I had a story removed with the explanation, “Submission is about using LLMs.”

I don’t really care about getting the moderation decision reversed, but I am not sure I really understand the new policy. Surely, it’s on topic to talk about (in this case) using an LLM to read your logs to try to improve service robustness? “Will this improve the reader’s next program?” If your next program is logging/telemetry related, maybe the removed article would, no?

score 1 · canonical

I was withholding judgment on this change in policy, but after seeing how moderation has been applied in the latest OpenBSD story, together with the mod statement that there’s now a “strict stance on removing posts that neutrally discuss whether a project used AI”, I have to say I’m incredibly disappointed.

I realize it’s tiring to rehash the same points over and over in the comments. I’m as uninterested as anyone else in discussing the minutiae of whether something is slop or not. But the mods have to realize that these discussions are a symptom of the internal division in this community. Banning these discussions won’t solve this, it just makes the problem invisible and it automatically gives preference to one side. I’m not terribly surprised, since this site has made a similar choice before: discussing the ethical considerations of our job remains off topic for the most part.

I just wish the mod team was direct enough to say that they don’t consider a large part of this community valuable enough, and are once again siding with the default path of pretending everything is a-ok as long as we ignore the problems in the industry (mainly, how the present changes are a direct threat to the livelihood to the majority of software developers, the rise of anti-intellectualism in the industry, and the devaluing of knowledge).

score 4 · canonical

First of all, I’m not personally affected since I’m not filtering out the vibecoding tag and although I’m mostly not interested in LLM-authored content, I find the balance on lobste.rs fairly reasonable. I’d like to try and summarize my understanding of the discussion and some options I see.

The vibecoding tag’s name and description was fairly ambiguous and as a result, different groups of users started using it differently:

• Some lobste.rs users are fine with LLMs on an abstract level and just want to get rid of low-quality content (me included). The current set of changes seems to cater to this group by treating low-quality slop the same as any other low-quality content (flagging it as spam).

• There seems to be a second large group of users who just do not want to see anything connected to LLMs in any way. From the discussion here, I gather it’s not just about filtering low-quality vibecoded projects or LLM-authored writing, but about creating a completely LLM-free safe space. This group (arguably mis)used the vibecoding tag to filter the content since that was the only available option (to the annoyance of the first group) and is now left without an alternative.

• I expect there are more groups along the continuum between these two but they do not seem to be as strongly represented in recent discussions. Therefore, it seems reasonable to focus on solving the issue for the second group (at least for now) rather than trying to create a spectrum of tags to cover options in between and complicating the discussion.

I’d be in favor of adding an llms tag that can be used as an LLM trigger warning to mark all content that mentions LLMs in any meaningful way. Naming it llms instead of vibecoding should remove the associated negative stigma and avoid endless discussions about whether something is actually vibecoded. I prefer the name llms over ai-usage and similar since the sentiment seems to be specifically about the current wave of LLM hype, not about AI in the abstract. Users that want to see LLM-related content should ignore the tag and not expect any valuable information out of its presence other than “this post mentions LLMs somewhere”.

On a related note, I’d be in favor of banning all discussion about the LLM purity of a submitted post and debating non-technical context of LLMs on articles that do not themselves engage that topic. Such discussions tend to be extremely repetitive, reductive and off-topic for the original article.

score 1 · canonical

I am pleasantly surprised by this. Yes, lets focus this site on computing! I want more deep discussions on that.

Machine learning has been at the center of computing since the beginning. Alan Turing wrote “Computing Machinery and Intelligence” paper in 1950. He talked about building “a child machine”. John von Neumann was influences by McCulloch and Pitts model of artificial neurons in 1943 and later wrote The Computer and the Brain. Claude Shannon built Theseus in 1950s, a mechanical mouse that learned to go around a maze. He founded information theory which is deeply connected with LLMs. Even Ada Lovelace talked about how machines could only do what they are told and could not originate anything, she was wrong. But the point is ML was at the very beginning of computing.

So if you love computing but aren’t studying machine learning (which LLMs are part of), you are missing a huge part of what computing is about. Lobsters is about computing and LLMs have deep connections with computer science, math, and information theory.

What sucks right now are the groups controlling this technology! Fortunately the Chinese labs have been a huge thorn in their side releasing open weight models. DeepSeek just released their inference engine software code as open source.

score 1 · canonical

deleted by creator

355E3B, can you explain what makes an LLM-related post acceptable or not acceptable on Lobsters?

Recently you removed posts about using agentic monitoring tools and a podcast about where the craft of programming is going, both for the reason of “Submission is about using LLMs”. But isn’t the ai-tooling tag for exactly this kind of thing? What is the goal of banning discussion of an AI-based monitoring tool?

score 1 · canonical

Irritatingly my vibecoding filter has been disabled and comments that should be hidden keep appearing in my feed.

score 1 · canonical

Hey mods, for whatever little it’s worth, you’ve got an enthusiastic thumbs-up from me. There is more than enough pessimism on the Internet, I hope these changes steer the boat into more productive and level-headed winds.

