Welcome to November, and welcome to the Stubsack, your first port of call for learning fresh Awful you’ll near-instantly regret.
Any awful.systems sub may be subsneered in this subthread, techtakes or no.
If your sneer seems higher quality than you thought, feel free to cut’n’paste it into its own post — there’s no quota for posting and the bar really isn’t that high.
The post Xitter web has spawned soo many “esoteric” right wing freaks, but there’s no appropriate sneer-space for them. I’m talking redscare-ish, reality challenged “culture critics” who write about everything but understand nothing. I’m talking about reply-guys who make the same 6 tweets about the same 3 subjects. They’re inescapable at this point, yet I don’t see them mocked (as much as they should be)
Like, there was one dude a while back who insisted that women couldn’t be surgeons because they didn’t believe in the moon or in stars? I think each and every one of these guys is uniquely fucked up and if I can’t escape them, I would love to sneer at them.
(Credit and/or blame to David Gerard for starting this.)
AI 2027 tracker has updated again
-On the main page chart, they decided to map the Anthropic unreleased models from the METR report for some reason. Setting aside how mapping unreleased models is inherently flawed because we don’t have the full picture, the METR report also said that their measurements come with a high uncertainty
Onto the list of predictions
-Late 2026 is at 84%
-Still doing that weird thing where 80% is both partially accurate and confirmed accurate
-The stock market prediction should be lower, since it didn’t get anywhere near 30%
-Imo the AI narrative prediction should also be lower, since theres conflicting reports about AI being amazing/AI being trash all over mainstream news
-One prediction got a 100% score even though part of it didn’t happen (under OpenBrain releases Agent 1 mini: “The part that did not happen is OpenBrain blowing away the competition: several labs shipped cheap, near-frontier models within weeks of each other.”)
-The key metrics seem to be mostly based on estimates rather than confirmed figures, so likely subject to change
Overall seems like they got a lot of ‘easy’ stuff right (ie business gurus shilling AI) but missed the harder stuff
deleted by creator
Anthropic employee has had their brain destroyed by rationalist circlejerk
‘we need to make armageddon happen as fast as possible even though I dont think it will be very good. a trillion dollars please’
I’ve been going down a Weird Little Guys archive binge for a little bit here and this eschatology is even less hinged than the most ridiculous accelerationist nazi terror screed.
They have not called their elected representatives to inform them that this is what they believe
“Hello, Ms Congresswoman? The computers! They’re going to kill us all! The only way to save humanity is to spend a trillion dollars on killer computers! I work for a soon-to-be publicly listed company making maybe-killer computers.”
A tragedy in 3 acts.
2026-03-05 Why I Don’t Vibe Code
Like many developers, I have a whole folder of draft hobby projects that have never been finished. […] From the LLM perspective, these are folders of failures and I could indeed use LLM to make an app a day or whatever challenge I want. However, the process was far more important than the product (again!). Not every whimsy needs to become a reality. Often, I get more from the fun of brainstorming and the process of learning enough to know that I don’t need to continue and finish the job.
2026-04-29 The LLM Is Not a Junior Engineer
Contrary to how I might seem at times, I do think there could be a place for LLM models in modern software development. I do understand why some developers are entranced with the ability to create applications in hours and turn prototypes into products. I see how it could be highly useful for home hobbyists, back-office bureaucrats and expert engineers to build the low-stakes software they’ve never had the time or skill for.
2026-10-25 My Local LLM Stack
And yet, as a developer, I do feel like I need to work with these tools to understand their potential for augmenting my own software development work. There are some cases (basically as a limited coding assistant) where a LLM can be useful. Some recent examples I have used across several projects:
- Look through this codebase and find all the areas where I have
# FIXMEcomments. Collect them in a report and analyze if any no longer apply. - Improve the test coverage for this file to be above 90%
- Update the openapi spec for this project in the repo to match what the API actually implements today […]
- Look through the data here and find all cases where there are likely errors. For instance, if a person has an end date before their start date.
tangential sneer
Now that I had a machine, the next step was finding an ethical model for coding purposes.
Apparently IBM Granite is the most ethical model out there (read “ethical” as “least likely to get you in legal trouble by using its output”.) With all this agentic power at our fingertips, what are we going to use it for?
The 8B model can do many simple programming tasks (like searching through syntax trees or renaming functions).
Who needs birdshot, when we can aim 16-inch cannons at every sparrow?
