Mastodawn

Trevor Bramble Mar 14

Systemd has denied to revise their policy in regards to AI.

they've also marked evidence that people gave regarding its effectiveness as off-topic, then locked the conversation.

I believe the authors have not understood the weight of the issue. Later this day, I will begin drafting an open letter to Systemd's authors under the Starlight Network umbrella of projects. EDIT: or perhaps I will take a different approach. There's many more issues I want to talk about.

Disallow usage of generative AI to write code · Issue #41085 · systemd/systemd

Component No response Is your feature request related to a problem? Please describe Generative AI is actively killing people, driving up costs, and plagiarizing work from many open source developer...

GitHub

@alexia funny to see the guy go "oh but it's good at doing code reviews because we find security issues"

buddy code reviews isn't how you find security issues, code reviews is here to check if the code will fit in with the rest of the project and can be maintained. if you do it for security, you're doing an audit, not a review.

@alexia it's also why i get agitated when people go "you should review the code your LLM gives you" because review is here to know if it looks right and can be maintained. you know, exactly what LLMs are good at faking.

yes, i'm being incredibly pedantic, but words have meaning and we have workflows built around those

Kim Crawley 😷 (she/her)Mar 14

Ohhhh, this is me. Avoiding Gen AI and Covid, no matter what it tskes.

https://mastodon.social/@plaguepoems/116206954876505716

@SRAZKVT @alexia

neobot_glare_sob

i hate AI bros RAHHHHHHHHH

if only someone made a systemd-compatible init that was also good...

Rexo the Trans Bus 🏳️‍🌈🏳️‍⚧️🚌Mar 13

@alexia the tone of this comment bothers me so much. you have to respect it. you must accept the bots.

Disallow usage of generative AI to write code · Issue #41085 · systemd/systemd

Component No response Is your feature request related to a problem? Please describe Generative AI is actively killing people, driving up costs, and plagiarizing work from many open source developer...

GitHub

@rexo @alexia I initially just went past that being like "oh, average AI bro comment". Then I went back and saw it was by Lennart Poetterring

aneobot_explode

FoolishOwl Mar 14

@tulpenkiste @alexia @rexo "There is a time when the operation of the machine becomes so odious, makes you so sick at heart, that you can't take part. You can't even passively take part! And you've got to put your bodies upon the gears and upon the wheels, upon the levers, upon all the apparatus, and you've got to make it stop! And you've got to indicate to the people who run it, to the people who own it -- that unless you're free, the machine will be prevented from working at all!"

-- Mario Savio, 1964

Marta Threadbare Mar 15

@rexo @alexia "It has been finding real CVEs in the past weeks, and it's just too good to ignore now."

Even assuming that's true, the dude himself says it's been better for just weeks, at most it would be advisable to just *start considering its viability* for a later decision, not just "yeah whatever let's allow it right now".

Also what about regressions? even openAI had a blunder with GPT-5. You can afford to use a system with regressions for a hobby project, not systemd.

@alexia completely fucking ridiculous. at this point, we’d all be better off just recreating everything that AI has infected instead of arguing with these people.

@mrmasterkeyboard easier said than done, even just getting a hard-fork off of the ground is quite difficult, letalone a drop-in replacement

@alexia if me and the others working on EVi could do it, i'm sure anyone with enough knowledge could!

@alexia @mrmasterkeyboard i mean there's already openrc and s6 and the rest of systemd can be replaced with extra independant components on the side

putting everything in one place is exactly how we got into this mess, why recreate it ?

@SRAZKVT @alexia Just because you can recreate it doesn't mean you need to make the same choices as them too. That's the fun part of recreating things.

@mrmasterkeyboard @alexia the problem is the very design of a lot of systemd components is that each depend on several others in an incomprehensible web, so unless you're willing to cut some features, you'll need to redo the whole thing in mostly the same way

if you're fine with just a subset though, you should be mostly fine, but again, it'd be a subset, not 100% compatible

Ra (Freyja) (it/its)𒀭𒈹𒍠𒊩 Mar 13

@SRAZKVT @mrmasterkeyboard @alexia ngl, I wonder if folks can do a port of SMF to Linux? that would make me a lot more likely to want to use Linux tbh