I come here for technical content, not politics. Being anti-AI is very firmly a political choice. I have had to hide my interest in AI technologies and locally-hosted model in some threads for fear of being lambasted by the small (but surprisingly vocal!) anti-AI crowd that came here just because the presence of a “vibecoding” tag (or rather, how it was usually applied) gave the impression that anyone having a non-negative opinion of AI was fair game for ridicule in this space.

To be clear, I hate low-effort slop as much as anyone else. But when I see it (or anything else that doesn’t interest me), I just close the tab and move on with my day. Life’s too short to be wasted shouting into the void.

score 1 · canonical

I’m ready to try this scheme for a couple weeks and then reflect. If you’re thinking of leaving, I invite you to try delaying that decision and see whether you do in fact have a bad time here. Granted I have been reading the tags before I click rather than filtering things out of view, so it’s not an immediate shock to me, but selfishly, I’d rather you all stay so I can keep learning from you.

score 1 · canonical

This comment will probably not be seen, but I want to thank you all for your hard work on this site; not just its existence but for the work you put in to keep the “vibe” (see what I did there?)

score 2 · canonical

I agree with the reasoning but disagree on the methodology. The new ai-tooling tag is too general, and encompasses a number of projects that fall under the umbrella of “a human putting in hard work with their fingers,” rather than what people actually want to filter out, which is “content that is not sufficiently technical or thoughtful because of its relationship to generative AI.”

For example, I’m working on a machine learning math compiler from scratch in C, but I’m writing it with my own two hands without assistance from AI tooling (my reasoning for this practice is not relevant). Does that fall under the umbrella of ai-tooling? Hopefully not, given the definition of content under the ai tag is “Developing artificial intelligence, machine learning, LLMs,” but much of the community is not aware of that definition.

If you attempt to rid the site of a “tag of shame” by lumping in things that AI-cynical people may actually want to read, all you’re doing is hurting interesting content that relates to AI in subject matter but not in quality. What you (moderators) are trying to do is separate the quality of the content from its subject matter, but what you are accomplishing in practice is removing the proxy by which we may quickly judge the quality of the content, as unintended and emergent as that mechanism may be.

I see the solution to be to judge content not on its topical nature, but by the things I wrote above: technicality and thoughtfulness. Allow us to publicly judge the quality of the content multi-dimensionally rather than with up/downvotes and “tags of shame.” What people actually want is to know before they click on a link whether the thing behind it is worth their time. Technicality and thoughtfulness are much better proxies for that than the “is it related to generative AI” mechanism.

This said, I do not have a constructive perscription as to an effective solution because I am not a qualified community moderator of any kind.

score 1 · canonical

I’m already 200+ comments late, so many good things have been said. Some less nice things, too. I’ve been complaining about this situation, so the least I can do is to offer my own thanks and (quick) thoughts.

Firstly, I’m mostly glad a decision was made, even if I disagree with it. In the end, it’s better for all involved (staff and users) if Lobste.rs doesn’t turn into a war zone over ethical disagreements, however important they are. So thank you 355E3B, pushcx, Irene and aleph.

Of course, I would have preferred a more ethical approach to this decision. Koala has made this point most eloquently, so I’ll link to it.

355E3B, you said that you “started reading Lobsters because it had the most interesting viewpoints related to computing,” and that was absolutely the case for me too. However, what y’all taught me consistently over these last twelve years is that the ethics of those viewpoints are important, too, and sometimes even outweigh the potential benefits.

Now, I amwas filtering vibecoding, so it’s possible that I missed some kind of acknowledgement by the staff of the issues at play, in which case I apologise for bringing that up.

Where does that left me? Frankly, I don’t know. These last months, I’ve increasingly quit spaces where I had the bad experience of getting excited over something, only to find out that LLMs were used in a key role. So, maybe that will happen here, too. Or maybe I’ll still enjoy enough of the site to make it bearable, who knows. Either way, it’s been twelve wonderful years.

score 1 · canonical

I rather doubt this is the end of the story, but mayyyyybe it’s a step in a useful direction?

score 1 · canonical

Thanks, I appreciate these changes and the explanation behind them.

I respect people’s desire not to avoid posts that are largely LLM-generated or human-written posts about work that is mainly LLM-generated, but I think “vibecoding” as a tag was too broad, and Fillipo’s post was a good example of the tag being used in a “spoonful of sewage” way that always felt more like a witch hunt than a good faith way to help people filter out largely LLM-generated work.

score 1 · canonical

If this kind of community schism happened 20 years ago, 5 “AI free” lobsters forks would already have been created with their own growing communities figuring out where and how to draw the lines. That’s one of the core features of the internet that used to be taken advantage of much more frequently. I have a feeling the rise of social media and the oppressive nature of big tech platforms has had an effect on user agency.

I like the direction the mods are taking here. But I’m also open to exploring other alternatives. If anyone does create their own link sharing community please link it below.

score 3 · canonical