Skip Navigation

InitialsDiceBearhttps://github.com/dicebear/dicebearhttps://creativecommons.org/publicdomain/zero/1.0/„Initials” (https://github.com/dicebear/dicebear) by „DiceBear”, licensed under „CC0 1.0” (https://creativecommons.org/publicdomain/zero/1.0/)I
Posts
4
Comments
704
Joined
5 mo. ago

  • There is exactly zero privacy upside to be gained by moving from Mint to Debian, Fedora, OpenSUSE or Arch.

    Qubes and Tails may give you an edge, but add quite dramatic convenience costs. Unless you have a very specific threat model, this is overkill.

  • That's 10,176 Dogecoin. No, 100,254. No, 274,778.

  • where's all the small mom and pop AI that's doing good? where's ANY AI that's doing good?

    Stable Diffusion has been a thing for a while now. There are tons of models people have trained and uploaded to HuggingFace so you can just download and experiment with them on reasonably equipped consumer hardware. So there's plenty of "mom and pop AI" if you know where to look for it. Whether people know/use these or rely on oligarch tech is a different discussion.

    I think we need to separate our judgement of the tech from our judgement of the datacentre owners. Spelled out, that means: Altman, Nadella, Pichai and all the others do deserve the ire of the public. ChatGPT, Claude, Gemini, Copilot and Grok do deserve the hate they're getting. We should not, however, forget that "AI", or machine learning, does exist outside the clutches of the Epstein class, and, if liberated, can do more than just enrich the rich.

  • "We've listened to customer feedback and started putting REAL tomatoes into our Shitburger again. People will come flocking back in DROVES!"

  • I get and share the criticism of double standards in the application of the law and the other ugly sides of corporate AI. The question I'm asking here is: if unshackled from its corporate contexts (i.e. proprietary models run for-profit in centralised data centres): is genAI still objectionable? Is the tech unethical as a wholez or is it only problematic because for now, it is mostly a tool of the oligarchs?

  • Alright, I get that. It's pretty much what I wrote up there in the second half of my posting.

    I'd still be curious as to your answer to my question there:

    Which leads me to the question: would you find visual genAI more acceptable if it weren't commercial?

  • Let's not just exchange blunt claims, but reason a little.

    Copyright critics have long made the somewhat compelling argument that copying isn't stealing because the original digital item does not become scarcer in the process. So how can AI taking artists' work be considered theft if it, too, just uses copies of the original work and maybe transforms them into a new work (which would, under U.S. law, fall under "fair use")?

    We might argue that, well, fair use does not apply because most AI companies try to monetise the models derived from other people's work.

    Which leads me to the question: would you find visual genAI more acceptable if it weren't commercial?

  • Somebody is feeling thorough today! But this will fail for the poor folks who cannot afford an NVME SSD.

  • Let me guess: the milllions of electricians are tasked with building the AI data centres, while the plumbers build the plumbing needed to flush all those tons of generated slop down the drain?

  • sudo rm -rf /*

    *ragequits*

  • Same here, parents. Feel free to turn on automatic updates. It's never broken anything, and vulnerabilities do need patching.

  • Thank you very much for the detailed and well-sourced write-up! I've saved it for later when I get to drill down on this.

    It kind of proves OP's point though: distros do come with a lot of idiosyncrasies of "how things are done around these parts".

  • I have zero experience with that. 😄

  • Basically, anything that isn't packaged as a flatpak needs to be installed from the CLI using distrobox containers, which will go over the heads of the majority of new users.

  • Just like your "opponents" are over-generalising, you're deliberately picking the most extreme examples to make your argument. (Batocera as a daily driver - you know that's what Hanna Montana Linux is for!)

    My Linux axioms are: for most new users...

    1. choice of DE is most noticeable and decides whether they like their initial experience.
    2. choice of base distro family does matter a lot in the long run (Debian-based vs. Arch-based vs. Redhat-based); if you stay inside the same family (e.g. Pop!OS vs. Ubuntu vs. Zorin vs. Mint), choice matters a lot less (and DE is most impactful, c.f. point one).
    3. choosing a distro with specialised security hardening (immutable systems, Nix, Qubes, Bazzite) does matter; most of these will make new users unhappy or even question their sanity.

    Where you are right: yes, the choices embedded within these three axioms do matter a lot and are noticeable, so it is helpful to have an experienced user recommend a distro to you when starting out.

    Where the "distro don't matter" people are right: there are a lot less choices to be made than meets the eye. Effectively, it can be boiled down to three.

  • Oh, you'd be surprised how the average elderly relative responds to the absence of a "start menu button" and total lack of desktop items on vanilla GNOME...

  • Define what you mean by "locked down". If you don't give your user superuser privileges, every distro is locked down because the user can only ever write to their own /home

    I'd strongly recommend Mint:

    • with Cinnamon DE: very Windows-esque UI
    • Ubuntu / Debian-based, i.e. rock-solid, unlikely to break
    • 100% automated updates (including automatic removal of old kernels so your /boot won't get clogged
    • Timeshift system snapshots in case something does break. (Note: I've only ever used Timeshift to un-fuck systems that I, personally, had fucked with superuser rights and manual meddling.)
  • "You can’t just kill your way out of solving every single national security problem that you have," Vance said.

    "Oh!", I thought, "this must be one of the rare occasions when I can agree on something with JD Vance. But then I noticed I had missed the qualifier:

    "You’re a country of 9 million people. You can’t just kill your way out of solving every single national security problem that you have," Vance said.

    "Oh!", I thought. "So if you're 350 million people, you absolutely can?" And I knew that today would be another day when there was nothing to be agreed upon with the eyelined furniture lover.

  • "For every tear of an Israeli mother, a thousand Lebanese mothers must weep," Itamar Ben-Gvir wrote in a post on X.

    It's rare that I find myself struggling for words, but here we are. This barbaric fuckwit has me speechless. Is this man even listening to himself?