The Linux Thread - The Autist's OS of Choice

  • 🏰 Deploying update to the Tatarus security page. If you encounter issues, post here.
  • Want to keep track of this thread?
    Accounts can bookmark posts, watch threads for updates, and jump back to where you stopped reading.
    Create account
Sometimes I really wish we had this react from a sister site.
1789325227516.png
 
100% AI slop (according to Pangram 4.0 and my spider sense). Entertaining nonetheless.

And best way to combat slop is slop, so here's Claude's summary:

What it is​

It's an annotated rebuttal page, published 10 Sep 2026. It reproduces in full a short essay by pluie (Leah Amelia Chen), "Don't Give Them the Time of Day," posted 9 Sep 2026, under its CC BY-SA licence, and surrounds it with line-by-line commentary attacking it.

The page's metadata and clock give the author away: it's srid (Sridhar Ratnakumar), one of the three people pluie's essay names. So this is a response from the accused, not a neutral analysis.

The essay being annotated​

pluie's argument, in brief: DHH's Omarchy project may adopt Nix; DHH is influential and (in pluie's view) politically objectionable; people previously banned from the Nix community — Jon Ringer and srid — are using the moment to regain standing. The correct response is not to argue with them, because engagement feeds attention, which is the resource they want. Instead: ignore them, don't ban them from using Nix (that would prove their persecution narrative and violate the licences anyway), and win the internal narrative before the 2026 Steering Committee election.

The page's argument against it​

Three overlapping critiques:

1. The psychological frame. The page reads the essay through Richard, an Australian writer (d. 2024) behind "actualism." His key concept is the social identity: a psychological construct society installs over the instinctual self, containing a conscience or "guardian" that polices on the group's behalf and speaks as "we." Every "we must" in the essay is tagged as the guardian talking. Related terms it applies: psychic web (projection, introjection, gaslighting), peasant mentality (resentment of the influential), victimology. Richard's list of ten identity types is turned into a bingo card scored against pluie's public writing.

2. The hypocrisy ledger. A table of the essay's claims against what its own linked sources say. Examples: the essay says Ringer was suspended for bad-faith debate, but the linked moderation notice describes his language as civil and unobjectionable; the essay says srid insists queer identities aren't real, but the linked page says srid will use preferred pronouns; the essay calls "woke" an alt-right coinage, when it's Black American English predating that by decades. Plus the structural one: a 1,300-word essay telling people not to feed trolls, naming the trolls, published the week before an election.

3. The hacker-ethic scorecard. Steven Levy's tenets from Hackers (1984) — unlimited access, information should be free, mistrust authority, judge people by their work — scored against the essay and Nix's Code of Conduct. Nearly all marked as failed. The claimed inversion: open source succeeded by asking what you built; the essay asks what you are and whose side you're on.

It closes by "giving the time of day" — a live clock, three of pluie's sentences answered in Richard's ornate style with dictionary footnotes for words like ultracrepidarian and tergiversation, and a direct address suggesting pluie investigate the fear rather than the enemy.

How to read it​

The receipts-checking is the strongest part and is verifiable — you can follow the links and see whether the captions match the sources. The actualism framing is the weakest as argument: it's an unfalsifiable lens. Once "social identity" explains everything a person writes, any reply from pluie would also be the guardian talking, which means the frame can't lose. It's also aimed asymmetrically — Richard's categories would apply just as readily to the author's own grievance.

Worth knowing that both parties are participants in a long, bitter governance fight in the Nix/NixOS project going back to 2023 (moderation bans, a constitutional assembly, a Steering Committee). Neither this page nor the essay is a disinterested account of it, and the page doesn't pretend otherwise.
 
"pluie — Leah Amelia “pluie” Chen, GitHub pluiedev. In pluie's own public words: a Chinese open-source developer based in Germany; maintainer of “a dozen packages in Nixpkgs” without commit access; Ghostty developer and community manager; 2025 Steering Committee candidate offering “an intersectional perspective that can help organizations avoid common blind spots that can be hard to notice for tech communities dominated mostly by cisgender, male, white, Euro-American members”. Site footer: Made with Lume, Tailwind, 💜, caffeine, and a smudge of estradiol. (candidate statement, home page.)"

