I was generally reminding all people that this thread is about the moderators closing a topic stating their reasons in a last post. It’s not about if LLMs are good or bad, or if it’s too much to ask for labelling of AI generated software.
1 Like
This is where I think the conversation has not landed yet. The HaikuPorts mailing list has a substantially smaller and more self-selected participant base than this forum. When discussion about project policy migrates from forum to mailing list, what migrates with it is not only the topic but the pool of people who effectively shape the outcome. That can be a legitimate operational choice for several reasons, but it is a choice with consequences for representativeness, and worth naming as such, independently of any single decision.
1 Like
The justification was right there in nephele’s comment. OP’s concern that I share is the locked thread keeps receiving comments from the moderators without anyone being able to answer. What are these comments for? To justify the closure one more time, because first one was not enough? Or to talk about moderation and flagging? Whatever the idea was there it looks like the resulting effect on the community has been negative so far.
You added another comment there about some other closed topic. Now, I’m curios to read that first closed topic or ask if it was deleted afterwards, but where should I ask you about it? I cannot reply there, so should I post my question here and start the off-topic discussion, or should I start yet another thread?
Don’t you guys see how those comments-after-close bring unnecessary friction and turbulence? If you really need a moderator announcement thread, maybe consider creating one.
4 Likes
Oh,another month,another drama 
Don’t these pro-LLM guys have anything better to do than trying to push their opinion on everyone?
The dedicated forum category for slop is really great,makes it super easy to mute that nonsense.
Unfortunately this one escaped the echochamber and here we go again,discussing the moderating decisions of voluntary moderators because they tried to redirect a discussion to the place where it belongs.
This doesn’t serve any other purpose than trying to split the community,does it?
I can’t tell you how happy I am not to be a moderator or someone doing voluntary work in these crazy times.
Speaking about frustration,as 3dEyes noted…
I actually see some similarities to my personal viewpoint,but the outcome is very different.
Haiku has also always been a breath of freedom for me,freedom from proprietary garbage and Big Tech monopolies,from ads and trackers and enshittification.
A small isle where things were done differently,more quality-focused and human-focused,not with profit or marketshare or the latest trends in mind.
I’m unfortunate to live in a decade where the human being isn’t really worth a lot anymore.
In many cases,humans aren’t the customer anymore,but the thing they sell to advertisers.
Things aren’t built anymore to make people happy,but to squeeze as much money as possible out of them.
At work,humans are only seen as a cost factor that needs to be removed,ideally replaced by some dumb machines that are cheaper and don’t ask questions.
Yeah,really not great times to be alive as a human.
Having a small project that still values human work and the idea of technically good software to escape from a otherwise crazy and awful world has been great,but now some people try to force corporate bullshit right into my face and don’t understand a “No”.
There used to be a general consensus on ethics that wasn’t written down but still clear to everyone,like not stealing other peoples work or not DDoSing other peoples servers.
No,I’m not going to iterate all the negative effects of LLMs over again,that would be pointless and it’s absolutely clear that some people simply don’t care at all,but it should be noted that the “tools” under discussion are actively harming the infrastructure of Haiku and other open-source projects and the social contract of FOSS licensing.
Having to discuss the obvious over and over again with the other side just stating “so what,I don’t care,and others are doing it too” is really,really frustrating and makes me question why I’m still contributing to open-source.
I’m sure some people would like to see me gone,maybe they’ll be happy in the not too distant future if things continue going downwards.
Speaking about moderation,I sometimes find this forum not strict enough.
There are people comparing this place with a gulag or fascism,*what the ***?!?!
Let me tell you,you’re free to leave if you hate this place so much.
I think that the moderators should remove such tasteless insults against the whole community and the project.
In my opinion,the forum and the moderation policies should benefit the whole community and the project as a whole instead of letting every single person speak out freely while driving away others that prefer not to read such drama and offensive language over and over again.
This once was a place for constructive and peaceful discussion and a place where users help users,it should really become that again.
If that means that some people deeply unhappy with the choices of the current maintainers leave and create a LLM-generated fork,so be it.
7 Likes
To be clear, this thread is not about LLM policy, and I have not raised the merits of that policy here. Both my opening post and the reply to waddlesplash stayed on moderation form throughout.
Agreed. A short procedural note such as “closed, please continue on the HaikuPorts mailing list” is reasonable, and the closed thread is the natural place for it. That was never the objection.
