While most of this thread is very revealing and traditional sealioning from someone using the fig leaf of a "nuanced, moderate position" to justify selling out the software commons, I have to admit that "our permissive LLM policy is for weaning people off of using them" is a novel take I haven't heard anywhere else. Props for that, I guess! Too bad it's at everybody else's expense.

https://mastodon.opencloud.lu/@paolo/117518911250763421

So yes. Paulo and the rest of The Document Foundation have solicited money with a blog post called "Yes, no AI is now a feature", put on the posture of a principled stand, and shipped slop to everyone who keeps up to date via their package manager. And yes, if that was your money, given on the premise of that simulated principled stand, you have been defrauded. I stand by that statement more strongly than ever.

RE: https://fosstodon.org/@libreoffice/117518007825825230

Profoundly embarrassing behavior from the org that solicited donations on the basis of being a shelter from LLMs.

I say again. Demand your money back, and tell your friends to only use older versions. Regardless of all other issues, this is a picket line, and if you don't learn not to cross those for other people, nobody will have your own back when you need it.

I think sometimes that Mozilla would like to be making almost anything other than a browser.

And I mean, not an email client either, the open disdain for Thunderbird has been a black eye for the company for like 8+ years.

But what Mozilla's greatest hits have in common is that they are product categories which *should* (especially in 2026) be *boring.* Which is deeply antithetical to their current leadership's attitude that feels ripped from a proprietary playbook - catching people's attention with unique, differentiating value that keeps them from using Chrome, Safari, Edge, or one of their own forks. Every gooberish product decision can be traced to:

1. The question being "what can we offer, that people have to come to us for?"
2. Not accepting "be a trustworthy, portable, obvious set-it-and-forget-it appliance" as a valid answer.

Prego is still one of those brand names that's a never-ending font of cheap comedy. I'm having a pasta and it's yours.

Last post before I have to get busy on real stuff today, but...

SINGLE LEADER PROJECT: Risks being a cult of personality, but crowd feels empowered to reject leader and fork if leader acts like a chud.

FULL DEMOCRATIC PROJECT: Needs good voting protocols, but can be awesome. Schisms in project leadership represent genuine schisms in community because those two are the same thing. Forks are not difficult to organize because labor organization/discourse are already established.

COMMITTEE LEADERSHIP: Amorphous, semi-entangled with community but not usefully so, divine right of democracies rhetoric without democracy, awkward to reject when they sell you out to the scab-o-matic or other egregiously dumb decisions. Usually build infrastructure investment which forks cannot match and which projects depend on. Basically the devil.

Oh and I have to keep saying this, but I do not want to hear a fucking word about how "projects need committee leadership, that would solve these problems" when COMMITTEE-LED PROJECTS DO THIS CONSTANTLY.

While usually being irreplaceable because of infrastructure investment in ways that individual-led projects are usually not.

I hate to have this chip on my shoulder but I'm so done at this point.

Okay. If the LibreOffice promotional posting about their "keep AI optional so people can do office stuff without it" stance was persuasive to you, and you gave them money, you have been defrauded. And you should demand your money back. And we should all be ever more skeptical of carefully-worded "nuanced stances."

No money for quislings.

https://circumstances.run/@davidgerard/117510420175177743

https://blog.documentfoundation.org/blog/2026/09/03/yes-no-ai-is-now-a-feature/

If you contrbuted anything to Debian or Bevy or the Linux kernel, you have a more valid claim to "look what they took from you" memes than any Greek statue PFP motherfucker alive.

If people contributed to a project with the understanding that it would contain LLM contributions, that's their prerogative, and some people are fine with that. But the reality is that most of these projects are older than that. People invested their effort on a different social contract, and it was changed out from under them. It's like relicensing but more consequential.

I think that's a much bigger theft than scraping. Scraping doesn't steal our communities.

What happens when you only get 12 or so function declarations in your stocking for Christmas and you need to make them last all year round, lest mother yell at you again. We must be thrifty to survive the bitter St. Louis winter this year, and there is always a job opening for naughty children who can fit inside the machines at the factory down the hill.

Thankfully not everything in my life is meatspace chaos right now. Some of it is reading the man pages for networking system calls and rolling my eyes at the grab bag of disparate models that all got dumped into the concept of a "socket," which makes the docs a dense thicket of the favorite words "if" and "unless" and "when," which in turn can only filter into our simple three dimensional world by ignoring most of it and thinking in a single use case slice at a time.

Welp. It wouldn't be the C standard library without some crimes, now would it?

Hey, so, crazy life/housing security stuff going on right now, through literally no fault of our own. Might blow over, might not. I'm probably not going to post much for awhile or make progress on projects while me and my wife circle the wagons and get our fundamentals established. We are on track to be fine thanks to local connections, don't worry about me too much, but I probably won't do any walk recaps for a bit. I may not even live near the golf course next week! Fun times!

Thank you for your patience, yadda yadda yadda, progress will resume when things are stable again I guess.

So if you want concise precedence that isn't an a priori lookup table, there's... a thing, you can do.

assert(4+1 * 3 == 15)
assert(4 + 1*3 == 7)

Allow a layer of implicit paren-ish grouping for non-whitespaced operators, and within groups, operations flow left to right.

assert(4+1*3 == 15)

What about unbalanced whitespace? Well, this is used for unary postfix and prefix operators, which are allowed to have distinct behavior, but some have obvious choices.

assert(-1 == 2-3)
assert(5 + -1 == 4)

The crazy thing is this gives us unambiguous dereference syntax for free, for things that need it.

So awhile ago I was a bit concerned about a Prone syntax decision, because of the following tension:

1. Gosh it would be nice if the parser could be stable forever, even as operators come and go in various stdlib editions. Both concatenative and LISPy languages are good at this.

2. People really like their operator precedence, though, because it makes it easier to translate math from proper mathematical notation. And precedence is kind of an arbitrary table that relies on knowing about all operators.

Well I just figured out a clever solution. I'm not married to it. The problem is that it seems bioengineered to piss off as many people as possible. I remember how people felt about significant whitespace when Python did it...

#pronelang

Clearly the world is in the state it's in because we didn't listen to them and we should have listened to them more.

True taking-the-advice-of-big-tech-security-consultants has never been tried.

I've heard that cheap RISC-V chips and small lithium-ion batteries are mostly being used to drive little LCD screen animations on vacuum-driven heater element rigs...

But I don't know about all that. Sounds like some vaperware to me.

Womp womp

Well. I think I just figured out where the "Posted from ${appname}" bit at the bottom of toots come from. In fact, if this status actually submits properly, I think I'll have conclusive evidence that it's one of the most annoying (but, from Mastodon upstream perspectives, reasonable) answers possible.

For the record, this status was posted without using madonctl at all.

Late addition, but I do think conciseness is the best measurement of "leverage" as I used the term above. You make a lot of things happen on your behalf by asking a little. And we're all used to having leverage, even if it's not always paired with a lot of control - you can start a Windows computer and an unholy plethora of services and network requests will happen on your behalf, each of which involves millions of function calls and kernel operations, all for the mere price of pressing the power button.

Conciseness is maybe a better thing to put on a pedestal than leverage, because it's a more specifically *thoughtful* kind of power. Like my microframeworks/microlibs conversation the other day. "What's the shortest thing I can say that allows the subsequent code to be dramatically shorter?" Which is still imperfect on its own. Because you want to minimize cognitive real estate in the human mind as much as you want to minimize line count.