Old Fucking Punk @lwriemen@librem.one

It’s interesting that mathematicians see the release of the 700+ purported mathematical proofs (we don’t know how good they are until mathematicians actually read them) as a threat to their existence, and not a tool that can help them move more quickly; and that I think speaks volumes about their recognition that none of these papers were published *to further mathematical curiosity*, but rather as a tech demo for private corporations.

Mathematicians don’t do math like a factory worker makes widgets. They pursue directions that they find *interesting* and that may someday uncover *useful tools* to model natural behavior. Mathematicians didn’t *ask* for Anthropic and OpenAI to help them with math; these companies simply wanted a press release, essentially claiming that their product rendered math academics obsolete. And their release was so voluminous that it would take months if not years to tell whether their claims had any merit—they basically DDoS’d the math world.

#ai

That every single software, app, or platform I rely in has gaping feature and design holes in it – all of which cost me and every other user money in lost productivity – and yet those fuckers keep ignoring the shit they should be fixing in favour of more "AI" features is just constantly incredibly annoying

The shit we're forced to use is full of beginner design mistakes and then they just turn around and keep inventing new ways to break things every day

This industry is so frustrating

Can we coin a fallacy name for the "just use a newer model" thing that LLM addicts subscribe to?

It's such a weird pattern. "Oh, you're using Sloppo F, that's old and no good, you should use Sloppo G". But in 6 months, it's "Sloppo G is no good, try Sloppo H". A model is only good if it's new, apparently!

There are strong elements here of:

- Lack of self awareness
- Failure to observe patterns over time
- Straight up gambling ("the next hand will be better")

The best name I can come up with is the "Latest Model Fallacy". Which is pretty boring. Got anything punchier?

Since people love hate-sharing these, I thought I'd get in on the action. https://www.dailykos.com/stories/2026/10/30/800120153/cartoon/brain-dead-op-eds/

"Writing is Magic; Machine Writing is Ash"

https://www.baldurbjarnason.com/2026/machine-writing-is-ash/

My take on why everybody should be writing. Bad writing. Sloppy writing. Unstructured. Whatever. Writing is magic even when you don't have readers and are writing just for yourself.

Matt Stoller thinks #AI software is going to replace established #saas tools because he vibe coded a utility with no logging, no security, no audit/compliance and very little complexity and but thinks he's 85-90% of the way there and that...

> poorly done AI-built software is just not that much worse than what exists now

Oh my sweet summer child, if you truly think that, you ain't seen nothing yet! Just wait for all the CEOs to f*** around and find out that vibe coding "systems of record" might not be the best idea if you want to keep your S/ME solvent/afloat and free from legal challenges and fines.

I do agree with the main thrust of "SaaS companies are evil, enshittifiying gatekeepers" but to claim AI could soon just magically replace this stuff is completely mad.

https://www.thebignewsletter.com/p/monopoly-round-up-the-2-trillion

/Cc @davidgerard

@kawazoe @zzt I unfortunately am forced to use llms in my job and I also spent many years as an academic studying language models and transformers which, at the time I was excited about but now am basically ashamed of because the llm industry has essentially nerfed any positivity anyone has towards machine learning in general. A lot of my work was showing limitations of transformers and using different models and the refutation was always "bruh we don't need this we have jippity now".

I'm incredibly sad and angry about what Clammy Sammy and his friends and frenemies have stolen from me and my colleagues. I understand LLMs more than most people and that's why it's especially infuriating seeing people claim they are intelligent, that they will become AGi, that we'll solve hallucinations and all those super scammy claims that are basically just not possible with this family of algorithms.

So yes an impure position is possible: I have a PhD in "LLMs" and I write an anti-llm blog. Fuck anyone who tries to draw false dichotomies.

Seeing others who are strongly resistant really takes the edge off the general anger and mourning so thank you for seeing this shit for what it is!

When you pay full price for select titles thru 11/6/26, we’ll donate 40% of each sale to legal aid for the Prairieland defendants and the Minnesota 15.
Let’s show up for them. Shop all titles and benefit merch at pmpress.org or see the link in our bio.

RE: https://tech.lgbt/@lettosprey/117534851823121382

one of my big missing-stuff-but-i-find-it-useful Theories i am kicking around these days is that silicon valley realized -- first with windows, then with the iphone -- that if you wanted to really generate money, don't just focus on software: control the entire computing environment. with AI they set out to make an entirely new, thoroughly proprietary computing environment, but they didn't quite realize that the environment itself isn't where the profits come from

@lettosprey I work in student support at a university. We got told we can get paid copilot access soon if we want it.

