lobsters score: 21 · tags: announce, meta
The last couple years have been the biggest upheaval in the field for decades. While it’s acceptable, within limits, to express frustration towards circumstance or technology, it’s not acceptable to direct that behavior towards individuals whether they be members of the community or not. Other people can’t be your punching bags and Lobsters can’t be your boxing ring.
Thanks for sharing this perspective. I consider myself quite skeptical of AI and share many of the ethical concerns raised here. But the negativity directed at certain people and projects has sometimes felt more like a witch hunt. I’ve increasingly felt that the atmosphere in this community was becoming toxic, and that’s not the kind of environment a garden party should foster :).
The changes look good. Keep up the good work and thanks for moderating this garden party.
score 2 · canonical
Pending an upcoming deploy, we no longer allow submissions which are code repositories
I’m disappointed by this. I didn’t find spending the two minutes looking at a repo to decide if I’m interested more of a burden than looking at an article to decide if I’m interested.
score 1 · canonical
vibecoding tag is now marked as inactive
It covered a lot. For a site focused on the software craft, the gripes about low-effort content and projects are legitimate, and it’s important that quality bar exists. There’s a lot of blogs I’ve learned about here who excel at technical depth, and that’s exactly what I’m here. Many of those authors even participate here!
I’ll probably always be trying to balance LLM adoption to pay the bills and learn quickly, and the fun and confidence of figuring things out and being inspired by people teaching me.
new tag
ai-toolingcovers “The design, ux, development of AI/LLM-powered tools themselves”This tag is great since I’ve been trying to get up to speed and learn this.
careers
It’s a little sad to see this go. I haven’t found anywhere to be a good place to learn about career progression.
score 1 · canonical
I am in a generally negative headspace at the moment due to unrelated issues, so I will have to let this decision play out for a while before deciding what to do. For a long-ish time I have been contemplating just leaving the site, and maybe this will be the final push.
I don’t envy the mod team. Going from the mission statement, it’s impossible to have a forum focused on computing that excludes GenAI now. The “vibecoding” tag was the compromise - if you don’t want to interact with this stuff, just filter. That’s now gone, as far as I can see. I would have preferred a straight re-name to something that wasn’t seen as pejorative, but oh well.
score 1 · canonical
There is an impossibility here of catering to those who open a link, see any amount of LLM-generated content, close the tab, and flag the post as spam, and those who don’t mind it/are excited about the results regardless. I’m hopeful this will cut down on much of the noise (esp. slopped together projects w/o explanation), but I suspect that posts detailing work only accomplished as a result of LLM use will now slip through, and I would definitely appreciate a tag for this as well. I do think LLM usage is shameful practice, and I do not want to engage with it, and I’m far from alone in that, and I’m certainly not leading crowds with pitchforks to those websites. I still believe there should be some such tag such that we can filter away such posts if we don’t want to engage with them.
score 1 · canonical
Why does that matter? If you need to hide it it’s probably because it’s a low quality project. The recompilation scene isn’t shy about AI use. It’s seriously impressive. I just don’t want to see it, even ignoring ethical issues (and I won’t), it has nothing to do with how I interact with programming. I have as much use for it as emacs posts. Just let me filter it.
score 1 · canonical
I loathe this argument. Sure, you are functionally correct, but you can never know if anyone is lying to you. I hope people would have the common decency to disclose, and otherwise when I find out they haven’t, they end up in my ublock list. I’ll miss some things, perhaps, but that misrepresentation is their failing, not mine.
Besides, it is almost always completely obvious when people don’t disclose.
score 3 · canonical
What you said is true. The userbase of lobsters has bifurcated on the question of AI, maybe it’s time for the website to do the same.
score 1 · canonical
It’s unfortunately been pretty difficult to try and keep the properties of the tag without also bringing along the negative externalities it’s existence has caused. I’ve thrown around the idea of adopting the ATproto “stackable moderation”, which would offer a better way to apply these overlay filtering long term, but that’s gated behind some serious technical work and a few UX issues.
With regards to posts detailing the work which was only accomplished as a result of LLM usage, my suspicion is that the post themselves will also only be acompished as a result of LLM usage and already fall under our definition of “spam”.
score 5 · canonical
With regards to posts detailing the work which was only accomplished as a result of LLM usage, my suspicion is that the post themselves will also only be acompished as a result of LLM usage and already fall under our definition of “spam”.
Unless I’m misreading you I don’t think this is going to be consistently true. Like, the proximate cause of this latest kerfuffle: I think it is fair to say the result it describes was a result of LLM usage, but the post certainly had meaningful human authorship.
score 2 · canonical
With regards to posts detailing the work which was only accomplished as a result of LLM usage, my suspicion is that the post themselves will also only be acompished as a result of LLM usage
This often holds but the number of cases where it does not is numerous. With this information, the best choice for us who do not wish to see the results of LLM usage is to flag even human-written posts about LLM usage as spam. There is no longer other recourse.
score 4 · canonical
The problem is that many of them will not be about ai-tooling, but a result of it; this is what I would hope to avoid.
score 2 · canonical
work only accomplished as a result of LLM use will now slip through, and I would definitely appreciate a tag for this as well.