@freya @SRAZKVT @alexia i'm not sure what SMF is

Ra (Freyja) (it/its)𒀭𒈹𒍠𒊩 Mar 13

@mrmasterkeyboard @SRAZKVT @alexia oh! it's a very finely scoped init system that's kind of like "what if we did systemd, but actually done properly"? Doesn't use shell scripts, has its own awareness of service states, does a lot of the same things systemd does in terms of initing but it doesn't like do more than that

@freya @SRAZKVT @alexia oh, that seems cool

i looked it up and it's a Solaris thing, i think Solaris (during the Sun days) was pretty neat

Ra (Freyja) (it/its)𒀭𒈹𒍠𒊩 Mar 13

@mrmasterkeyboard @SRAZKVT @alexia yeah! I use Solaris on a daily basis and I think SMF is one of the best components

@freya @SRAZKVT @alexia i need to get back to Solaris honestly and see if i could make it a part of anything... the only one i used was Sun Solaris 9 on UTM...

Ra (Freyja) (it/its)𒀭𒈹𒍠𒊩 Mar 13

@mrmasterkeyboard @SRAZKVT @alexia omg niiiice

Ra (Freyja) (it/its)𒀭𒈹𒍠𒊩 Mar 13

@mrmasterkeyboard @SRAZKVT @alexia if you want an account on one of my Solaris boxen let me know, hey?

@freya @SRAZKVT @alexia hmm, sounds cool, i'll think about that and see if i do if i start trying to use Solaris more.

(might take a while, i've got a lot of stuff on mind at the moment and... honestly, i do too much nowadays, this will take a while to get through everything i want and i keep procrastinating. apologies for the involuntary vent.)

Ra (Freyja) (it/its)𒀭𒈹𒍠𒊩 Mar 13

@mrmasterkeyboard @SRAZKVT @alexia nah is totally ok!

@freya @SRAZKVT @alexia although i'm curious, is it a SPARC machine and what version of Solaris may i ask?

Ra (Freyja) (it/its)𒀭𒈹𒍠𒊩 Mar 13

@mrmasterkeyboard @SRAZKVT @alexia SPARC yes, UkltraSPARC-IIe running a heavily modified version of Solaris 10 1/13 with the 2021-10 security patch and the FractalKit Solaris upgrade kit

@freya @SRAZKVT @alexia ooh, niice!

Ra (Freyja) (it/its)𒀭𒈹𒍠𒊩 Mar 13

@mrmasterkeyboard @SRAZKVT @alexia not a terribly high spec machine, Sun Blade 150, but holy shit if it's not the most stable thing in the history of ever

SIN001E/R-SINE | ΘΔ🦈 | 🏳‍⚧Mar 14

@SRAZKVT @mrmasterkeyboard @alexia i don't think you even really need to recreate systemd in the first place though

like, i've been running runit for a while now, and have had practically zero issues; turns out the vast majority of things that "depend on" systemd literally just need their service file rewritten, which in most cases takes 5 minutes

and on a distro built around runit like Void, that's of course already done in packaging

@sinewave @mrmasterkeyboard @alexia i don't think runit is perfect (even though i run it as well, and have for 4 years now), because runit doesn't handle service dependencies, but still, s6 and openrc exist and both support those

SIN001E/R-SINE | ΘΔ🦈 | 🏳‍⚧Mar 14

@SRAZKVT @mrmasterkeyboard @alexia true, although there's a very simple (and very runit-y) solution; you just run sv check dependency and exit with nonzero if it fails

the overhead isn't particularly high since it waits a few seconds between attempts; and as it turns out not that many services even need explicit dependency handling

@mrmasterkeyboard @alexia I actually didn't understand any of it. I know about systemd. could you explain in simple words what is actually happening there?

@Rose @alexia Basically, they've allowed AI contributions by providing AI agents instructions on how to contribute, plus I also think they did some AI sloppery in documentation? (Then again, I don't really use systemd other than on my Ubuntu laptop which I will now refuse to update, and will be switching to FreeBSD soon which I predict I'll work on in around another... month or more?)

@mrmasterkeyboard @alexia another one 🤦🏼‍♀️ Any other distro that doesn't use systemd? I don't do much on my laptop, a browser it's almost all I need.

@Rose @mrmasterkeyboard

There is many. Personally I like Alpine and Void Linux. Chimera Linux is also something to look at.

@alexia @mrmasterkeyboard I would look at them, though I'm not that techey. I hope I will find something user friendly and easy to install

@Rose @alexia though at this point, Linux allowed AI into it too, so you're better off moving to FreeBSD to escape AI... or even a hobby operating system.

