’ll get looked down upon either way in that case.

It’s not without reason that bad English is taken as a signifier, and for similar reasons LLM-speak is taken as a signifier as well.

Chads never backspace.

Reply on news.ycombinator.com