If such content is welcome on this site, then I definitely agree with this, I’d be filtering that out by hiding this hypothetical tag. I think this is what a lot of people wanted or assumed the vibecoding tag was for… Like the post says, it became an overloaded tag.
I am assuming the mods will try out this new ai-tooling tag and then will be willing to reassess a little bit down the line. Hopefully this second tag for work only accomplished as a result of LLM will be considered. Again, assuming such content is allowed on this site — which I’m personally not clear on whether or not it is.
score 3 · canonical
I don’t love this change but I think it’s an improvement and I really appreciate the work that goes into moderating the site. Thanks.
score 1 · canonical
Thanks!
Other people can’t be your punching bags and Lobsters can’t be your boxing ring. That’s not acceptable.
Are you going to start removing derailing and personal attacks directed at AI users?
score 3 · canonical
Yes.
We’re starting with a strict stance on removing posts that neutrally discuss whether a project used AI. This morning we had a thread go from that, in six responses, to purity-testing whether a project was sufficiently anti-AI and naming individual targets for bile. This can’t go on.
“Derailing” is very hard. Comments that connect stories to prior art and related topics are some of the best on the site. Comments that do so to pick fights look very, very similar. I’d welcome attempts from you or others to define it better than a “know it when I see it” judgment call). Then maybe when we’re better-resourced and have re-established rhetorical norms we can lighten up to that standard.
score 14 · canonical
I would like to carefully express a disagreement with this approach. I’m sympathetic about the troubles caused by derailing, and I understand that such an approach may seem reasonable from the perspective of reducing derailing. However, it leaves me very uncomfortable about the community that this will foster, for two reasons:
-
The update described in this story already makes it difficult for people who would like to avoid LLM-related content to do so. Outlawing neutral discussions about AI usage makes that even harder, indicating that there is no on-site mechanism to find out about LLM usage. The only two remaining approaches are to a) look at the content yourself, which becomes a denial-of-service attack on us at this scale, b) to create an external, optional web extension to share such marks. With the approach taken by moderators, first with pandering to LLM proponents ostensibly due to personal attacks, and now with removing neutral discussion, I have no confidence that something like b) won’t be considered a tool for discrimination and thus lead to bans or being shooed out of the community.
-
This choice strikes me as trying to “keep the peace”. I have seen abusive situations play out, where people kept quiet to placate the abuser, and it never reduced the harm, just made it harder to stand for yourself. I am also sure that you’re aware of fascist tactics, painting all-but-complete support as form of “compromise”, and themselves as victims. I am not saying either of the situations is happening here, but I want to point out by analogy that your choice is not at all a compromise. It is very explicit support for LLM boosters, and it removes all the tools from people like me, all within the span of two days.
I have other qualms about moderating decisions. The fact that I have seen no toxic behavior from LLM avoiders on this thread, and that the discussion has been extremely respectful, contrasts with the purported “boxing ring” behavior as described in the story. While I don’t doubt that such events have happened, the fact that this hasn’t occurred here indicates that you’re throwing the baby out with the bathwater and making a lot of people responsible for the unwise decisions of a select few.
I believe that this change in moderation rules is not something that should be kept in comments and lost to time, but rather expressed in a more public way, just like this story was. In my eyes, it’s a very major difference from what the story portrayed, and it shouldn’t be swept under the rug. I wanted to stay on Lobsters for a while and check posts for AI usage manually. I believed that you’re trying out a compromise and have everyone’s best interests in mind, and that you might consider adding a trigger warning-like tag. But making this decision on moderation, without constructively responding to my or anyone else’s grievances about what this means for people who would like to avoid seeing LLM-related content, indicates that I am not welcome here at all. While I respect that this is your site and you can do anything with it, it leaves me deeply hurt that a place I enjoy so much is not-so-subtly hinting me to move out.
score 10 · canonical
-
deleted by creator
The vibecoding tag was the thing that kept this site tolerable in the deluge of crap that’s subsumed the web. With it gone I don’t really see myself sticking around. What a bummer.
score 2 · canonical
I am new here but I was so happy to find a remaining site focused on the art of coding when the rest of the world is now just “how to tune LLM”. I’m a bit sad that it feels this last dam may fall soon with that news. I have seen it all around me, even with talented folks, their technical skills crumbling because according to their own words it does not matter anymore (even if they initially were in the camp “it’s just a tool”). I don’t mind if people want to talk about how to manage LLMs, which is much more akin to a manager trying to micro manage their engineers in the vague hope they do as they are told. I want to talk about the art of code, even if some believe it’s a lost art. Is this the site for this? I can’t see how the mission statement of the website mentioned by OP and the disappearance of AI flag are compatible.
score 3 · canonical
same. I am hoping there can be an additional ‘tag of shame’ for content not explicitly related to AI/LLM tooling but still in some way contains LLM generated material.
score 8 · canonical
I was also filtering out the
vibecodingtag. I hope we find another way to discourage the constant influx of vibe coding relating news. I probably will stick around as the moderation here is otherwise excellent and the discussion very good. :)
score 5 · canonical
Agreed. I’ll try and keep an open mind while sticking around for a while and watching the actual outcome of this, but I fear the worst.
