oil capital can't help but pollute open source, huh. Gotta get those oily fingers into every fucking open source project, twould seem.
fuck it man, gonna make my own vim out of vendored copies of readline and ncurses
I did get a response on whether Helix has a policy re: LLM usage or not

https://github.com/helix-editor/helix/discussions/15408

I wouldn't call it encouraging because I don't agree that it's useful (nevermind ethical) but I wouldn't call it
discouraging, either, because they're clearly aware that it can produce a lot of unhelpful, time wasting crap.
Policy on accepting PRs that include LLM generated code? · helix-editor helix · Discussion #15408

Hello! I've just been checking out Helix recently, and I quite enjoy what I've seen so far! It's fast, intuitive, ships a lot of nice defaults, and works well with what I've used it for thus far! I...

GitHub
I haven't had the time or mental fortitude to construct a diplomatic reply that might encourage them to consider the ethical side that isn't just me ranting but if anyone wants to politely and gently nudge them towards considering the ethical drawbacks and license implications...
on the other hand, never a better time to greenfield the entire field of computing

just fucking NIH the whole damn stack

/s sorta
Microsoft, OpenAI, Anthropic, etc, are defense companies with a large PR team.

@aud I've got some vague idea that the ideal of "greenfield" development has been one of the problems with the ideology around the software industry. Come to think of it, that might be why Musk & Co. get so worked up about Mars.

On the other hand, I was just doing my rant yesterday about hierarchy being baked into the basic models of user accounts and filesystems.

@aud yeahhhhhh we fantasize about what our Forth could become ...... if there were world enough, and time...
@ireneista it feels low key ridiculous to start wondering if the time is now to make a genuine linux fork because I am not qualified for that buuuuut
@aud we certainly know people with every relevant field of expertise, the problem with it is just that it's too big. the way to go would be to make a new kernel (or adopt an existing smaller one)
@ireneista I almost added another comment saying exactly that: that it takes a village... of, in this case, communities and companies.

A lot of time and money goes into updating the kernel, frustratingly. And it sucks because Linux has hit a real sweet spot with running games -_-
@aud it does suck but the game thing is, like, due to investment of capital, yeah? like it's part and parcel with it. if we're trying to grow something outside of capital (and since capital aligns with hate when hate is on the rise, we don't really have a choice about that), ... that seems like one of the things we need to accept not necessarily being as good at
@aud an awful lot of that game compat stuff is in userspace anyway tbh. the fact that Steam doesn't completely fall apart on NixOS suggests to us that it can't be that hard to do a freebsd-style ABI emulation
@ireneista that and the graphics stack, yeah. To some extent it might arguably have been faster had Valve invested in a fork ala Android, and I'm glad they didn't.

But you're right. My biggest thought was about driver compatibility; it's already been a decades long slog to get manufacturers to do decent Linux drivers, and they're a far cry from
actually doing it. But assuming a BSD fork had a good policy on LLM usage, I wonder if that would be a pathway?... (for software almost certainly. But drivers? hmmm. I just don't know enough).
@ireneista like at least SystemD can be cut out via OpenRC and such thing. Linus seeing no problem with LLM usage in Linux PRs is a very different issue than the Poettering issue...

hmmmm. Not for the first time, I wish I knew kernel architecture. What IS the game and graphics situation like on a BSD nowadays? Hmmm. I wouldn't say
no to dualbooting and contributing if I could and if there was a path...

It would be great if something like Qemu/KVM's VirGL existed for BSD, but I admittedly don't know if you can reliably use that even now for something like games on Linux...
@ireneista this would be a great time to buy two new drives and triple boot Gentoo+OpenRC / that Arch fork with OpenRC along with a BSD

thanks, LLMs, for driving up that cost, too.
@aud cc @linear does your demoscene background give you any insight into this sort of question?
@ireneista @aud ... what demoscene background
@ireneista @aud that comment aside,

i think freebsd supports virgl but i don't use it, so cant comment on that

i dual boot gentoo and freebsd on my laptop. all three major gpu vendors work fine, including proprietary nvidia drivers if that's your thing. my desktop environment on freebsd is the same kde plasma 6 wayland desktop as i use on gentoo, with the same fancy thirdparty graphical effects plugins. i run steam on freebsd and can play factorio with it. some more complicated games work, more things don't. proton works too, more or less, there's a script for launching steam that adds natively-built proton.

netbsd's situation is a bit more dire, with no proprietary nvidia support if you care about that, and the open source drivers generally being further behind
@ireneista @aud the graphics support in every other operating system tends to just come from linux, most of the DRM/DRI stack is dual licensed MIT/GPL, so other projects are free to take from it

this does mean that they are not immune to the importation of AI code. freebsd also does not seem to take as strong of an anti-AI stance as i'd like. netbsd does.
@linear Can you point me to your KDE Plasma FreeBSD? Or do I need to get good? 🥺
@Filene i just install the meta package from the package manager, and launch the session manually after logging in on the command line (same way i would on a Linux system, if not using a DM like SDDM)

you will, of course, need the graphics drivers installed and configured correctly first.

