RE: https://mastodon.tetaneutral.net/@pulkomandy/117191319412476681
For anyone bothered by the "LLM deluge" in the world and in software development (or for anyone who doesn't understand those of us who are bothered by the "LLM deluge" but sincerely wants to understand), I highly recommend giving this post a read. An unpacking, an elucidation, and a lament all at once.
I don't know if I have something useful or interesting to say, this is probably a bad blogpost written without enough time to think about it, and badly organized. I may or may not rework it later.
🎉 Haiku R1/beta6 has been released! https://www.haiku-os.org/get-haiku/r1beta6/release-notes/
And now, well, everyone walks around with little mini "seeing stones" in our pockets. And to most people, the functioning of these devices is so far beyond their understanding that it may as well be magic. Even those of us who work in tech and can explain the principles behind each step of their operation, couldn't make one ourselves, and some part of the manufacturing processes will probably be opaque or even "magical" to us too. Isn't that what Gandalf is warning about?
Because the palantiri weren't made for evil! They had legitimate usages! But Gandalf wasn't just saying that they're dangerous because Sauron has one and can deceive you with it, he's saying they're dangerous because they are far beyond anything they were capable of making or doing at that time. Even the wizards. And that made them dangerous to use even if Sauron didn't have his hands on one of them.
RE: https://mastodon.online/@danirabbit/116977287188735133
So apropos of this. In TWO TOWERS, after Pippin touches the palantir and Gandalf is taking him along to Gondor, in the books they have a long conversation during the journey. (The movies understandably had to cut it for time.)
But in that conversation about the palantiri and their history and power, Gandalf says this line: “Perilous to us all are the devices of an art deeper than we possess ourselves.”
I think about that quite a fair bit.
In case anyone's curious what a guarded heap fault looks like on Haiku, here it is:
Months ago, in Haiku, I refactored the kernel's debug malloc()s to be runtime-swappable & configurable on nightly kernels instead of needing a rebuild. Thanks to that, a kernel panic reported today was traced to the source in just minutes, instead of needing multi-hour recompile-reinstall cycles. Working as designed!
(Now if only I'd thought to do that refactor years ago, I'd have saved myself and others a fair number of spent hours, at least...)
https://blog.lauramichet.com/thoughts-about-interfaces-acting-like-people it's honestly not hard to imagine a world where people are legally forbidden from selling software or hardware that refers to itself in the first person or is referred to as a person by any supporting materials (eg marketing), and where the intent embodied in such software is always explicitly attributed to the human beings who produce it and set its policies. i don't think that would fix everything but it'd help. that's a moderate position IMO.
I'm going to try and host a Haiku "lounge" / "office hours" open conference call tomorrow (Friday, 05/29) at 2:30pm EDT (8:30pm CEST.)
There's been a lot of contention in the community about "LLMs" lately, so I expect that will come up or even be the dominant topic of conversation (and hopefully this might help clear the air a bit in a way the forums haven't.) But hopefully we'll talk about other Haiku things, too!
More details & stream link will be posted at: https://discuss.haiku-os.org/t/office-hours-lounge-group-call-friday-05-29-2-30pm-edt-8-30pm-cest/19293
Of course, much of the ("developed") world now seems committed to using "AI", so dealing with people who are using it (or who want me to use it) is basically unavoidable. But having to live alongside people doing grievously unethical things (and who brush off prompts to consider their unethicality) goes back practically to the dawn of humanity, I'd wager. So that's nothing new, though its oldness doesn't make it much less uncomfortable...
(The philosophically astute may notice there is, logically, a third possibility: that "intelligence" is not real but purely illusory. In that case, all basis for morality collapses into nothing, and one's left only with raw power and its totally unrestricted exercise. I find that idea to be so ludicrous, not to mention even more abhorrent, that it's not really worth considering.)
When it comes to so-called "AI", there are only two real possibilities:
1. It's not intelligent.
In that case, machines are being used to produce things (e.g. art, code, words) that up until five minutes ago were only ever produced by humans, with intentionality, meaning, purpose, and other things non-intelligences can't supply.
2. It's intelligent.
In that case, it's enslaved to an almost unimaginable degree.
Both of those sound abhorrent, and I want no part in any of it.
This month in Haiku: a lot of bugfixes and little improvements… oh, yeah, and the ARM64 port boots to the desktop (in QEMU, at least)! https://www.haiku-os.org/blog/waddlesplash/2026-04-13-haiku_activity_contract_report_march_2026/
@icculus I see your question from last night on OFTC#haiku. StartWatching() takes a BMessenger, so anything that can receive messages will work. Simplest case is a BLooper, which will be its own thread, but if you don't want to spawn another thread you can probably create a port and then poll it now and again from the same thread.
Another month, another @haiku activity report... but this time, I wrote it from within Haiku itself, using the newly available Go port to preview the website live! Read it here: https://www.haiku-os.org/blog/waddlesplash/2025-12-12-haiku_activity_contract_report_november_2025/
linux is like if text based adventure games were an operating system
@leoluk @GossiTheDog For the most part, none of these packages have any need for new versions to appear in anyone's builds for *months* if not years after publication, unless someone *specifically* has read the changes to the new version and sees a new feature they want from it.
LPM platforms should be designed around this basic principle that updates are mostly unwanted.
The only way anyone should ever get unexpected updates is if there's a serious security problem, in which case there should be a description of the problem and a small comprehensible patch prominently displayed.
IOW LPMs and similar platforms should behave like Debian Stable.
You may now add me to the list of FOSS folks directly impacted by unethical AI scraping effectively performing Denial of Service attacks.
My wife informed me this morning that our billing system had been knocked offline.
The reason? #Amazon, #OpenAI, and similar bot scraping traffic blew up my access logs to the point of filling that server's entire drive. They're constantly scraping and re-scraping my FreeBSD wiki.