While not perfect, the tag and it’s usage was the main differentiator between this site and a few comparable ones for me, and this change has the potential to destroy all of that.
score 1 · canonical
I agree, sadly.
I don’t see how repurposing the spam flag to also include LLM output is going to make things better. People will complain the same for being “unfairly” tagged as spam (of course they will!).
In my opinion it will make things worse. Despite complaints, vibecoding-for-all was clear and it didn’t prevent anyone what wanted to read that content from reading it, while “ai-tooling” plus “spam” are going to be misused and the conclusion will be content harder to filter by people that don’t want to read anything related to LLMs.
I feel like there is a missed opportunity here to make everybody happy. If the reason to not use “vibecoding” was “a tag of shame” that it would push people away, what do the moderators think this new situation is going to accomplish?
It took 5 meta posts to get to this change. I don’t see why we can’t have one meta post every week if things aren’t working for everybody.
score 3 · canonical
the conclusion will be content harder to filter by people that don’t want to read anything related to LLMs.
This is why I’ve advocated for the tag to be called “llms”.
score 1 · canonical
To start this off, from an ontological perspective, Lobsters has always been closer to a collection of record store bins versus a library catalog. We have tags which overlap in weird ways, and tend to have fuzzy definitions on overlap. That’s generally been fine, and quite frankly more of a feature than anything else. With that in mind, if the only problem with the “vibecoding” tag was that it cast too wide of a net, we wouldn’t be having this discussion.
We are having this discussion because while the “vibecoding” did allow for community members to filter out content that they did not want to see, it also was a catalyst for other community members to act poorly towards each other. That behavior does not foster a healthy, supportive community.
I will also point out that Lobsters, as a community, has evolved from where it started and will continue to do so. I don’t consider these actions to be an end state.
score 2 · canonical
Thanks for the response.
My impression is that the biggest issue was the negative connotations of “vibecoding”, not it being too wide. If the tag was neutral and someone disagreed, it would be easy to resolve in most cases. Obviously, I could be wrong.
Now lobsters won’t accept submission of repos, despite them having a README that can be enough to start an interesting conversation, and LLM generated content will be tagged as spam, again with negative connotations.
I also liked to read some posts that now are off-topic under the “careers” umbrella. People sharing experiences about burnout, and the responses in comments, helped me personally. I’m sad lobsters is not the place for that now.
And most importantly: it is complicated, IMHO. People looking to use lobsters instead of contributing won’t pay attention, and for those that didn’t want to read LLM stuff and yet enjoy lobsters, things are going to be harder.
Anyway, thanks for trying. The work mods do is very appreciated.
score 1 · canonical
How would you feel about six tags corresponding to the six identified overloaded meanings in the OP?
score 1 · canonical
Some of those meanings are already covered by flags, so you shouldn’t need six new tags.
score 2 · canonical
deleted by creator
An influential person was able to force through a change to the lobsters community by complaining about it on social media.
It’s actually coincidental.
score 1 · canonical
Eh, Valsorda had 137 comments here and around 1800 karma. Respectable yes but influential on this site, no.
score 1 · canonical
The post was drafted before Filo’s departure and it had no influence on the situation. We don’t attempt to recruit or retain notable developers.
The way to be influential here is to contribute or help write good policy, whether that’s insightful criticism of problems, suggesting new guidelines, etc. Not saying anything to mods and then publicly denouncing pretty much rounds to zero influence, especially because taking it seriously would encourage people to brigade us.
score 3 · canonical
Where do you get that from? I see no indication of it in the OP nor in 355E3B’s other comments and more importantly the comment I’ve linked to says this is not the case.
score 3 · canonical
So there’s going to be no way to distinguish LLM-generated blogposts and code repositories in submissions that isn’t opening the link and wasting my time on it? The slop flag or an “this article is primarily AI-generated” would be a great introduction for that reason alone.
score 3 · canonical
I think this is a helpful change. I, like many here, have serious ethical reservations when it comes to LLMs, but an old man once told me “people don’t get better by making them feel worse.” People shouldn’t be made into punching bags. Their shit choices should be called out, slop should be discouraged out as a waste of everyone’s time and attention, but we’ve had slop for way longer than LLMs. Echo chambers don’t make people better, no matter how right you think your side of the room might be. If you’re not listening and thinking, you’re part of the problem. Being “right” doesn’t mean shit if you can’t also be kind.
score 2 · canonical
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
A mod explained their reason for removing that story and another LLM-related story in this comment thread.
score 1 · canonical
Note also that these changes included expanding the ai tag to include development of LLMs. Hopefully this will be obvious and infrequent enough to ignore such posts quickly.
score 2 · canonical
As I understood it it did before too, and this is just a clarification of the description. At least my impression is posts about development/research of LLM already mostly is tagged
ai. It’s just that such posts are not that common.
score 3 · canonical
Yes, but the change did explicitly occur, and the redirect to vibecoding (which now should be directed to ai-tooling, or, in other cases, has no specific labelling anymore) was removed.
score 1 · canonical