Top
Best
New

Posted by usefulposter 1 day ago

Don't post generated/AI-edited comments. HN is for conversation between humans(news.ycombinator.com)
4142 points | 1630 commentspage 17
loeg 1 day ago|
It's an interesting guideline, but will require self-enforcement.
LtWorf 1 day ago||
I think it's hilarious that whenever someone complains about it they're a luddite, and now this happens on a website that is filled with LLM enthusiasts who have done nothing but overpromise.
lapcat 1 day ago||
I had been wondering if and when HN would update its guidelines for this. Glad to see it.
Jeffrin-dev 20 hours ago||
the ai humanizers are getting out of hands, any experiences ...
rc-1140 1 day ago||
The next step is to forbid generated/AI-edited posts.
forgetfreeman 23 hours ago||
There's an element of cognitive dissonance to the community's response to AI that I find fascinating. Nearly unanimous rejection of AI-generated content while simultaneously breathlessly touting AI tooling in significantly more sensitive (and lets face it riskier) environments like the company codebase.
plewd 22 hours ago|
I think people care less about risk and more about human creativity & genuinity. Personally, I get disgusted when I see AI encroaching into artistic fields because I hope new technologies will be used to replace our monotonous work, not take away from authentic discussion/work.
forgetfreeman 18 hours ago||
This and other social media are hardly platforms for authentic discussion, and as far as artistic fields go AI is perfectly incapable of encroachment provided you accept Stephenson's definition of what makes "art":

"Hard art demanded commitment from the artist. It could only be done once, and if you screwed it up, you had to live with the consequences." - Neil Stephenson, Diamond Age

I feel like what you're arguing for here is "it's fine as long as it's convenient for me".

nomel 1 day ago||
I would enjoy a "block user" feature, to help this. I personally want to live in an online bubble of interesting thoughts. This seems close (or better, since people I enjoy can contradict my own flags) [1].

[1] https://news.ycombinator.com/item?id=47141119

arjie 1 day ago||
Haha, I feel the same way. I want to block and be blocked so I made this: https://overmod.org/

It's pretty easy to rewrite if you want. Just point Claude Code at the repo and go. But I think there's a little bit of network effects in that I want to subscribe to some trusted people's blocks too. But overall it's quite helpful. See how much fewer I get:

    849 comments | 138 hidden | 87 blocked | 23 green
SauntSolaire 1 day ago|||
Excellent thanks, I've been looking for something like this. Now just need more people using it to get the friend-of-friends feature useable
kelnos 1 day ago|||
I'm torn on this. On one hand I do agree with your goal about wanting to live in a bubble of interesting thoughts. But on the other... I know I have my biases, and I'm sure I might end up blocking people who actually are insightful and interesting but either a) had an off day and shitposted, or b) says insightful things in ways that make me angry and get past my sense of reasonableness.
nomel 1 day ago||
Good news, it doesn't block! It just puts a red mark next to their name, so you can put less effort into that comment, if you choose.

And, it's social. If someone you've marked green is also using this, and they marks someone green that you have marked red, then you'll see a contested red-green next to them, which is a good "you should probably reconsider" indicator.

b112 1 day ago||
A good idea, but I lament the downfall of Slashdot.

They had the same sort of system. Friends of foes, they calldd it.

krapp 1 day ago||
I suggest Comments Owl for Hacker News - one of many available plugins that make this place tolerable.
crossroadsguy 1 day ago||
Apple's proofread is essentially spell-check and punctuation until it isn't and even in a few-sentence-long para you'd see it has sneakily changed a lot and Apple being Apple you, the customer, obviously has no way to set it to "only fix spelling, punctuations and leave everything else including grammar as it is" and I've a feeling a lot of folks are at least using proofread or something on those lines. But then I really don't think browser's "spell check" ought to be kosher either if the content has to be the human's because those mistakes are also makes such text human and in some way unique. I don't think it's an easy line to draw but weird seeing just comments "targeted" here.
schappim 1 day ago||
I have a kid with severe written language issues, and the utilisation of STT w/ a LLM-powered edit has unlocked a whole world that was previously inaccessible.

What is amazing is it would have remained so just a couple of years ago!

DennisP 1 day ago||
What is STT in this context?
schappim 1 day ago|||
Speech to text
zahlman 1 day ago|||
Does your kid post here?
ranger_danger 1 day ago|||
Agreed... there's often other perspectives people never thought of like this, which is why they say "strong opinions about issues do not emerge from deep understanding."

Even if you're just inexperienced in the language you're communicating in and are trying to have better conversations, it's very helpful.

For cases like that, I say just don't tell people... I think it's unlikely anyone will be able to tell either way.

ex-aws-dude 1 day ago|||
Come on dude, its obviously just to prevent spam and not for your super specific case

These are just guidelines

schappim 1 day ago|||
Title literally says “AI-edited comments”.
zamadatix 1 day ago|||
Sure, despite another guideline saying:

> Otherwise please use the original title, unless it is misleading or linkbait; don't editorialize.

the title being the changelog is still probably the better choice because the discussion here and linked are about guidelines in the page rather than absolute rules or a discussion about the title alone.

Many of the other guidelines have exceptions too, and various strengths. E.g. "Throwaway accounts are ok for sensitive information..." is a pretty weak guideline in practice while "If the title contains a gratuitous number or number + adjective..." is often over-enforced by automatic tooling and stuff like "Please don't use uppercase for emphasis..." CAN sometimes just make sense where a use of italics might easily get missed WHILE OTHER TIMES BEING THE REASON THE GUIDELINE WAS ADDED.

Edit: Well I wasted my time writing that as dang said it better anyways https://news.ycombinator.com/item?id=47342616

jasonlotito 1 day ago|||
> HN is for conversation between humans.

It also says that.

The intent of the guidelines are important. Using AI to generate the STT is fine. The conversation is still between humans.

djohnston 1 day ago||||
nuance and basic common sense left the chat about ... 8 years ago.
majorchord 1 day ago|||
How is it obvious?
eudamoniac 1 day ago||
[flagged]
xbryanx 1 day ago|
Great message...but gosh, can someone throw 15px of padding on that <td>? I know HN is supposed to be minimal, but I had to check the URL to confirm that this was a real page because of the odd design.
zahlman 1 day ago|
It also says:

> Please don't complain about tangential annoyances—e.g. article or website formats, name collisions, or back-button breakage. They're too common to be interesting.

Feedback such as this is better as an email.

xbryanx 1 day ago||
Thanks! I will share this.
More comments...