Skip Navigation

InitialsDiceBearhttps://github.com/dicebear/dicebearhttps://creativecommons.org/publicdomain/zero/1.0/„Initials” (https://github.com/dicebear/dicebear) by „DiceBear”, licensed under „CC0 1.0” (https://creativecommons.org/publicdomain/zero/1.0/)C

corbin

@ corbin @awful.systems

Posts
5
Comments
77
Joined
3 yr. ago

  • Stubsack: weekly thread for sneers not worth an entire post, week ending 21st December 2025

    Jump

  • Sadly, it's a Chomskian paper, and those are just too weak for today. Also, I think it's sloppy and too Eurocentric. Here are some of the biggest gaffes or stretches I found by skimming Moro's $30 book, which I obtained by asking a shadow library for "impossible languages" (ISBN doesn't work for some reason):

    • Moro claims that it's impossible for a natlang to have free word order. There's many counterexamples which could be argued, like Arabic or Mandarin, but I think that the best counterexample is Latin, which has Latinate (free) word order. On one hand, of course word order matters for parsers, but on the other hand the Transformers architecture attends without ordering, so this isn't really an issue for machines. Ironically, on p73-74, Moro rearranges the word order of a Latin phrase while translating it, suggesting either a use of machine translation or an implicit acceptance of Latin (lack of) word order. I could be harsher here; it seems like Moro draws mostly from modern Romance and Germanic languages to make their points about word order, and the sensitivity of English and Italian to word order doesn't imply a universality.
    • Speaking of universality, both the generative-grammar and universal-grammar hypotheses are assumed. By "impossible" Moro means a non-recursive language with a non-context-free grammar, or perhaps a language failing to satisfy some nebulous geometric requirements.
    • Moro claims that sentences without truth values are lacking semantics. Gödel and Tarski are completely unmentioned; Moro ignores any sort of computability of truth values.
    • Russell's paradox is indirectly mentioned and incorrectly analyzed; Moro claims that Russell fixed Frege's system by redefining the copula, but Russell and others actually refined the notion of building sets.
    • It is claimed that Broca's area uniquely lights up for recursive patterns but not patterns which depend on linear word order (e.g. a rule that a sentence is negated iff the fourth word is "no"), so that Broca's area can't do context-sensitive processing. But humans clearly do XOR when counting nested negations in many languages and can internalize that XOR so that they can handle utterances consisting of many repetitions of e.g. "not not".
    • Moro mentions Esperanto and Volapük as auxlangs in their chapter on conlangs. They completely fail to recognize the past century of applied research: Interlingue and Interlingua, Loglan and Lojban, Láadan, etc.
    • Sanskrit is Indo-European. Also, that's not how junk DNA works; it genuinely isn't coding or active. Also also, that's not how Turing patterns work; they are genuine cellular automata and it's not merely an analogy.

    I think that Moro's strongest point, on which they spend an entire chapter reviewing fairly solid neuroscience, is that natural language is spoken and heard, such that a proper language model must be simultaneously acoustic and textual. But because they don't address computability theory at all, they completely fail to address the modern critique that machines can learn any learnable system, including grammars; they worst that they can say is that it's literally not a human.

  • Stubsack: weekly thread for sneers not worth an entire post, week ending 14th December 2025 - awful.systems

    Jump

  • I got jumpscared by Gavin D. Howard today; apparently his version of bc appeared on my system somehow, and his name's in the copyright notice. Who is Gavin anyway? Well, he used to have a blog post that straight-up admitted his fascism, but I can't find it. I could only find, say, the following five articles, presented chronologically:

    Also, while he's apparently not caused issues for NixOS maintainers yet, he's written An Apology to the Gentoo Authors for not following their rules when it comes to that same bc package. So this might be worth removing for other reasons than the Christofascist authorship.

    BTW his code shows up because it's in upstream BusyBox and I have a BusyBox on my system for emergency purposes. I suppose it's time to look at whether there is a better BusyBox out there. Also, it looks like Denys Vlasenko has made over one hundred edits to this code to integrate it with BusyBox, fix correctness and safety bugs, and improve performance; Gavin only made the initial commit.

  • Stubsack: weekly thread for sneers not worth an entire post, week ending 14th December 2025 - awful.systems

    Jump

  • They (or the LLM that summarized their findings and may have hallucinated part of the post) say:

    It is a fascinating example of "Glue Code" engineering, but it debunks the idea that the LLM is natively "understanding" or manipulating files. It's just pushing buttons on a very complex, very human-made machine.

    Literally nothing that they show here is bad software engineering. It sounds like they expected that the LLM's internals would be 100% token-driven inference-oriented programming, or perhaps a mix of that and vibe code, and they are disappointed that it's merely a standard Silicon Valley cloudy product.

    My analysis is that Bobby and Vicky should get raises; they aren't paid enough for this bullshit.

    By the way, the post probably isn't faked. Google-internal go/ URLs do leak out sometimes, usually in comments. Searching GitHub for that specific URL turns up one hit in a repository which claims to hold a partial dump of the OpenAI agents. Here is combined_apply_patch_cli.py. The agent includes a copy of ImageMagick; truly, ImageMagick is our ecosystem's cockroach.

