On the closure of the "On LLM policy of HaikuPorts" thread

Feedback Websites
41

This reply belongs on the HaikuPorts mailing list, not here, as the discussion is happening there. Note my reply was merely clarifying that the thing you and others seem to be worrying about is not being discussed at the moment, and not advocating one thing or another (I am indeed advocating many things on the mailing list, but not here.)

1 Like
42

You are misunderstanding it. The point is to make OS itself simple and efficient to give more resources to user payload. For example BeOS pushed hardware to its limits with real-time media tasks such as audio/video encoding/decoding/mixing. LLM processing fits perfectly to this concept where OS do not waste resources for itself and give maximum efficiency on running LLM training/inference using all available resources (CPU, GPU). In is in spirit of Media OS.

Of course nobody tell to preinstall LLM models to Haiku, but users are capable to run them if they want.

5 Likes
43

Since my (now closed) thread on the forum resulted in a lot of activity on the forum, I think I should share some thoughts on the problem:

  1. The only defence agains slop, any kind of slop, is having a taste
  2. LLMs allow to produce slop in industrial quantities
    1. And it takes no effort to do so
    2. Using LLMs for autonomous programming (“vibe coding”) ensures producing slop
  3. There are legitimate use cases for LLMs that do not result in producing slop
  4. If code is good, it does not matter how exactly it was produced (modulo copyright)
  5. If code is slop, it does not matter how exactly it was produced
  6. Obvious slop should be rejected on spot
  7. Contributions of a person who keeps submitting slop can not be trusted
  8. In contrast, if a person keeps submitting good contributions, he deserves some trust

I believe that solution proposed by @pulkomandy in Create pull_request_template.md by pulkomandy ¡ Pull Request #14201 ¡ haikuports/haikuports ¡ GitHub is adequate to the problem. But a full ban as proposed by @waddlesplash is not.

That’s a shame, you’re the one with common sense.

4 Likes
44

+2 from me here!

This already came up a few times when local checks failed and asked for some kind of build log, I think this is also a good idea following up on the comment above.

4 Likes
45

Congrats on chasing away a leader of the community. You know who you are: the people who insist on shoving your crap sideways into everything after you’ve been told repeatedly that it’s not welcome. Even if it wasn’t for all the technical and ethical issues that y’all refuse to acknowledge, it’s the attitude.

We had just made a decision. Together, as a group. But y’all couldn’t be happy with it, because everything has to be about you, all the time.

This isn’t even about slop, or moderation. It’s about who we want to be. Well?

2 Likes
46

Did you read the original thread though? I explicitly asked to remove my contributions if they are not welcome.

1 Like
47

Thank you for proving me right on every account, but please stop playing the victim. Nobody wants to silence the conversation about these issues. Besides: you were given a space for discussing LLM-assisted software all you wanted. Except it’s not enough, is it? Never enough.

1 Like
48

Um can we add a longer timer like a 50 minutes timer because it seem like things are starting to get heated

49

I’m a honest man and demand honesty from others. This makes some people uncomfortable.

If my contributions are not welcome, they are to be removed. I will submit a PR removing opensmalltalk-vm from haikuports as soon as any haikuports contributor will say my contributions are not welcome.

If my contributions are welcome then this fact contradicts the rule that any LLM usage is not allowed.

1 Like
50

You have misinterpreted what happened. A specific user (Atomozero) has been trolling us. He created an insulting thread and got the ban he deserved. He then went on to brag about being the rebel in the Haiku space and being banned for saying out loud what other people don’t dare to (he didn’t, he just trolled and insulted people). The ban was for 30 days, but he didn’t use these to reflect on his behavior, then came back and did the exact same thing again.

What else can we do than close that topic?

No actual topics discussing LLM policies have been closed. We would rather have a discussion in the mailing list for Haikuports, because, as far as I know, that is still the official place for Haikuports discussions. Maybe that rule is out of date, but it’s the rule. We can have a separate discussion (on the mailing list) about officially moving things to the forum.

So, I don’t think the rift comes from moderator actions. From strong disagreement on the policy about LLMs, maybe, but that debate is doable. As long as there is not a troll insulting people in there…

6 Likes
51

The trolling you mentioned didn’t start in a vacuum. It wasn’t the cause of this rift, but a direct consequence of the frustration built up around these heavy-handed moderation decisions. When people feel procedurally silenced, sarcasm and satire are often the only tools they have left to react to absurd situations.

Banning users for jokes or sarcasm is a slippery slope. It feels like people have become too fragile and easily offended lately, always looking for a reason to feel attacked.

If we want to stop the trolling, we need to address the root cause of the community’s frustration, not just ban those who react to it with humor )

8 Likes
52

That’s not how it works, you can’t behave inapropriately first and then go around claiming that you acted inapropriately because you got banned afterwards. The moderation is not a cause, it is a symptom of rude and antisocial behaviour.

Yesterday alone there have been 5ish people on irc who told me they do no longer wish to use the forum as a direct consequence of andrea’s behaviour, and that we have so far tolerated it (by not banning him completely), one even asking for their account deleted.

3 Likes
53

Can’t speak for others, but for myself it isn’t just the behaviour of Andrea, it’s the reply from others too …

1 Like
54

Andrea only reported a question I asked on Telegram, finding your decision to distort the forum by favoring one channel over another. A moderator should close duplicate topics on a single medium, not across different media. A user might not intend to subscribe to a mailing list or consult a service with a terrible interface and full of ads, like freelist. Furthermore, it is interesting to note that on one hand it is justified that if a discussion starts on a channel it must continue on that channel, without considering the purpose of the OP of the closed thread, who simply self-reported asking for the removal of content probably not to your liking. My question was not followed by a response on the same medium, Telegram, but you preferred to answer me through a closed thread, evidently to prevent me from replying. This behavior drives users away from the forum, sends them to other channels, and prevents them from participating.

3 Likes
55

Some kind of childish nonsense :sweat_smile: :rofl: :joy:

56

To be honest, I’m personally quite happy with current moderation team work.
Sure, things could be improved but at what cost.
IMHO the only way that a moderator can’t be accused to enforce their opinions is that they are not participating to discussions. Therefore, they would be neither a dev nor a user and never become one of them. That means that you would have to pay for this person or these persons. Since you don’t them to be accused to enforce Inc views either, the bill would have to be paid by forum users. At the end, people would complain that the forum isn’t open to everyone and that they would prefer to see that money spent to pay another developer. This wouldn’t make sense.

57

So the dev story is atom zero is Andrea? And he was causing problems and keep trolling then ran to the internet to complain is what I’m hearing from the developers and then he now starting a another thread because he didn’t get his way again, and got banned again then he currently made another thread(this thread) to make a new change that was unnecessary.

Ai assisted user are just talking about all decision should be made in forum not on a separate mailing list or Then people who hate ai think this thread is about ai

With people throwing around random words like fascism without know the direct meaning behind it is bad

So this is what I’m getting from this thread and that mean everyone is confused

58

On Telegram we say the same about you… and we are more than 5

1 Like
59

This thread is very confusing: most are ignoring the content of the original post, focusing on the OP for their recent past, with users hallucinating — truly ironic — believing it is once again a post about LLMs.

2 Likes
60

Indeed, but it is not clearly identified. “Proprietary and other software”? I could swear it was called something else the first time I saw it. If we created it specifically to get a ceasefire in the LLM Wars, why isn’t it in the name?

It may not sound like a big deal, but I foresee that for years we will see newbies posting their first app only to be snarled at to “put your slop in Proprietary and others!”

3 Likes