This reminded me of @rhamph’s post last week. Is this an archetype of guy that is concious of the harms of LLMs, but cannot trust his own gut enough to see they are fundamentally wrong? So they try to come up with their own super-ethical spin on the beast and try to change the rotting carcass from within. But was anything gained here? I feel that an ideal of simple, elegant, and empowering code was lost.
- Look through this codebase and find all the areas where I have
An interesting question from mcc:
I wonder if at this exact moment one could succeed in filing a issue against an open source project for an entirely ficticious vulnerability, get it assigned a CVE and a github score, and then get the upstream project to accept your patch for the vulnerability, with all these things presented as the issue, the CVE, and the PR being written by generative AI, but in fact you wrote them substantially by hand and the patch for your nonexistent vuln is actually a backdoor
Very probably the more your submission appeared to be written by generative AI, the more likely it would be to be accepted. If some person came in with an ungainly, unreadable patch and was aggressively pushing a project to accept it you’d get suspicious and wonder why but if an AI did that you’d be oh well AIs are just like that and AI written code is just like that
deleted by creator
I am beginning to like the cut of this publications jib.
https://defector.com/there-is-no-place-in-a-functional-society-for-elon-musk
The federal government is in tatters, the judiciary seeded with sneering MAGA goons like bear traps on a forest floor, the executive branch a metastatic growth of varyingly murderous criminal enterprises in unholy union with a psychotic, apocalyptic tech industry in suicide-cult thrall to bumbling chatbots.
Happy Saturday. Pleas enjoy a very schadenfreude GitHub issue
But Sam Altman said AI is smarter than the smartest human scientists! I wasn’t expecting shutting down a server to be one of the deep, intractable mysteries of the universe, but here we are.
@samvines @BlueMonday1984 wait is this guy blaming it on using the wrong version of the idiot robot?
https://github.com/openai/codex/issues/59060#issuecomment-6601721599
You haven’t tried shutting down a remote server and having it wipe out the remote machine instead in the last 3 hours? Omg bro the newest models would never make that mistake! Just one more model bro!
That should cause an issue like this I assume.
Should have taken the effort to type ‘shutdown -r now’ yourself.
Prob even quicker than asking an llm.
Ιt could have been part of a larger thing where it made sense to want to keep evidence of the shutdown in the chat context.
It’s not like they design or promote these things in any way that encourages user initiative, can’t have full cognitive surrender if the user is running system commands on the side.
Good point, had not thought about that. And that makes, in a way stuff even worse, as when it goes wrong people go all ‘skill issue’ despite the systems being designed and marketed for full cognitive surrender.
I will assert again that ~nobody who claims that prompting is a skilled practice also offers a guide to how to develop that skillset. And it would be effectively the same claim as having cracked the use of English as a computer programming language, so it should be pretty easy to produce that guide, right? Anthropic should have this shit front and center! First clickable link on their front page!
Too many good sneers in this one.
Also, it was only yesterday that I learned that several great conservative thinkers on X thought that making a Harry Potter x Star Wars crossover with a genderbent character would be an exceptionally powerful cultural message that would greatly influence the outcome of the midterm elections. https://www.404media.co/republicans-learn-x-is-an-echo-chamber/
There is a theoretical timeline shimmering somewhere out there in the void in which humanity figured out how to roll out LLM technology - which does have many genuine positive uses
makes wanking motion with right hand while rolling eyes
uwu pls don’t take away my sense of technological progress - it is the only thing that gives meaning to my life and culture
that making a Harry Potter x Star Wars crossover with a genderbent character
And evil characters they claimed were right even. This is just ‘hitler did nothing wrong’ with extra steps.
There is a theoretical timeline shimmering somewhere out there in the void in which humanity figured out how to roll out LLM technology - which does have many genuine positive uses
God damn it
@istewart@awful.systems Dave Karpf is coming out with a book which will be of interest: Vaporware, Inc.: How Silicon Valley Sells the Future to Control the Present (W.W. Norton, In Press) bookshop.org - Amazon
For thirty years, techno-optimists have promised dramatic technological innovation as long as we–the public, journalists, regulators, and policymakers–keep out of the way. The results have been great for Silicon Valley but haven’t gone so well for everyone else. They promise to stop climate change while building resource-intensive data centers, evade governmental regulations at our expense, and watch wealth inequality soar as they enjoy a permissive tax code. Venture capitalists and the companies they fund have amassed wealth, status, and political power, but their bold pronouncements have amounted to little more than vaporware.