pluie: He is nonetheless very influential, well-funded and well-connected, and it is completely natural for his opponents to feel intimidated, and for his proponents, vindicated. This is already quite evident by the outbreak of hysteria coming largely from the political left in the Nix Community<a href="https://time-of-day.pages.dev/#fn2">2</a> who rightfully criticize him for his increasingly reactionary and appalling ideas regarding ethnicity and immigration, which are extreme even within his cultural context. Many fear that Nix will face an onslaught of right-wing fanatics that oppose our Code of Conduct on respect and inclusivity. We must resist this fear.

pluie: Or Sridhar “srid” Ratnakumar, who got banned for insisting that queer identities aren’t real and are only a malformed political ideology and that systemic racism is only a figment of your imagination. They rally around DHH’s move while parasitizing on his popularity in the Tech Bro Ecosphere™, in order to regain legitimacy for themselves. We must not allow them to do so.
GEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEG that was fast! These fucking busybody content police worms come right on out as soon as there's the barest hint of rain. I pray to Vishnu that saar DHH gives them all the south Calcutta dirk so that one way or another, Nix and Asahi can be reclaimed from the hands of dirty trannies. Not sure how much of a W that is if it passes from the hands of the mentally infirm to soon-to-be mentally infirm AI addicts, but hey, tranny seethe is always nice to see. The only thing that worries me a bit is that a potential mass ousting might result in them migrating to Guix en masse. The janny tranny machinery is there with Ludo's reluctance to be a proper BDFL and leaning on a CoCuck for moderation.
 
I'm not really a Linux power-user, but I do know that I would be able to recover from accidentally nuking my desktop environment or most of the other issues that have popped up in the Linux challenges that I've seen. Unfortunately, a lot of that knowledge came from something going wrong on my personal box in ways that I never really encountered on Windows.
So this is a bit of a tip to get out of "I've nuked my desktop environment."

Enable SSH on your machine as it is normally `sudo systemctl enable --now sshd`. Having a username/password for SSH is fine if the machine isn't directly exposed to the internet. If you ever nuke your environment, you can use another computer that has an SSH client to fix your machine (as long as it boots and connects to the network). So you can google/ask the AI while problem solving.

Most of the time if you nuke your desktop environment, it is because you've removed some base package the DE needs, and it just uninstalled all the other packages as they depended on it. So you just reinstall the meta package, and you are normally fine. I've done this a bunch of times and I just look up the meta package (or if it is Debian, just use sudo tasksel). If you uninstalled the session manager, you just reinstall it and enable the service e.g. sudo systemctl enable --now <sessionmanager service name>, normally this can be autocompleted e.g. GDM is almost always something like gdm.service.

If the machine doesn't boot, as long as the partitions are there, the installation exists and just isn't bootable. If you have some sort of installation media, you can mount the partitions under /mnt on the live installer and then chroot in and then troubleshoot. If you are already used to mucking around in VMs, you can familiarise yourself with the process.

Very rarely is it impossible to recover an installation. The only time I would say it is impossible (and I mean really impossible) is either drive failure, you've lost the encryption key (if you've set one) or you zero'd/scrubbed the drive.
 
So this is a bit of a tip to get out of "I've nuked my desktop environment."

Enable SSH on your machine as it is normally `sudo systemctl enable --now sshd`. Having a username/password for SSH is fine if the machine isn't directly exposed to the internet. If you ever nuke your environment, you can use another computer that has an SSH client to fix your machine (as long as it boots and connects to the network). So you can google/ask the AI while problem solving.
Unless you have to copy and paste a bunch of stuff couldn’t you just drop to TTY on the computer, instead of having a ssh server running all the time?
 
Unless you have to copy and paste a bunch of stuff couldn’t you just drop to TTY on the computer, instead of having a ssh server running all the time?
You can, but it can be slower due to it being awkward having to switch between one machine and another. Running an SSH server isn't resource intensive and I don't think it is a big security risk to have user/pass on your LAN, as your ISP and the router have a firewall configured.
 
You can, but it can be slower due to it being awkward having to switch between one machine and another. Running an SSH server isn't resource intensive and I don't think it is a big security risk to have user/pass on your LAN, as your ISP and the router have a firewall configured.
How about use tty to turn on the ssh server instead of running it 24/7?
 
How about use tty to turn on the ssh server instead of running it 24/7?
Sure, you could do a number of things. I gave a suggestion on how you could achieve a particular task.
I personally like the ability to remotely administer the machine whenever. I find myself using quite often.

While you could fuck around turning on/off, is there any real benefit? Probably not. If you are worried about using password auth on a local network, you can just copy over the public key and disable password auth.
 