The asymmetry I pointed to was not justification as such, but a substantive, multi-paragraph defense of the moderation decision posted after closure. That goes beyond meta-information and effectively secures the last word on the moderation question in a venue where no reply is possible.
So the fourth option missing from your list is the one already in use in this very thread: brief procedural closure notes inside the closed thread, and substantive meta-discussion in a separate open thread when warranted. I am not asking that moderators be unable to close threads, nor that their decisions be made unaccountable through endless re-litigation. Only that, when the justification grows beyond a short procedural note into a substantive argument, it should be open to reply somewhere accessible. Which is what is happening here, and seems to be working.
2 Likes
As humdinger pointed out, this thread is strictly about moderation mechanics, not whether LLMs are good or bad. And I’m not trying to be rude but you pasted this big wall of text that so incoherent to read and completely missing the whole point of what was being said.
Nobody here is trying to force corporate tech into your face. We are discussing a basic procedural question When a thread is locked, should moderators be allowed to post an unanswerable final argument, or should closures be strictly neutral system notices?
Also I fully agree with you that wording of gulag shouldn’t have been mentioned and it completely out of pocket and fascism(I don’t know if this shot at me) but all I said was it stratocracy which means a military government in this case moderator are running the forum as a military which leave less of democracy for the people in the forum not to mention a lot of people don’t have access to the same mailing list or same type of stuff with the forum being the only way to talk openly about weather a decision should be made on the forum as that is the most accessible thing for people to type on.
And I understand your angry about LLM but you posted this big blob of text that missed the point
1 Like
Just to be clear, here are the two messages that this entire topic is about:
From Waddlesplash:
I have been informed that some people (on the Telegram group) are complaining about nephele’s moderation decision here. Well, I can tell you that if the other moderators disagreed with this decision, we would have reversed it. (Indeed, even before I made this reply, someone reported nephele’s post, and another moderator besides nephele or myself rejected the report.)
And then from me:
I will add that this is a re-start of a previous topic that had already been closed a few hours earlier. Please don’t do that again. When we close a topic, it’s not to have a very similar one pop up a few minutes after.
Just clarifying a moderation decision since there was contestation that it was a single moderator going rogue.
This gets us accused of fascism, of being a gulag, and whatnot.
I had enough. I am resigning from my moderator and admin role. Find someone else to run this forum. I wish them good luck.
1 Like
Genuine question: is this the same person who described this situation very differently on Facebook a few days ago? Because the tone gap is… significant.
You raise the point of asymmetry — but that could apply to tone as well, depending on which space you’re writing in.
1 Like
User(s) in a different discussion group seemed to think the decision was nephele’s alone and were blaming him for it. So I chimed in to clarify that this was not the case.
What is “taboo”? We haven’t actually declared, anywhere, that any particular topics are off-limits. Certain topics get redirected or merged when they get too heated, this is true. But there are no restrictions on debating LLMs, at present. It’s exhausting that they come up so much, but in fact we haven’t made any rules against it.
The community has changed, yes. But the community changed because the world changed. The technologies causing all this division functionally did not exist, or were not very widely available, a year or two ago. So I can very much understand wanting to go back to the way things were… but how can we do that? The closest we could get is by banning LLMs from Haiku entirely. Which is more or less the position I and others have been advocating for, which you don’t like.
I agree, this is all very frustrating. But what are we to do here? There is no way to go forward without something changing, because we live in the world, and the world has changed.
Monopoly? We are having an argument in this very forum. (And also the one on the mailing lists.) Do you see opinions getting “censored”? So what are you referring to here?
The most “censoring” I’ve done so far was to remove some swear words from a post. I think terms like “gulag” and “fascism” are not the best, but I understand that people are feeling very strongly about these things, and so they are going to throw around emotionally charged words. Personally I don’t have a problem with that, though I also understand it makes other people uncomfortable or angry. So how are we supposed to moderate that? If we leave it up, some people are deeply unhappy. If we remove it, other people are deeply unhappy. Either way, we lose…
Open source projects basically never are “democracies”, they’re “republics” at best. Haiku is no exception to this. Those who have power/authority ought to use it for the good of the community, and not be tyrants, which means listening to the community, attending to their needs and wants, and taking all that into account when decisions have to be made. But at the end of the day, decisions get made, and not by taking it to some democratic vote, but an “aristocratic” vote at best. Basically any open-source project works just the same way.