the main graphical plugins i use are kwin-effects-better-blur-dx and KDE-Rounded-Corners, both of which i build from source.
@Filene contrary to outdated documentation you might read, my experience with everything except SDDM has been that it "just works" on FreeBSD 15, even with wayland
@Filene bugs i thought were FreeBSD (e.g. firefox weirdness) turned out to actually also be present on Linux on closer inspection
@linear @aud we must have you confused with someone, sorry
@aud @ireneista I daily drive Alpine (musl) with openrc. It's also very small; and also afaik you can boot it from an USB drive (or CD) but commit any changes to your regular disk partition without it conflicting with the existing installed systems.
@aud .... but our self-monitoring heuristics tell us to be cautious of .... this thing https://irenes.space/leaves/2024-09-29-technology-community-idealism
technology, community, and idealism -- Leaves Given to the Wind

@ireneista agreed

@aud and we don't know whether this is a situation where we should override that heuristic or not

but at the very least it seems like we should be looking for ways to organize that aren't quite as much work as starting over

@aud we coined the term "the cybernetic bias" for a slightly different impulse in another of our Leaves a bit after that piece, but trying to reference it just now, they feel like two facets of the same thing to us. we may have to write another piece sorting that out

@aud but yeah as we were saying elsewhere

there really is a rising tide of hate and willful negligence projecting themselves throughout the free software world

we really do need to think about more-organized ways to fight back

@aud at the end of the day, to the forces of hate, software is a tool in the demeaning sense - "only a tool", its purpose is the thing it achieves for you

whereas to an awful lot of disabled, neurodivergent queer techies, software is a tool in the sense of "this is the thing I use to make my art, so I have learned everything about its history and how it's made and I love it deeply"

and the latter is a stronger position. it's not even on the same level.

@ireneista this is so well, and succinctly, stated. Software is a thing that allows me to express myself; to make art; to communicate; to interact with the world; to expand my boundaries; to meet other queers; to hear music; to make music; to have community in isolated places...

... and to them, well, it's just a way of disempowering labor and doxxing/harassing minorities.
@ireneista to a young, isolated closeted trans queer in rural mormon Utah in the 90s, with no queer elders, no physical transportation, no money, no hope of community... the internet was it. That was literally it. All the possible ways I had to interact physically with the humans around me invariably involved religion and queer hate. So perhaps that's why it all feels so deeply personal to see it attacked.

@aud @ireneista

The first time people were nice to me on the internet I sat down and cried. It's home.

Our home and the future is being stolen from us.

@violetmadder @aud we will win. always believe that. the challenge we face is just to figure out how.

@ireneista @violetmadder @aud YMMV, but I take solace in the facts that:

- Those currently despoiling the internet are part of an inherently unsustainable cult. The way in which it collapses matters *a lot* since they could take the habitability of the planet with them, but their collapse is literally inevitable.
- If I have to become a strange hermit using weird alternate tech that the rest of society mostly doesn't understand, I'm 100% okay with that. I'd *like* to win a culture war and help liberate a bunch of people (myself included, because imagining oneself as a liberating hero is a great way to set oneself up for failure), but when contemplating the difficulty of that and the chances of failure, I find it reassuring to imagine stages of partial failure in which I retain at least some agency.

@aud yeah and when we finally got into the industry we looked around and realized that nobody else cares about any of this in the same way.... except the other queer people, who had been having very much the same kind of isolated story as ours, in parallel, but each of us alone...

it's a very, very powerful thing to no longer be alone, and more than that, to realize that it was always just because society wouldn't allow us to find each other, that we always had the raw numbers

@aud the assault feels very personal to us, too

but people like us actually outnumber Nazis, going by every indication, and we're the ones who are actually decent at strategy

@aud thank you. we've put, you know, very nearly forty years of thought into our feelings about software, and the last few have given us good reason to put this part of it into words, so 💜
@aud we are totally gonna write our own vi soon though. we made two attempts at that in the 90s (both of them cut short by food scarcity and domestic abuse) and have been meaning to circle back to it ever since

@ireneista @aud I've already seen a Vim fork that when I have the energy I might switch to:

https://codeberg.org/NerdNextDoor/evi

I've also fantasized about just making an entirely fresh OS, even though I likewise lack the skills. There's gotta be... Thousands? Tens of thousands? of people in this bucket right now...

I think the deeper problem is governance? How exactly are you going to ensure an equivalent thing doesn't happen to this new project? I think clear politics baked into the project is probably best, but that's hard to get right.

evi

EVi, a hard-fork of Vim v9.1.0 (Jan 2024) before AI was used in the project.

Codeberg.org

@tiotasram @ireneista

I think the deeper problem is governance?annoying, but entirely predictable and expected, that the same problem plaguing society and nation states plagues software.

