Replying to
“no you don’t understand it’s weird and janky and broken because” no I understand perfectly thank you
Remote
[object Object]
@zzt@mas.to
mastodon 4.7.2infrastructure admin for awful.systems, where the jank is a feature (tm)
2299 Followers
326 Following
50 Posts
Joined November 04, 2022
lemmy:
writefreely:
remember: if you want to do something incredibly unpopular like use LLMs in your project, start a discussion thread and make sure to flood it with the most noxious pro-AI guys you can find. if some of them are racists, that’s perfect! then when the rest of the community reacts strongly to you wanting to do something horrid and the discussion thread about it being filled with horrid people, lock the thread because it’s “too heated”. then you can do whatever you want because the discussion has concluded!
isn’t that right, KDE?
every CEO when they get on a podcast: “I have a problem with AI”
CEO: “the problem is we’re not using enough of it and my employees aren’t taking enough personal responsibility for when it’s dangerously wrong”
Boosted by @jwz@mastodon.social
Replying to
apple announced a major OS update whose only features were AI and surveillance (which are one and the same) and you clapped
it’s not too late to do better, but you have to do better
when your government or enterprise consists entirely of virtual pets you raised yourself, that’s neopetism
“our discord full of the worst people you’ve ever met said we should go all in on slop and that means the community has spoken” yeah i see
Replying to
it’s opt-in, but the features will mysteriously toggle themselves on constantly, a thing every corporation doing AI loves to do to juice their numbers
so you opt out, but you’re not actually opting out of any of the privacy violation or resource consumption you care about
https://arstechnica.com/cars/2026/09/aftermarket-driver-assist-under-federal-probe-following-fatal-crashes/ ahahahaha after that ass-clenching video where that one guy installed comma in his rivian and thought it was amazing because it hesitated less than the rivian self-driving and it was open source and he utterly disregarded that it blew through a stop sign and kept doing dangerous shit, it’s very satisfying that the fucking obvious consequences are coming for the horrid open source self driving device that manages to be even more horrid than the commercial ones
Replying to
do you see how this works
have you fully absorbed that this isn’t the same apple you were a fan of?
it’s fine if you got here late
it’s not fine if you still insist there’s no danger
Replying to
please don’t tediously reply to this thread with a bunch of horseshit copied from apple’s marketing about how it’s good for journalists or “evidence” when there’s a ton of other technical and non-technical ways to authenticate doctored images, and none of those get fed to the slop machine
Replying to
this is the thing we were shouting about, it’s happening now
your consent materially doesn’t matter to apple
do you understand why we were shouting? why aren’t you shouting?
can’t believe the K in KDE stands for klan
Boosted by @jwz@mastodon.social
Replying to
@davidgerard@circumstances.run also, you’ve gotta love the typical polite fascist trick:
no, this isn’t akin at all to randomly punching someone in a restaurant and getting thrown out. this is akin to overhearing another restaurant patron spewing hate, calling them a nazi, them going “damn right I’m a nazi”, and then the owners of the restaurant throw you out instead of the nazi
it’s real weird how the facade of neutrality breaks when you ignore the weird, incorrect implication that the accuser is guilty of any kind of violence
every number is imaginary if you think hard enough
Replying to
@davidgerard@circumstances.run “What happened is akin to someone walking into a restaurant and seemingly randomly punching another patron. They're going to get thrown out.” not even slightly, but ok. I think walking into a restaurant and punching someone spewing hatred who admits to being a nazi is a favor to the restaurant, actually
Replying to
I may be reading too much into it, but it always felt grating that none of these emoji ghouls could write “crafted with love”, an ordinary thing I might say about an actual project of mine. no, it was always “crafted with heart emoji” because love was too strong of a word
and now these people love the automated emoji ghoul because it lets them feel like 🤔🤯 is a complete thought
Replying to
an unbroken streak of making emacs materially much worse for literally no fucking reason
your programs will be faster sure but debugging might maybe be marginally harder and also a couple of data structures would have to change
emacs lisp isn't like machine code except for when it gets compiled into fucking machine code
this post is from 2026
why don't you stick to something you know better than emacs or lisp, muffins. I'm not sure what that is to be honest, but emacs and lisp ain't it.
Replying to
Replying to
it’s weird how open source projects banning the inclusion of license-incompatible code didn’t lead to witch hunts and bullying
almost like every time we see anti-LLM people get accused of doing witch hunts and bullying, it’s actually the pro-AI guys crybullying because they got told no
Replying to
- you can’t afford any new tech products, but none of them are for you. the only product categories are revenge porn and surveillance, and of course they overlap.