The entire question here is how we should develop. If users had no opinion about how we should develop Haiku and just told us to do “some developing” however we chose, then the flamewars would be much smaller, because they’d only be between programmers.
What you are advocating for here would then lead to moderators and developers closing more discussions, and creating a subforum where only programmers could respond (and everyone else could only read, at most). I think I’d actually be fine with that. But would everyone else?
2 Likes
You’re right I should’ve did more research, I’m inexperienced in this topic and this will be my last reply, ima gonna do more research on how open source communities work and it completely my fault for speaking with no knowledge on the topic.
I hope you have a great day and thank for educating me
This is really really weird interpretation of 3deyes words. The point is that Haiku gave a sense of freedom and freshness. People freely shared new ideas, made contributions, ported software they use, made native software. And then suddenly new technology appear: LLM and AI agents. Instead of getting happy that out childhood dreams about machine understanding human language and doing things that we told it realized, about what 20-th century sci-fi books and cinemas told a lot, it suddenly got rejected. Haiku turned from BeOS-inspired future dream to some conservative innovation-rejecting thing.
NVIDIA give a present of portable GPU driver, so we can run hardware-accelerated LLMs on Haiku, isn’t it great? Why reject progress and cool things?
LLM is just yet another technology, I see no reason of considering it as turning point and prohibiting it in any way.
12 Likes
Because, as it turns out, it doesn’t understand human language. It’s not the thing from the sci-fi books. That’s one of the biggest reasons for the rejection.
Not all innovations are good. Haiku has always been “conservative” in some ways, for example in UI design most obviously, and in preserving ABI compatibility with an OS from 1998, those aren’t the only ones.
In other ways, indeed we are quite innovative and adopt new technologies quite rapidly. But we don’t always; there are any number of things that we rejected as irrelevant, unnecessary, or not worthwhile, which didn’t blow up in the past because they either faded out of relevancy or were just minor issues in the first place.
The LLMs controversy seems to be the biggest controversy we’ve had since the package manager, and at this point I’d say it’s bigger even than that. That drama had a lot of fallout and some people left the community permanently over it. I don’t think there’s much chance the LLM drama gets resolved without the same thing happening, unfortunately.
2 Likes
This is touching of my view on the subject. BeOS was always about state of the art, inventing and being legacy free, while Haiku seems to be more about just preserve the legacy only.
10 Likes
I’m not going to participate in yet another drama about LLM with the exact same LLM-lover telling me the exact same bullcrap. Don’t be fooled; his point is not the so called “assymetry”, or mailing lists, or whatever. That only serves as an excuse to open yet another thread about… but the holy LLM, of course, what else?
I’m just saying I countersign @nipos 's post above, 1000%.
1 Like
I guess we simply don’t have the same childhood dreams. I think I was exposed to slightly more dystopian sci-fi?
Haiku was a flawed project from the start, from, he moment it decided to have binary compatibility with BeOS, then. There have been several other projects going wild with innovation keeping that part of the BeOS spirit alive. They’re all dead now, because not having a stable platform also means having no users. I think Haiku has a good balance, between providing a stable platform to run apps, and innovations such as the package manager, the aukland layout system and stack and tile, and many more of a somewhat smaller scale. There is also a lot of room for experiments outside the core OS thanks to its stability (lately the obvious examhle would be the Sen semantic desktop, but I’d say x512 work on the risc-v port and on the nvidia drivers also qualifies).
Is this an unreasonable development model? After all, it is how innovation worked in BeOS days as well. The translation kit started its life as a 3rd party dev implementing something similar to Amiga datatypes libraries. The gcc port was made by Fred Fish independently of Be inc. And probably many more things. Personally, I never got to know BeOS during Be inc’s lifetime. When I used it, it was already a frozen system, with support coming from 3rd-party patches and apps. And, indeed, in Haiku this is even more so, when the mission of the project is “preserving and continuing the innovative legacy of BeOS”.
But another imhortant element of BeOS, besides its innovation, is it’s simplicity and efficiency. You could almost call it a precursor to the current trend of permacomputing. LLMs at least in their current implementations certainly are not compatible with that.