@mrmasterkeyboard @alexia how will be the hardware hardware compatibility on BSDs? Again I'm afraid of 'hard' techey stuff

@Rose @alexia FreeBSD is the best for hardware compatibility, but it's also nowhere near Linux.

Some of my machines that ran perfectly on Linux don't even run correctly on FreeBSD. (Missing touchpad drivers, no Wi-Fi, etc)

Honestly no OS is perfect and it depends on your hardware. If you don't want to fix stuff constantly, don't go to FreeBSD (or a hobby OS, those will be way worse...)

My advice is to keep on a Linux version prior to 2025 or if you're paranoid enough then 2022.

@mrmasterkeyboard @alexia so I think the end of the story for dummies like me is we are stucked

neocat_floof_sad

@Rose @alexia yeah until someone can properly get something good up and running. for now I’d say resist updating Linux.

@Rose @alexia A few that I know allow you to select other init systems:
Gentoo (OpenRC or systemd)
Devuan (Debian but OpenRC, sysvinit or runit)
Void Linux (runit)
Alpine too (OpenRC)

Ian Jackson Mar 14

@mrmasterkeyboard @Rose @alexia I've been running Debian with trad sysvinit for years now. It seems fine.
The procedure for a fresh install is unfortunately shell runes including halfway through the installer but should be doable by your local Debian expert. IIRC the instructions are on the wiki.
My thanks to the Debian init diversity team! (Some of whose most important contributors are also working on Devuan.)

@alexia genuine question, why is this such a big problem?

@alreadydeadxd There are many reasons, but to cut it short, the effects of the continued use of Generative AI go far beyond just code quality. It introduces licensing ambiguity, contributes to devaluing software development as a skill, and encourages increased energy use and expansion of data centers which has shown to negatively affect marginalized communities.

@alexia
ah, i see, the code quality argument is really starting to fall apart in 2026, haha

i don't know that much about the license ambiguity, but to be honest i don't care that much either, maybe it's because i think all code should be open source, and copyright shouldn't exist, lol. nevertheless, i'm aware things don't work like that in our current dystopia, which sucks

i've always took the devaluation of software development as a skill to be just marketing and tech bro hype, to do ai-assisted programming properly you still need to have all the knowledge a software developer that doesn't use ai would have in the first place, you can't just be like "claude, build systemd or go to jail" or whatever

the energy consumption and data center situation is worrisome, it could be done in a way more sustainable manner, but when did big corporations care about the environment and marginalized communities, am i right? i also feel like people tend to overemphasize this issue in particular, it's not the only case of corporations and/or government being evil at the expense of others, not the biggest such issue by far, and not the only problem in which the population/consumers are supporting bad things happening through their behaviour. for example, if you complain about how data centers are bad for the local communities and the environment, but then go to the supermarket and buy some beef, i don't even want to talk to you, haha

@alreadydeadxd
> if you complain about how data centers are bad for the local communities and the environment, but then go to the supermarket and buy some beef, i don't even want to talk to you

First, this is whataboutism: people do not have to behave perfectly to criticize bad behavior, otherwise no one would ever be able to criticize anything.

Second, I've been a vegetarian my whole life. I have never eaten beef. Will you listen to my criticisms of LLMs now?

@skyfaller yeah, i get that, i just find it very disheartening to see how some people suddenly develop more empathy and environmental awareness when it's favorable to their arguments against something they already dislike, but then fail to keep up those same values when it comes to other issues, it's almost like they never cared to begin with

of course, i'll listen to your criticisms of llms

@alreadydeadxd The really short version is that the tech industry was already an accelerating environmental disaster:

https://solar.lowtechmagazine.com/2015/10/why-we-need-a-speed-limit-for-the-internet/

LLMs caused the tech industry to give up on all of their sustainability/emissions goals. At a time when we need to be zeroing out emissions from every source, LLMs and techbros are doing their part to make the world uninhabitable.

The climate crisis really ought to be enough to end the conversation on LLMs. I resent having to come up with other reasons.

Why We Need a Speed Limit for the Internet

The energy use of the internet can only stop growing when energy sources run out, unless we impose self-chosen limits.