In Vaporware, Inc., political scientist and tech commentator David Karpf looks back at the history of the digital future to understand the ideological project of techno-optimism, and why it has gone so wrong for the rest of us. With sharp wit and keen insight, Karpf debunks the foundational myths of the techno-optimists and traces how tech billionaires amassed power by selling the idea of the future. Eye-opening and essential, Vaporware, Inc. is a critical reappraisal of how Silicon Valley works and how we can make it work better.
Anthropic appears to be suffering from low $T
Anthropic is going ahead with their IPO
This is looking like its going to be a bloody disaster but if the world wants to feed my schadenfreude by all means keep it coming
Struck some gold on slobstack
this is good for bitcoin
you unfortunately have to dig quite deep for sneers in this weird post about datacenter votes in the recent US elections, but they’re there
especially “Noosphere89” smugly stating that datacenters will just move somewhere else, like Iceland (retracted after someone points out that Iceland is probably even more woke than, say, Vermont) and Paraguay (which recently had a digibank seized and closed, leading to the president stepping down).
Then there’s this gem
Data center inputs, unlike almost anything else are much more subsitutable than any other input, and can be built in places where humans don’t care to raise NIMBY concerns.
So cutting through the jargon, “inputs” in this case means
- power
- cooling
- construction / planning
- the actual fucking contents in the racks
- data
All of these are constrained in various ways. The places with abundant accessible power tend to be relatively developed parts of the world. Same with water which is the most efficient mode of cooling.
Building your datacenter in Texas means you’re on familiar ground with local laws and regulations, and can draw on skilled manpower to actually construct the thing and keep it running. I’d wager that the actual hardware is cheaper than almost anywhere else. Once it’s built it’s in the security zone of the USA, which means property is efficiently defended by LEO and emergency services.
Paraguay? You need to find a skilled workforce there, or import it. Same for all the bits and bobs needed. Then you probably have to pay more for the same, as well as having to deal with an unfamiliar planning process. Not to mention political instability. And you might have to build a shit-ton of supporting stuff yourself, like power generation and fiber comms.
These people are delusional.
especially “Noosphere89” smugly stating that datacenters will just move somewhere else
there’s no fucking way people will happily accept data centres near them after Big Tech has so happily made clear that they’re on the side of MAGA and fascism
Also I can’t imagine the MAGA crowd being excited about their tech minions operating in someone else’s territory. Data sovereignty is all fun and games until some Paraguayan commie starts getting chummy with China or whatever.
iceland also has gdpr, and i’d like to see how it fits with data thievery of ai companies, but they prefer to put excess electricity generation into aluminum smelting
especially “Noosphere89” smugly stating that datacenters will just move somewhere else
Why didn’t they start out there then?
I think they’re saying that stopping a datacenter won’t do squat to stop AGI takeoff
at least this a reading of
Efficiency gains from AI are absolutely massive, and while they don’t apply to training, they do apply to inference, and even accounting for endogenity, state bans only slow AI progress by 9 days at the absolute most pessimistic.
so we can re-state this as a form of Pascal’s wager:
- if I stop a datacenter near me, the world will still be taken over by a superhuman intelligence, albeit 9 days later, and my local environment will be moot
- if I stop a datacenter, and no superhuman intelligence appears, I won’t have a datacenter near me
I’m not Blaise but I feel that stopping a datacenter is a win, all in all.
state bans only slow AI progress by 9 days at the absolute most pessimistic.
I don’t know, I think that’s insanely optimistic
And this is why they invented the basilisk.
Sometimes you miss a few shots. Well, sometimes you miss 40 shots in a row with those anti-data center ballot measures. I’m sure their streak of bad luck will end eventually and they’ll find tons of people who love having data centers near their homes.
just bribe 'em, bro!
They’ll just skip straight to the Dyson Sphere, dontchyaknow
“How to be Rational: Use Drugs”: https://www.lesswrong.com/posts/8qYuy9Cr6BNpgjJjT/help-your-friends-get-adhd-and-depression-meds
this only works when you are veterinarian
–
there is a case for broadening access to mental healthcare, this is not it
actually, no – i’m not a veterinarian, but i follow these rules and it works beautifully. but i’m in the position where i can actually afford paying those few hundred zlotys a month.
i take you’re not diagnosed by a chatbot, because this is part of rationalist healthcare
Dont think that getting people who need those meds in meds is bad actually. (Do go through the proper channels however, me getting some therapy with meds was quite useful).