6 Likes
I see that my words about the “Gulag” have been completely misinterpreted. Perhaps you are not familiar with the established term “Digital Gulag” — it was a metaphor about excessive control and censorship, not a literal comparison.
You are fundamentally misinterpreting my stance if you think I am here acting as an “LLM lover/activist”. That is not the point at all. If you started discussing a total ban on all GTK-based software in HaikuPorts, I would be saying the exact same thing.
My personal opinion is that GTK, GNOME, and much of the software written using them is awful. It’s often spaghetti code, hard to read, and poorly written. But! I would be absolutely, categorically against banning them in HaikuPorts. My personal distaste for a technology hasn’t stopped me from porting a huge amount of software that I personally dislike, because ultimately, the user should decide what they want to run, not me.
The real issue here is that some moderators on this forum are projecting their deep personal hatred for LLMs onto their administrative duties, putting a spoke in the wheel of any discussion they personally dislike. Closing threads, banning topics, and silencing dissent based on personal technological preferences is fundamentally wrong. This is exactly what is causing the rift in the community, not the LLMs themselves.
A repository policy shouldn’t be dictated by the personal phobias of a few individuals.
16 Likes
just throw an “I used an LLM to write/augment/debug this code” button to haikuports and lets be done with it. I don’t feel like these threads, as cathartic as they may be for everyone, are doing anything but dividing the community. they feel good but leave damage behind. like heroin. speaking from experience there.
I’m not the most active person on this forum, but there are too many valuable community members getting digs in at each other. familiarity can breed contempt, sure, but I don’t feel like there’s any truely valuable discussion happening here. just bad faith misreadings and subtle digs when the technical issue at hand can be a button.
I do a lot of development with the help of LLMs, but I’d never try and hide it. just make it a button and lets get back to haiku. just my .02.
2 Likes
I have to say that I really haven’t seen this.
I can think of 1 moderator who could be said to have a bit of a hair trigger on this topic, though I think their decisions have all been defensible on the basis of forum policy.
I do not think heavy-handed moderation (to the extent that it can even be argued to have happened) is responsible for the tension here.
There is a fundamental split here over to what degree LLM-use is acceptable; the OS developers are split, the external developers are split, the porting group is split and the users are split. None of the Haiku sub-communities agree within themselves over it.
And this is a bad situation, but it can be resolved without anyone being driven out if we want it to.
But that involves compromise.
Those of us (yes, including me) who would prefer that LLM-generated material just be outright banned in any Haiku-official space will have to accept that the community overall will not support this stance.
But those who would prefer that LLM-generated material be accepted as normal and without comment will also have to accept that the community overall will not support that stance.
The only effective middle ground I can think of is labeling (a responsibility for the pro-LLM folks) and filtering (a responsibility that falls on the anti-LLM folks).
Maybe someone else has a better idea; I hope so, because the labeling/filtering approach is also not getting widespread support at the moment. 
(But this is not the thread for that conversation.)
The decisions have been made for two Haiku properties:
- the OS itself forbids LLM-generated code
- the forum has set aside a space where LLM-generated code is expected to be identified as such, but where it also is expected to not be challenged
The HaikuPorts LLM policy is being discussed on the mailing list thread at https://www.freelists.org/post/haikuports/LLMs-policy.
That is the place to make a contribution to the HaikuPorts policy.
6 Likes
To be clear, the portion of the discussion about what software should be allowed in HaikuPorts has been dropped, at least for the moment, and the remaining question is whether to allow LLM-generated contributions to HaikuPorts itself: recipes, patches, etc.
@waddlesplash I completely understand the fear of a PR flood, especially without pre-build automation. To protect reviewers, we can implement two simple, quality-focused rules:
- Any new recipe PR must include a build log or a screenshot proving that the package compiled and ran successfully on the contributor’s local machine. If there is no proof, the PR is closed.
- Since our reviewer resources are limited, maintaining high quality must be the top priority. If a contributor repeatedly submits broken or untested recipes and shows no effort to fix them, we should reserve the right to restrict their PR access.
Ultimately, it shouldn’t matter who or what wrote the recipe - what matters is how well it is made. Explicit bans on LLM-assisted code are practically unenforceable anyway, as very few people can reliably distinguish between poorly written human code and poorly written LLM code. Why waste our limited time trying to police the tools, when we can simply judge the quality of the end result?
9 Likes