@aud @tiotasram it's hard to get right but that is a thing we have some experience with, for whatever it's worth

@aud @tiotasram but yeah we sometimes feel like.....

working together is the oldest problem humanity has ever faced and still by far the hardest

@tiotasram @ireneista @aud i think the values (what i think you mean by politics) are deeper than governance. governance structures are certainly important, but if they are held more important than values, that governance structure can be captured. if you have a community working together that all share some basic values, attempts to subvert those values will be met with resistance.

i personally think a good set of values stems from supporting the autonomy of the folks operating, creating, and otherwise affected by the software (or other project). resistance to large language models follows from it, for instance.

@tryst @aud @tiotasram we do agree with that. politics is the means by which we fight for such things

@tryst @aud @tiotasram and yeah governance is hard because .....

well, the larger and more complex a governance structure, the more it cares about perpetuating its own existence to the exclusion of the purposes it was nominally created for... and when that gets bad enough it's an oppressive thing that needs to be dismantled in its own right

@ireneista @tryst @aud yeah. I'm strongly anti-hierarchy for this exact reason, to the point where I'm not confident what's traditionally seen as governance is possibly fit for purpose here.

I think the fact that Vim got forked so fast is perhaps evidence of a workable strategy: be so persistent and attract and cultivate enough of a principled following that attempts to co-opt your projects will be met with a water-like flow-around response and although the co-opters may build their own version which might even attract attention for a while, it won't supplant the original and when it eventually rots the original will still be there.

I have a lot of confidence in the long-term superiority of the non-LLM open-source software universe.

@tiotasram @tryst @aud right so like our ideal world personally consists of many small power structures which hold each other accountable and which have a lot of redundancy across them, so that it's easy for people to dismantle individual ones as it inevitably becomes necessary

@ireneista @aud @tiotasram i am tentatively hopeful that a reasonable operating system can be built without any one piece being so large as to require such a large and complicated governance structure.

the big problem being, of course, a web browser. but that can be put off to some extent through virtualization (like seL4’s “cyber retrofit”).

@tryst @ireneista @aud @tiotasram THE HURD CAN NEVER DIE WHILE THE DREAM LIVES WITHIN EACH OF US!

@ireneista @tryst @aud good point; values is a much better way to think about this.

Sadly although I agree that an anti-LLM stance does logically flow from values of freedom for users, I think there are many who perceive themselves as valuing user freedom but who embrace (or don't strongly oppose) LLMs, Linus Torvalds being perhaps a decent example?

It's not an insurmountable problem, but the same statement of a "value" will mean different things to different people, and in some cases like "freedom" you've got layers of propaganda all over the place trying to distort what that means.

@tiotasram @ireneista @aud yeah, part of that is that it can’t just be a simple statement of values (though those certainly can exist), but also an ongoing discussion and community experience of the values.

@tiotasram @tryst @aud yes. there are two implications -

in any written document that attempts to state the values, it's important to address all the points of tension we can reasonably anticipate; and

since society changes over time, and new issues rise to prominence, the written statements need to change with it

@tiotasram @tryst @aud we've watched what a lot of different groups do and we think it's best to have these things be living documents that regularly get renewed through a process based on grassroots engagement

we think treating them as semi-immutable constitutions to be amended only rarely is exactly backwards, because it's the community processes, the everyone talking about stuff, that give the values both their legitimacy and their adaptability

@aud I wish I didn’t hate writing code so much… Then again, I’m quite sure that maintaining a popular oss project is more or less a guarantee for burnout -.-
@schrotthaufen yeah... there's a degree of me that sympathizes with why some developers/maintainers are turning to LLMs. But then there's also a part that is just like, well... you're throwing out the baby with the bathwater, here.

Not for the first time, the "convenience of applying an LLM" highlights a social problem that would be better addressed in a different manner, but in this case, while there's more agency to select a better solution, there is almost certainly less money and time to do so (as compared to, say, companies pushing LLMs).
@aud Sounds about right, unfortunately.
@aud Hell yeah. Back in the 60’s, my grandma worked for a small company (2 engineers and a few dozen workers) that used to make computer chips in the Bay Area and I’m always like. We need to go back 😂
@aud I feel this so much. I can live without harfbuzz (the privilege of using a latin script to communicate), I can live without systemd and linux even if I liked them a lot. But mesa just doesn't seem replaceable. Most of my own things are wayland related, which now means the vast majority of code I have written in the last five years runs on a stack tainted by LLMs, which is quite upsetting since I don't wish to endorse that in any capacity.
@aud I am toying with the idea of just setting up a simple hypertext based UI system coupled to a lisp runtime running in a BSD framebuffer. Basically a shitty emacs that doesn't need a windowing system and ideally replaces the entire userland. If I can read/write mail with it and have a simple org-mode reimplementation then it would be a viable "lifeboat" out of the slop seas.