Students already feel the university is faceless as staff time reduces and student numbers go up.

I already have to do more email than meeting students due to resourcing. So I try to make emails count; personalised to each student.

I need to read source info myself cos I am already spotting gaps and trying to get them filled. An AI would miss the issues entirely.

Twice as many plants and animals sold in #wildlife trade as previously thought, says report - https://www.theguardian.com/environment/2026/oct/30/twice-as-many-plants-and-animals-sold-in-wildlife-trade-as-previously-thought-says-report "100,000 species are traded globally, whether illegally smuggled, sold as ornaments or in fisheries, timber and fashion industries" awful trade by awful people

So, for folks who have moved from hands-on tech roles like devops or software engineering into maybe tech-adjacent non-hands-on but ALSO non-managerial positions, what did you move to, and how did you approach the problem? Bonus if it doesn't require costly education or training.

Asking for me. Like I have no idea, I've done mostly two things in tech for decades and have zero real skills or experience in anything else that seems relevant. I've siloed my career and I want out, the only other alternative I see is just retiring early and hopefully not starving.

#AskFedi

@zzt My position on genAI will basically ensure I don't have a future in tech at this rate. My current employer is at least willing to tolerate my rejection, but for all intents and purposes my decades-long career is dead.

@muddle @iain @zzt better than that - NFTs came about and he dived in and said "HEEEEEY i invented those"

“I’m anti-AI too, I don’t use LLMs for anything. But”

that’s adorable. I’ve been forced to use LLMs in my professional work or my ability to support myself financially would end. it was a horrid experience. it isn’t the reason why I hate LLMs, that vastly predates me being forced to use them under threat, but it gave me firsthand experience in how fucking backstabby and dishonest teams are when LLMs are used, and how much these things impact people mentally.

it’s great you have the privilege of not being forced to use this crap! some of us are knee deep in shit and we don’t want to be. I know you’re used to speaking for other people, but you sure as fuck don’t speak for me.

that’s why the 404 article about proton resonated with so many of us, while a lot of the privileged fuckers pretending at being anti-AI didn’t even notice it. we’ve all felt that type of abuse. we know what it is to know you’ll be fired if your company finds out you’re not onboard with this horrid crap.

you know, now that I’ve calmed down I get it

if you’re a big name in infosec or programming, you’ve made it. you don’t feel the same type of anxiety as someone who’s been threatened with being fired or has been fired for not complying with an AI mandate. the type of horror we saw in the 404 article about proton is theoretical to you.

but the horror isn’t theoretical to me or to anyone else who’s struggling, whose career is precarious due to this crap.

you declare yourself to be anti-AI then declare that the anti-AI people have gone too far, because you were never in any real danger.

and at this point I’m not just talking about that infosec dickhead from yesterday. you’ve seen big name programmers do this crap. anil dash (but just a dash of it, please), simon willison, that one rust fucker I used to like, torvalds

all of these people are comfortably wealthy, and their careers are in no danger. of course they think we’re being irrational, or religious, or doing purity tests.

Blogs recently:

"I wrote this cool project I've always wanted. I used a coding assistant"

"I did this detailed analysis of something fun. I wouldn't have written 5kloc to do this if llms didn't make it so easy"

Meanwhile, oligarchs:

"Your computer shouldn't have any ram; it should just be a terminal for our ai service"

There is something deeply irresponsible about these bloggers and folks asking for nuance about llms; it's like that guy who says "I've only had one drink, I can still drive"

Agriculture $200B
Health care $150B
Military $800
Chatbots $3,600B
Utilities $150B

someone who is good at the economy please help me budget this. my country is dying

our no AI notice/splash screen which is available for free now also has a dark mode version: https://eniko.itch.io/cute-fox-no-ai-notice

with hand drawn lettering and a relaxing cute fox to put your audience at ease. options include with/without fox, opaque/transparent, and 4k, 2160p, 1440p, 1080p, and 720p resolutions

The executives and investors and lawyers and crisis management (PR) experts are throwing some kid under the bus they're driving in flames off a cliff, not expecting there to be any survivors at this point.....

But they're in a full panic because any competent incident responder would document how this incident was a failure of leadership, and this has to be prevented at any cost.

Their individual careers and dignity must be preserved, so let the victims all burn.