Last edited:
Trouble in paradise down in Guixville, Ian Eure, a core maintainer, has stepped down due to the failure of GCD008 to ban AI-gen commits. An ill omen considering Guix was already having issues with maintainers and the obsession with making any effort to change project policy as slow, difficult and annoying as possible. I wonder if others will follow suit.

Edit: As far as I'm aware GNU already has a hard stance against AI code, so I am not sure why this was even a question for Guix. Even RMS chimed in during the deliberation. It would be a grim, grim, grim day if Guix were to stop being GNU over this.
 
How about use tty to turn on the ssh server instead of running it 24/7?
Why, though? And besides, some distros offer a web interface for configuration, in case you have an issue with SSH specifically, even with 2FA for the extra paranoid. My openSUSE does exactly that. I can configure nearly everything from a different PC on the same network via the web interface, just like with Proxmox. It's really neat.
 
What’s the use case for colours?
Weren't the GNOME devs the reason why it took so many years for the DEs to agree on the default set of colors?

Is this another op by Drew DeVault?
Also it's funny that the antifa distro is allowed to exist but not the apolitical ones.

>People to avoid
>Vox Day

Really? In 2026?

>We have the capabilities to build a better future together. We must not go gently into that good night.
Wow calm down, theater kid.

>Webpage created by Brennan Kenneth Brown
Queer, métis and C*nadian.
image (2).png hero-image.png
Queer Métis poet, writer, and JAMstack developer
Brennan Kenneth Brown is a Queer Métis writer exploring Indigenous identity, digital culture, and creativity.
Not sure if the native version of the Sonichu medallion will help him fight the chuds.
 
Last edited:
Trouble in paradise down in Guixville, Ian Eure, a core maintainer, has stepped down due to the failure of GCD008 to ban AI-gen commits. An ill omen considering Guix was already having issues with maintainers and the obsession with making any effort to change project policy as slow, difficult and annoying as possible. I wonder if others will follow suit.

Edit: As far as I'm aware GNU already has a hard stance against AI code, so I am not sure why this was even a question for Guix. Even RMS chimed in during the deliberation. It would be a grim, grim, grim day if Guix were to stop being GNU over this.
IMO AI code is still better then jeet code. It was shaky for a bit but Qwen 3.8 27b has passed the threshold and now as long as you don't let every idiot and their cousin submit PRs that were written with a very bad AI prompt it should be fine.
 
IMO AI code is still better then jeet code. It was shaky for a bit but Qwen 3.8 27b has passed the threshold and now as long as you don't let every idiot and their cousin submit PRs that were written with a very bad AI prompt it should be fine.
The problem IMHO is that it's way too easy for people to let go of the reins, especially since LLMs are very good at writing meta-prose (comments, commit messages, PR descriptions, etc.) indicating they know what they're doing, and then the small issues they introduce accumulate extremely quickly.
A good example is the kernel itself, it's an insane slopfest now (ironically Spengler of grsec also shows how to correctly use LLMs in between highlighting how insane upstream has become, especially LTS backports).
 
IMO AI code is still better then jeet code. It was shaky for a bit but Qwen 3.8 27b has passed the threshold and now as long as you don't let every idiot and their cousin submit PRs that were written with a very bad AI prompt it should be fine.
The trouble is that Guix is a tiny project with a handful of core maintainers that almost unanimously dislike AI and do not want to use it. Ludovic Courtes himself, the founder of Guix, the Shepherd init system and a very prominent contributor to Guile, forbids contributions in all of his other projects, but is too chicken shit to take the reins of Guix as a proper BDFL. Thanks to one sealioning faggot ~2 mo back, the AI ban proposal for Guix failed, and people are now acting accordingly. Whether AI is finally useable or not, I have no clue, I do not and will not use it, but the way I see it is as @Hannibalistique said, give an inch and people will take a mile. The Kernel contribs are full of very skilled programmers, and still it is a horrible mess. For a distro that has maybe a couple thousand users like Guix, letting it accumulate sloppy cruft is both against the ethos of the maintainers and, in their own words, just puts the burden of checking and debugging that dubious AI code on them.
 
SOMEONE clearly hasn't spent enough time in leftist circles to witness leftist cannibalism unfold in real time. It's 100% sincere, and Geerling and Kling are the subjects of anathematising because they didn't toe the party line hard enough
I've had my run-ins with the "frog posting is racist" crew. If that helps.


 
Back
Top Bottom