Skip Navigation

rook

@ rook @awful.systems

Posts
1
Comments
191
Joined
3 yr. ago

  • Featured

    Stubsack: weekly thread for sneers not worth an entire post, week ending 26th July 2026

    Jump

  • This is funny yet also awful: adversarial tokenmaxxing suggests that writing everything as l33t$p34k increases the cost to process a document with an llm because the initial tokenisation step produces far more tokens.

    This seems like it shouldn’t be too hard to work around, if it became commonplace (which it won’t) but the prospect of any anti-llm places doing this in the meantime does not spark joy.

  • Featured

    Stubsack: weekly thread for sneers not worth an entire post, week ending 26th July 2026

    Jump

  • I’m deeply suspicious of this whole thing, because it looks a lot like a marketing exercise showing off how dangerous and powerful and autonomous their product is.

    Also, “sandbox” is one of those words that the llm companies have ruined, because they use it to mean a strongly-worded sentence telling an llm not to do something.

  • Featured

    Stubsack: weekly thread for sneers not worth an entire post, week ending 26th July 2026

    Jump

  • The author tends to cull his own social media posts pretty aggressively if they get too much attention, or the wrong kind of attention, and it’s possible there’s more he wanted to say in the blog post but he chickened out at the last minute and self-censored it.

    But yeah, it felt like half a post.

  • Featured

    Stubsack: weekly thread for sneers not worth an entire post, week ending 26th July 2026

    Jump

  • Tales from Earthsea is also good.

    I didn’t mind The Other Wind… it isn’t as good as the others, but that’s a tough act to follow. It does draw together a bunch of plot threads from the previous books and provides a kind of closure, which I appreciated. There was more to say of Earthsea, but le Guin wasn’t going to have the time or energy for that.

  • Featured

    Stubsack: weekly thread for sneers not worth an entire post, week ending 26th July 2026

    Jump

  • Nah, this isn’t the endpoint… you want something like prompt2exe, which is a performance art piece. Gets an llm to output machine code, and bundles it into an executable.

    no package dependencies and invokes no assembler, compiler, or system linker.

    Hardcore mode. No handholding for puny human devs by their tool chains. 🦾

    Probably impractical for normal people to run, unless they have an employer who wants their minions to be burning all the tokens and pays up for the expensive kind of accounts.

  • Featured

    Stubsack: weekly thread for sneers not worth an entire post, week ending 26th July 2026

    Jump

  • Daft ai project of the week: “lore”, a version control system that should not be confused with another vcs of the same name open-sourced by epic games a few weeks ago.

    lore does not track your code. it tracks the prompts, notes and decisions that produced it. you commit intent. when you want code, you run lore materialize: it replays the accumulated intent into a brief, and an agent reconciles the working tree to match it.

    the code is build output. the intent is the source.

    …which would be great, if the output of llms were deterministic, and generally it is not. This means rolling back a change to undo or fix something is just another spin of the roulette wheel, with the analogy being reinforced by the number of tokens you’ll have to buy and burn to do a rebuild. It looks like it still needs a real vcs behind the scenes, given that it isn’t entirely self-hosting, but maybe I misunderstood something.

    You might wonder if this is some kind of satirical work, or perhaps piece of performance art, but I’m fairly certain it isn’t… the creator and sole human author (naturally, claude is the only other “contributor”) is a big fan of something called “open audio protocol” and a project called “audius” which came about by someone asking what if spotify and soundcloud, but on the blockchain? With ai agentic integration?

    https://github.com/lorevcs/lore

  • Stubsack: weekly thread for sneers not worth an entire post, week ending 12th July 2026

    Jump

  • I’m more familiar with vox day than I’d really like. He hasn’t pivoted from the culture war stuff that he’s known for, but he has branched out into ai music and video these days.

  • Stubsack: weekly thread for sneers not worth an entire post, week ending 12th July 2026

    Jump

  • More ai stuff, this time from flathub: Democratizing Abandonware.

    Flathub has a fairly relaxed ai policy that both ai bros and strongly anti ai people are unhappy with. It was brought in to try and deal with the review burden of slop submissions where no human is involved, and a chatbot fields review comments.

    Turns out that ~75% of submissions that got a slop tag were abandoned… not just the submission, but the entire git repo behind it, too. The author is quick to point out that this is far from a representative study, but I can certainly believe that a) people who have invested little time or effort into their slopware will abandon it without much concern, and b) things like openclaw could definitely submit bullshit packages that are immediately forgotten as its internal state moves on. There’s no malice in the same way there’s no intent, just shitty tools being left running and polluting everything around them.

  • Ecosia’s odious greenwashing — now with AI

    Jump

  • what they can do is mitigate the effects of those data centers

    Oooor, and I know this is a bit of a wild idea, they could instead not offer llm search.

  • Stubsack: weekly thread for sneers not worth an entire post, week ending 12th July 2026

    Jump

  • I’m hoping my own qualifications sufficiently predate the llm era that I’d be safe from that particular filter, so I’ll only have to worry about being too old and/or too expensive.

  • Stubsack: weekly thread for sneers not worth an entire post, week ending 12th July 2026

    Jump

  • Stubsack: weekly thread for sneers not worth an entire post, week ending 12th July 2026

    Jump

  • A couple of bits of nice ai news recently, for anyone who hasn’t come across them already:

    Bosses Horrified as “AI Native” College Graduates Hit the Workplace

    new hires who were seen as “AI natives” are turning out to have alarmingly shallow ideas. So much so, the anonymous finance worker admitted, that his firm now actively avoids seeking out AI-literate STEM graduates, and opts to comb through humanities students instead.

