Post

I think it's another indication of how hard it is to deal with the firehoses that AI companies keep encouraging people to aim at each other.

Governance problem

Sounds like the beginning of the end

Ironically, the most obvious and most upvoted slop articles I've seen have all been AI critiques

It also has a user base accustomed to taking shortcuts (DRY) and the skills to evade common AI detection mechanisms (ai shibboleths)

I mean, my blog is regularly accused of having AI writing (a leap from the fact that I do use AI images) despite the fact that I never could have an AI write even a just decent "in-style" summary of my articles. People hallucinate too. Also, it's just a way to insult you nowadays.

Getting accused of LLM use because you used an en dash (not even an em dash!) gets old really fast, when you have always written like that. Most of the "AI vigilance" on HN is actually low-effort shitposting.

This is the core problem with detection-based approaches to AI content: they require someone to be the accuser. Civility norms make that costly. Disclosure-based approaches (provenance labels, self-identification) don't need an accuser. The immune system works without anyone pointing fingers.

hacker news has always been mostly slop, just the artisanal kind

I'm increasingly seeing people talk like LLMs, independently.

When I read the comments I tend to find many people reporting AI slop and criticizing it. Are you mostly referring to the fact that AI slop gets attention and votes?

I got praised there yesterday for having spelling mistakes because it made it seem "real"... We may be losing the oxford comma, em-dash, smart quotes and such, but we gained pause commas and ended the reign of grammar pendants. Writing can be jazz again :D

Wrote about slop and the importance of maintaining genuine voices. You might enjoy it! jola.dev/posts/the-so...

Your comment reminded me that this generation of children is being raised, at least partly, by LLMs. I know multiple parents who ask Claude questions regularly about this or that. I imagine anthropomorphizing them as a third parent will be common in the future unfortunately.

The best heuristic I have is to check for other posts or article by the same "author" - the slop accounts tend to use exactly the same length, style and shape of content for everything

I find the lobsters posts / discussions more based, but I do think we are ripe for someone to disrupt this space in the near future. Hoping to see an atproto based platform take off

There's also a weird witch hunt going on for exposing AI written content. I get that there's a philosophical discussion to be had whether AI generated content is worth consuming as a human, but if an idea is well articulated and gives you something, there's probably nothing wrong reading it.

I concur. Happens in almost every forum board-type community (for example, Spanish Meneame). Happened in good old Digg, too. "I like the message, I'll upvote."

honestly the "don't accuse it of being AI" norm is the part that gets me. without being able to even name it, how does a community ever build up a tolerance? feels like the slop wins by default. do you think there's a healthy way to call it out, or does that just turn into witch hunts?