LOW←TECH MAGAZINE

@skyfaller i'll read the article

a very weeny construct 💀Mar 16

@skyfaller wow this toot is 🔥🔥🔥

immediately i want to boost it but i don't want your conversation partner to get dogpiled if it's popular

definitely post the same sentiments standalone or if you already have pls point me at it

@alreadydeadxd
you can't just be like "claude, build systemd or go to jail" or whateverYes, true! however... the thing that is actually devaluing this work is that those who pay us value us less, because why would they pay us a high salary if they can pay 20 bucks a month for claude? Same goes for artists, writers and lots of creative jobs.

The fact that you still need to understand what you're doing doesn't matter to those we develop for.
if you complain about how data centers are bad for the local communities and the environment, but then go to the supermarket and buy some beef, i don't even want to talk to you, hahaI think there's nuance here, every single person cannot be absolutely perfect, and especially not every single person has reflected on every little thing they do; We can only care about so many things before we collapse. That said, the reason people care about these data-centers is because they went digging for reasons not to use AI, not the other way around. If AI hadn't been, issues with large datacenters would've probably gone undiscussed for many more years. Which...I guess is a good thing?

@alexia i don't think the issue you pointed out with managers being sold on ai and devaluing workers is going to be made any better by all the ai hate we see online, either the tech bros are right and coding is going to be obsolete in 5 years, in which case management is going to continue to seek profits as always, or everything will come crumbling down in the next years, and the people who bought into the ai hype are going to have to come back to earth with the rest of us

yes, it's like you said, people hate ai, so they started looking for reasons not to use it, that's why they are not consistent with their values

i'm personally kinda bothered by how the conversation around ai is going, on one side there is almost this religious fanaticism, and on the opposite side, where most people on fedi seem to be, there is a lot of hate, as if the technology itself is evil. most negative commentary i've seen around ai is just pointing out issues with state capitalism, and then blaming them on ai in particular. like, if wage slavery wasn't a thing, we wouldn't be talking about the devaluation of the worker. if people were more environmentely conscious, we would figure out how to make the tech even more efficient, and how to run it as sustainably as possible long term, not rushing directly into building massive data centers all over the world. if people were generally opposed to the centralisation of power, we wouldn't have only a couple of very big genai competitors that instill their own biases into the models, and so on

@alreadydeadxd I agree with basically everything you said, but to be pragmatic, I must work from inside the system and tear it down from there

I unfortunately cannot act as if all of these are non-issues due to the world we live in so this will have to do as the next-best thing

@alexia sorry for the late response. i do understand where you're coming from, but taking it back to the original conversation, meaning the systemd situation, and other such projects for that matter, i can't help but feel that this is a losing battle. imagine i'm a systemd contributor who wasn't polarized against llms already, there is practically nothing you could say that would change my mind. and from the perspective of the project itself, they can't tell if contributions are hand-written or ai generated, if someone chooses to use ai, they might just as well try to provide support to improve the code quality. from the user standpoint, i don't want to run poorly written or vulnerable software, but not all llm generated code is slop, you can totally use ai agents to write decent quality software, so this falls back to a case-by-case basis. do i have ethical concerns related to the use of ai as it exists today? totally. but i wouldn't go as far as switching from systemd to something else though. there's probably so much code that runs right now on my system, which was written by people with whom i don't share all my moral values. i think we need broader systemic and cultural change, the current version of genai is only a symptom of deeper issues

Bolet modo Sombrero de paja 🏴‍☠️

@alexia @alreadydeadxd

It's like people had an emotional response when their very source of work is being threatened and the skill they developed their whole life is rapidly devaluing. That irremediably leads to radicalization (in fanatism or hate) when not depression.

And a very important detail about this and the one that is the most infuriating is that the tech itself can be good. But the people in charge (your bosses, AI companies, economic powers) won't let it. They only want to use it extract more wealth and waste resources. It's one of the newest biggest technology innovation in years, corrupted only to make stupid business people think they are the next rockstar X10 developer.

A word manipulation machine that can seemingly do good, and by the actions of their users is instead destroying the very value of tech work itself.

@javascript @alexia totally agree

jaKa Močnik Mar 13

@alexia lennart was always a pain. shouldn't really entrust the init system to a guy like that. 🤷

Legion495 Mar 13

@alexia I am not amused.

Bernd Petrovitsch🔴🔴🔴♂️🏳️‍🌈🇦🇹🇪🇺Mar 13

@alexia systemd developer have long history of not understanding ...