    “We want critical thinking, not just AI,” the financier told the FT.

    I can’t help thinking that, funny as this is, the people who are really going to the be worst off here are a bunch of new grads with a load of debt and an education that has made them less able to do anything at all. They’re not all going to be grifters, after all.

    Meta's Zuckerberg says AI agent tech progressing slower than expected

    This is brilliant. They’re making so many mistakes they’re actually having to admit it. It’s amazing how incompetent zuckerberg is… late to every fad he’s tried in the last decade and fucks it up when he finally gets there.

    In retrospect, he said, the "trajectory of the agentic development over at least the last four months hasn't really accelerated in the way that we expected," and ​that the company's bets on the new structure "haven't come to fruition yet." Zuckerberg was referring to AI agents, automated systems that can ​execute tasks on behalf of a user.

    Conversations he was having "with our top people" when they started planning the restructuring in January and February "were that they ‌were ⁠worried that we weren't going to move fast enough to adapt," Zuckerberg said.

    I’m sure there was a third thing, but I found it yesterday when the site appeared to be down (at least for me) and now I can’t remember it or spot it in my million open tabs.

  • Stubsack: weekly thread for sneers not worth an entire post, week ending 28th June 2026

    Jump

  • Here’s one from the archives: https://spectrum.ieee.org/children-beating-up-robot

    Children Beating Up Robot Inspires New Escape Maneuver System

    tl;dr: small children have no empathy for your robot and will torture it and hound it to death if they can. The safest thing to do is to head for the nearest adult-sized people, who will hopefully be less inclined to kill it simply because it is a robot.

  • Stubsack: weekly thread for sneers not worth an entire post, week ending 28th June 2026

    Jump

  • I wouldn’t want to do that myself… personally too much tracking gear in there, and it’s easy to make a mistake and not disable it all. Also, you just know that if you get caught, they’ll try and prosecute it like you kidnapped and dismembered a regular officer.

    Now, I’m more surprised that they don’t get black bagged and tipped over. Maybe they only ever use them in super thoroughly surveilled areas with nearby human backup, but you’d expect at least one successful tipping to make the news somewhere.

  • Stubsack: weekly thread for sneers not worth an entire post, week ending 28th June 2026

    Jump

  • How it started: in 2025, the city of dublin, ohio (the latter detail missed by quite a lot of reporting,because there are no other dublins it might get confused with, I guess) gets an autonomous? ai powered police surveillance robot.

    City officials are encouraging residents to interact with Dubbot—ask questions, take selfies, and experience firsthand how AI is shaping public safety. The goal is to foster transparency and gather feedback to refine the robot’s role in the community.

    How it’s going

    The person-sized, camera-covered robot that looked like it rolled right out of a sci-fi movie did not identify any criminal incidents, issue any tickets or help with any arrests in its nearly 10 months on the job.

    On the other hand, I bet it didn’t shoot anyone’s dog, so who’s to say that the $64k was wasted.

  • “Omelas” seems like a great brand name

    Jump

  • Anyone who names a security company “sauron” has critically failed their reading comprehension. Mercifully, there’s no company naming itself saruman, because that character was breathtakingly dumb in so many ways that perhaps even the y-combinator set are dimly aware of them.

  • “Omelas” seems like a great brand name

    Jump

  • Who even has time for that? Do you think that the people behind palantir, icarus and sauron have time to read google summaries? They’re too busy remaking the world!

    Anyway, if you’re successful enough you’ll eclipse the original source in terms of importance and all the search engine summaries will be about you anyway, so any time spent learning anything before that will have been completely wasted.

  • TechTakes @awful.systems
    rook @awful.systems

    “Omelas” seems like a great brand name

  • Stubsack: weekly thread for sneers not worth an entire post, week ending 21st June 2026

    Jump

  • Some folks, who may be familiar to some or more of you, accidentally discovered that if your git repo symlinks CLAUDE.MD to, say, /dev/urandom, it breaks Claude code.

    the reason why this works is exactly the reason why claude code sucks so bad. there are protections against this in the file reading tool. however because everything in claude code is implemented in 5 million different ways, those protections are a completely orthogonal set of codepaths from how CLAUDE.md files are read. conversely, the file read tool seems to be completely naive to symlinks while the CLAUDE.md reader is not. this is the fucking swiss cheese security model of the fucking gold standard of what AI programming can do.

    https://neuromatch.social/@jonny/116779793188712173

    The thread is actually about trying to attract and manipulate autonomous coding agents, but they’ve only had limited success so far, which may have been slowed down by the above symlink trick.

  • Stubsack: weekly thread for sneers not worth an entire post, week ending 14th June 2026

    Jump

  • I think part of the issue is that historical software quality was an artefact of its time… if you can’t easily patch your released products, you need to work harder to ensure they’re functional. If the only way for people to learn about how your product works in the documentation you ship with it, the docs need to be useful and comprehensive.

    The combination of software needing no guarantee of merchantability or fitness for any particular purpose and the internet rendered those pressures obsolete. Ship shit, fix later. Mass-scale a/b testing over past decade or two shows that most people seemingly don’t care if their software runs like absolute garbage, and is covered in adverts, and harvests all their personal data and the leaks all of it that wasn’t sold.

    An incident-to-pr ratio that’s up by 250% is unfortunate, but it is not yet so bad that the end-users actually care enough to do anything about it, even assuming they can do anything.