- none of the corporations doing IQ and neurodivergence tests will make the connection between their products being widely hated and their policy of only hiring people whose brains excel at bullshit tasks
- none of the people cosplaying as psychologists will draw a connection between discriminatory hiring pipelines and bullshit products made by people addicted to imposter machines. the numbers they made up to justify all of this still look fantastic.
- nobody will reflect on anything at all, actually. reflection is obsolete.
Replying to
imagine living in 2026 and thinking your lisp shouldn’t have TCO
surely nobody will get bit by expecting a fucking lisp to have god damn TCO in 2026 and having their editor run like shit or crash when it doesn’t
“oh it’s harder to debug” then don’t use it???? you don’t have to use recursion if you’re afraid of it muffins
Boosted by @kcarruthers@infosec.exchange
Replying to
whenever I chant “bell labs speedrun” people think I’m talking about unix and transistors instead of a damning pattern of systemic academic dishonesty that nobody involved ever fully recovered from
between unix and a replication crisis, it should be a bit obvious which one we’re going to get from LLMs, right?
Replying to
@davidgerard@circumstances.run it’s literally the Kehoe Rule every time, isn’t it. you can’t prove that leaded gas is harmful, and also all the proofs of harm are out of scope for whether or not we should ban leaded gas
if unicode is so good where’s multicode
Replying to
ah, there's muffins:
From: Richard Stallman Subject: Re: bytecode tail call optimization Date: Sun, 04 Jan 2026 21:56:30 -0500
[[[ To any NSA and FBI agents reading my email: please consider ]]] [[[ whether defending the US Constitution against all enemies, ]]] [[[ foreign or domestic, requires you to follow Snowden's example. ]]]
I am currently looking into the bytecode compiler. It seems that there's no tail-call optimization, is that intentional? Would it be a welcome change?
Tail call optimization would make programs faster but would make debugging harder. All in all, I think we are better not doing that.
As for implementation, Emacs bytecode is not like machine code. An optimized tail call in machine code can be as simple as adjusting the stack a little and jumping. For Emacs bytecode, several data structures about function call frames would have to be fixed.
Replying to
@davidgerard@circumstances.run the rationalists call this tulpamancy or rocko’s modern basilisk or whatever and to them it’s more real than the actual jail they deserve
Replying to
@jwz@mastodon.social today is an excellent day for endless screaming. thanks for elisp byte compilation that doesn’t suck by the way!
Replying to
code for anyone having trouble picturing this (I guess I'm the documentation now):
(named-let recur ((x 0)
(l nil))
(if (< x 10)
(recur (+ x 1) (push x l))
l))
Replying to
that’s real by the way https://x.com/anilkseth/status/2101796550473073128/photo/1 (warning: twitter link, feel free to reply with an accessible archive for it if you’ve got one)
Replying to
some people will tell you this isn’t how it is but they’re just trying to sell you on how it isn’t
Replying to
@ireneista@adhd.irenes.space I’m pretty sure I have ADHD second edition
anyway roll for THAC0
Replying to
I’m laughing now but can you imagine how vicious the publish or perish parts of academia are going to get once having your name on some shit like “can large language models feel pain”, a sequel to the paper “can large language models feel fear”, is rightly seen as a poison pill for your career as a researcher
Replying to
@davidgerard@circumstances.run you don’t want… miserario? the marvel comics villain stand-in for clinical depression who appeared for like 3 issues in the early 90s and is rumored as a background character in the upcoming Doctor Strange movie?
I’m looking to add easy toy languages to my language workbench’s example set and on the wikipedia page for m-expressions (horrid things, but I’ll probably write a parser for them) I found “sweet” t-expressions from the 2013 Readable Lisp S-expressions Project and uhhhh
https://readable.sourceforge.io/ of course they keep citing Paul Graham in this and their SRFI https://srfi.schemers.org/srfi-110/srfi-110.html
the curly bracket infix notation is ok! the rest really hasn’t aged well
here’s the example from their site showing the uhh improvements? and then it goes off the rails when the SRFI introduces the sublist operator and it becomes the parts of Haskell I struggle to understand at first glance
Replying to
@davidgerard@circumstances.run yes I know that if miserario were real he’d be a gay icon
if you like lisp so much, where’s your IBM 704 vacuum tube mainframe?
the problem with chlamydia is it sounds wonderful
it’s not, but it’s a beautiful name
Replying to
@oxy@social.bsdlab.au @davidgerard@circumstances.run now now, screaming “I AM MISERARIO” while emitting a burst of radiation is how you ended up in jail in the first place
Replying to
we can’t have TCO I guess so here’s a weird replacement
it works exactly like let but the extra NAME variable is the name of the function you call to do tail recursion
so for example, if you set NAME to recur then you have an exact copy of the clojure tail call feature, and using it won’t blow up your stack
I don’t hate this, I just wish it was documented anywhere