The GitHub page is showing “404 not found” as of today.
More information about what happened.
I cannot speak to motivations here, and neither can Kahli right now so let’s not speculate.
I do know that Kahli was upset at me moderating his posts in that topic. He messaged me about it. I asked that he flag any problematic posts I missed instead of escalating with subsequent replies. He responded that he will continue escalating, so I should ban him. I did not ban him; I temporarily suspended his posting privileges on this board for a week. It was shortly after this that the repositories were made inaccessible.
In Kahli’s defense, I understand that his answers can be reactive sometimes, but every single time I saw him escalating, it was in a scenario like this:
- He posts something he developed (on his free time, using LLMs).
- Someone starts to criticize his work due to using LLMs, even though this was happening under his packages and not inside Julia core itself.
OR
- Someone (like me) post something related to AI.
- Another one criticize this post.
This happened many times; it was not once, it was not twice. It happened with me. However, this kind of feedback for Kahli is way worse given that one of his greatest contributions is a MCP server for the Julia language.
I understand that there are errors on every side. But I do understand Kahli’s behavior to be in the side of self-defense. Now, he is suspended, and the other offenders, who hijacked many threads, many times, were not. That’s why he is upset, and I agree with him.
There are a lot of people spending insane amounts of time and making ground breaking contributions. If mods aren’t letting them speak in-character it just creates a stifling forum for no good reason.
i wont comment here on what happened today in that topic, nor am i in a position to criticize moderation decisions, but its sad that we lost important and innovative packages just because of one argument, i sincerely hope that kaimon.jl and the rest will come back 
This is an instance of people locking horns rather obnoxiously (and, not for the first time). This cuts both ways: it’s not okay to harass people who use LLMs, as much as it’s not okay to escalate in response to such harassment. Any community guidelines we have are ultimately about people behaving respectfully, not about technology
Julia lost something truly amazing here today.
The packages certainly have generated a lot of interest, but honestly, the author’s somewhat abrasive behavior in interaction with me and other members of the community has made me avoid trying out the packages, despite having an overall positive view of what LLMs can enable. The point being: the social aspects of an ecosystem are at least as important as the technical aspects. I generally don’t think that just because someone makes valuable technical contributions, that means that poor behavior should be tolerated. Whether in open source, academia (where this is a long-standing issue), or a professional environment.
In any case, I think a knee-jerk reaction of deleting package repos because tempers are running high is not okay for a registered package: Registration in the General registry creates an obligation towards the community. But that’s maybe something to follow up on once this discussion has cooled down a bit.
I also find that, generally, absolute positions are always both wrong and unhelpful. This applies both to “AI is inherently evil”, and “AI is the second coming that has ‘solved coding’”. If you find yourself holding either one of these positions (or any absolute position on any topic), I would recommend further reflection, and restraint in how you voice your opinions.
So, let’s all just chill out a bit
I spoke with Kahli today on google meet just before this happened. He strikes me as a deeply passionate and caring person. He is abrasive, but really only because of how much of himself he puts into his work. I’m not gonna share everything we talked about but i will say he did not seem to be a bad person. He’s very, very passionate about what he’s been doing.
With Fons stepping away and mentioning him getting hate for Pluto.jl, i’m afraid this may be a more extreme case given the temperaments of those involved. It is a shame because what Kahli showed me was truly mind-bogglingly cool and ridiculously impressive.
If you are not including Kahli’s packages in that statement, please say so explicitly. If you are, that is simply not true. Yes, using AI—and computers in general—tremendously speeds up development. That does not mean everything developed with AI - or computers in general - is low effort or low quality. Kaimon and Tachikoma are definitely not.
One-sided obligations?
Yeah, was trying to evaluate Kaimon.jl when this happened (needed to understand better what it offers in respect to general coordinators as one can call them), then I knew about what happened, and I have to say didn’t expect that since I’m generally untouched by people critizing me, even more if I don’t believe in their arguments, but probably he is on the other side of the spectrum. Hope he can find a way back
I respectfully disagree with this decision, both in general and in this particular case.
