HN kills lots of posts. I try to be careful about my online footprint (since HN posts are forever), and try to switch to new accounts every so often. It's no use anymore, HN just kills any post I make from a new account, even when I spend 20 minutes researching a response and trying to get useful information.
replyIt doesn't even show you the post is killed, it looks to you like it posted fine, and you have to logout to see it's actually dead. It's an approach that's extremely hostile to the user.
It's specifically against the guidelines to keep registering new accounts, and this is a good reason why. We have to have ways of determining credibility and authenticity, now more than ever, and a track record of good posting is one of the best ways to do that. We are drowning with spam and low-quality posts/projects posted from brand new accounts. If it's a well-researched, high-quality post, of course we want to give it exposure. We just have to be realistic about what we're up against.
replyHN front page is about 25% LLM written blog posts at any given time.
replyThere’s no rule against submitting LLM-written—ahem, “cleaned up my notes”—articles, just comments.
replyBadly-written articles are still unwelcome on HN, wether AI-enhanced or not, and obvious LLM smell is definitely lowers the quality of an article. But it's true, we don't ban every article with any evidence of AI-assistance.
replydeleted
replyI have read enough “you are replying to an LLM” comments that I am pretty sure this is still a hit or miss process.
replyWhy do you think those comments are accurate? Maybe those comments are by LLMs? If you believe crowd wisdom on its face, you will have big problems with LLMs.
replyIt needs help. I often pipe my screed though an LLM and post it. I do request that it use a 10th grade reading level, and no emdashes.
replyFor giggles, here's how it would look for this comment. Rather meta, but in this case it removed the "It needs hellp" so here we are.
I often run my screed through an LLM before posting. I ask it to keep the writing at about a 10th grade reading level and to avoid em dashes.
The question is how reliable that detection is.
reply>HN autokills comments it detects as LLM
replyNo it doesn't. Unless you have proof.... ???