The “vibecoding” tag has been a bone of contention on lobste.rs since it was instituted in April 2025. Originally described as “AI/LLM coding tools; don’t also tag with ai”, it was gradually expanded by the user base to apply to anything remotely connected to GenAI use. For some, it was a neutral marker that allowed users not interested in reading about GenAI in relation to software development to filter out submissions so tagged; for others, it was a literal badge of shame stamping their hard prompting work with disapproval.
Various fixes were proposed, at least 5 contentious “meta” discussions held. The final one, It’s enough - vibecoding tag crossed from useless to harmful resulted in an official statement Updates to site guidelines and AI/LLM tags authored by mod 355E3B. In it, the following changes were made:
- the “vibecoding tag” was retired
- the “ai” tag was retitled to “Developing artificial intelligence, machine learning, LLMs.”
- a new tag “ai-tooling” was introduced, titled “The design, ux, development of AI/LLM-powered tools.”
Further official comments by mod pushcx clarified that any discussion regarding whether a submission was primarily authored with the help of GenAI was to be seen as off-topic “purity-testing”, and removed.
So if we designate the pro-GenAI faction within lobste.rs as the Royalists and their opposition as the sans-culottes, it seemed to be a clean sweep of victory for the Royalists.
However.
In the following days, some worried Royalists could see that submissions that touched on GenAI use in a positive way were removed by mods with the reason “Article is about (not) using LLMs.”. A clarification by 355E3B confirmed that
The litmus test I’m using for this tag is if the body of work still makes sense if I were to replace any instance of LLM with consultant and it still hold up, it’s more likely than not outside of the bounds of topicality for the site.
So “ai-tooling” was not, in fact, a straight-up less derogatory replacement for “vibecoding”. In fact, a careful reading of the tags showed that anything outside the definitions
The design, ux, development of AI/LLM-powered tools OR Developing artificial intelligence, machine learning, LLMs
is, in fact, off-topic.
At the same time, mods have aggressively removed any comments with even a hint of notice that a submission was authored with the help of GenAI.
Also implemented around this time was a subtle change in how users can “flag” a submission, thereby both alerting the mod team to potential infractions and lowering the submission’s effective score, and thus visibility. Previously, to lower the score, a user had to both flag and hide the submission. Hiding a submission prevents it from turning up in most views on the site. Crucially, it removes it from the /comments view.
Now however, setting a flag automatically hides the submission. Flagging is therefore a much stronger signal than before, as well as being a more final “I don’t want to see this” move. A submitter might be surprised or upset that their submission is flagged, but any followup comment to this effect will more likely than not not be seen by the flagger, thus removing a possible conversation that was in fact possible when a quorum of users had suggested the old “vibecoding” flag.
All this points to a situation where the mod team is aiming to enforce quietude by enforcing rules against both extremes while discreet policemen and informers wield the denunciations of flags against the Republic’s enemies.
Who knows what the future brings. Hopefully not a new Thermidor.