We are all thoughtful, well-meaning, well-educated people. In the rare occasion discussions become heated, those who are not interested can simply choose not to participate or to ignore the messages/thread.
How the community feels about this?
- Please remove the suspension
- Keep the suspension
- I prefer not to express an opinion
0
voters
One-sided obligations?
No, not at all! Certainly, the responsibility of the community towards the maintainers of registered packages (well, anyone) is to respect their work and dedication. I don’t think harassing Kahli with anti-LLM ideology is acceptable in any way. The appropriate response to such harassment is to report it to the moderators.
In a situation where someone announces that they’re looking to get into a fight and actively baits to be banned, a cool-off period is the only reasonable response. Or rather, the more measured response compared to doing what they’re asking for and banning them outright.
No, I’m explicitly not including Kahli’s packages in that statement! The packages are 100% fine, see also Policies around registering AI-created packages in General? - #175 by goerz where I had to repeatedly make the point about there being no problem with Kaimon and related packages being registered.
Update: I edited the original comment. That was not very well written on my part, since (as you picked up on), it implied that Kahli was pushing low-quality LLM content, which isn’t the case (and never has been). Let me try again:
I also disagree with this decision. I find it super weird that we are having a discussion about someone from the community here without them being able to reply or defend themselves.
But you keep coming back to “Kahli’s communication style”. I wonder if you worked with him on his packages or was a meaningful user of Kaimon.jl, to continuously bring this up. This, and the unnecessary heavy handed commenting from mods, can certainly feel like online harassment on someone that devoted so much passion and time.
If you don’t like a post, hide it, don’t need to point out how much the poster is morally wrong to say what he thought.
My understanding is that they themselves asked to be banned and that is in fact the listed reason for their suspension. I am unaware of a request from them to lift the suspension. As far as I can tell, the primary decision came from them. That said I think there is a need for some time for reflection.
The basic terms of use of this forum are governed by the Julia Community Standards. If those standards are being violated, the proper response is to report those violations as detailed in the community standards. Retaliating in violation of those same standards is counterproductive.
There is clearly a need for a more structured approach to the topic of AI use. We cannot tolerate continued harassment occuring in either direction. My understanding is that is being worked upon.
While what has happened here is tragic, I hope that something useful can come from this.
Well, in my view, it is pretty one-sided. “To respect” doesn’t actually mean anything.
“F” in “FOSS” stands for “free”. “Free from obligations”, among other things, except otherwise specified.
As we Brazilians say, “let’s not try to block the Sun with a sieve.” Acceptance is the first step to correcting an error.
The attitude toward Kahli was precisely what led to the current situation. I just cannot understand how anyone can find normal in someone announcing a package he developed in his free time, and willingly provided gratis to the community, to be harassed with LOTS of posts, frequent posts, saying that this is AI slop, slow down your AI enthusiasm, etc.
Notice, we are not talking about Julia core, but a personal package!
Yes! In all cases, Kahli reacted to people not being respectful to his work. Every single time, Kahli reacted, not started.
He can say the same about you. I remember that your first interaction that I saw in the Tachikoma.jl announcement post was to ask him to slow down. Notice, a package he was making available for free and that he could have kept to himself.
Worse is people that are not making ANY contribution whatsoever to a package criticizing the work of another person in this package that is generated by free and goodwill.
And here is my criticism to the moderators: I saw very few interactions (and no suspensions) when people hijack threads of package announcements just to criticize the tools one is using. Why are those people not asked to just close the browser and move on? Why they are not also suspended by frequent behavior?
Now, my point of view. We (Julia community) might have lost a great programmer, with a very good vision. Tachikoma.jl is/was a great tool, comparable to Bubbletea in a moment that TUI tools are becoming very popular. Now, check how many devs are behind Bubbletea. What Kahli has achieved was impressive to say the least.
Coincidence or not, three very important maintainer steps out from Julia in a very short amount of time: Kahli, Pluto.jl’s maintainer, and JuliaFormatter’s maintainer. The two first clearly had problems with the community. This is a very big problem.