  • Why Disney just put $1 billion into OpenAI

    Jump

  • Now I'm curious about whether Disney funded Glaze & Nightshade. Quoting Nightshade's FAQ, their lab has arranged to receive donations which are washed through the University of Chicago:

    If you or your organization may be interested in pitching in to support and advance our work, you can donate directly to Glaze via the Physical Sciences Division webpage, click on "Make a gift to PSD" and choose "GLAZE" as your area of support (managed by the University of Chicago Physical Sciences Division).

    Previously, on Awful, I noted the issues with Nightshade and the curious fact that Disney is the only example stakeholder named in the original Nightshade paper, as well as the fact that Nightshade's authors wonder about the possibility of applying Glaze-style techniques to feature-length films.

  • Stubsack: weekly thread for sneers not worth an entire post, week ending 14th December 2025 - awful.systems

    Jump

  • The orange-site whippersnappers don't realize how old artificial neurons are. In terms of theory, the Hebbian principle was documented in 1949 and the perceptron was proposed in 1943 in an article with the delightfully-dated name, "A logical calculus of the ideas immanent in nervous activity". In 1957, the Mark I Perceptron was introduced; in modern parlance, it was a configurable image classifier with a single layer of hundreds-to-thousands of neurons and a square grid of dozens-to-hundreds of pixels. For comparison, MIT's AI lab was founded in 1970. RMS would have read about artificial neurons as part of their classwork and research, although it wasn't part of MIT's AI programme.

  • Stubsack: weekly thread for sneers not worth an entire post, week ending 30th November 2025 - awful.systems

    Jump

  • Stubsack: weekly thread for sneers not worth an entire post, week ending 30th November 2025 - awful.systems

    Jump

  • Ziz was arraigned on Monday, according to The Baltimore Banner. She apparently was not very cooperative:

    As the judge asked basic questions such as whether she had read the indictment and understood the maximum possible penalties, [Ziz] LaSota chided the “mock proceedings” and said [US Magistrate Douglas R.] Miller was a “participant in an organized crime ring” led by the “states united in slavery.”

    She pulled the Old Man from Scene 24 gag:

    Please state your name for the record, the court clerk said. “Justice,” she replied. What is your age? “Timeless.” What year were you born? “I have been born many times.”

    The lawyers have accepted that sometimes a defendant is uncooperative:

    Prosecutors said the federal case would take about three days to try. Defense attorney Gary Proctor, in an apparent nod to how long what should have been a perfunctory appearance on Monday ended up taking, called the estimate “overly optimistic.”

    Folks outside the USA should be reassured that this isn't the first time that we've tried somebody with a loose grasp of reality and a found family of young violent women who constantly disrupt the trial; Ziz isn't likely to walk away.

  • Stubsack: weekly thread for sneers not worth an entire post, week ending 23rd November 2025 - awful.systems

    Jump

  • Indeed. I left a note on one of his blogposts correcting a common misconception (that it's "all just tokens" and the model can't tell when you clearly substituted an unlikely word, common among RAG-heavy users) and he showed up to clarify that he merely wanted to "start an interesting conversation" about how to improve his particular chatbots.

    It's almost like there's a sequence: passing the Turing test, sycophancy, ELIZA effect, suggestibility, cognitive offloading, shared delusions, psychoses, conspiracy theories, authoritarian-follower personality traits, alt-right beliefs, right-wing beliefs. A mechanical Iago.

  • Vibe nuclear — let’s use AI shortcuts on reactor safety!

    Jump

  • Linear no-threshold isn't under attack, but under review. The game-theoretic conclusions haven't changed: limit overall exposure, radiation is harmful, more radiation means more harm. The practical consequences of tweaking the model concern e.g. evacuation zones in case of emergency; excess deaths from radiation exposure are balanced against deaths caused by evacuation, so the choice of model determines the exact shape of evacuation zones. (I suspect that you know this but it's worth clarifying for folks who aren't doing literature reviews.)

  • Data Center Watch worries that anti-AI activism is working

    Jump

  • Unlike a bunker, a datacenter's ventilation consists of [DATA EXPUNGED] which are out of reach. The [DATA EXPUNGED] are heavily [DATA EXPUNGED], so [DATA EXPUNGED] unlikely to work either. However, this ventilation must be [DATA EXPUNGED] in order to effectively [DATA EXPUNGED], and that's done by [DATA EXPUNGED] into the [DATA EXPUNGED] and [DATA EXPUNGED] to prevent [DATA EXPUNGED].

    Edit: making the joke funnier.

  • Data Center Watch worries that anti-AI activism is working

    Jump

  • In my personal and professional opinion, most datacenter outages are caused by animals disturbing fiber or power lines. Consider campaigning for rewilding instead; it's legal and statistically might be more effective.

  • Stubsack: weekly thread for sneers not worth an entire post, week ending 16th November 2025

    Jump

  • Previously, on Awful, I wrote up what I understand to be their core belief structure. It's too bad that we're not calling them the Cyclone Emoji cult.

  • Stubsack: weekly thread for sneers not worth an entire post, week ending 26th October 2025

    Jump

  • Stubsack: weekly thread for sneers not worth an entire post, week ending 26th October 2025

    Jump

  • Hey now, at least the bowl of salvia has a theme, predictable effects, immersive sensations, and the ability to make people feel emotions.

  • Stubsack: weekly thread for sneers not worth an entire post, week ending 26th October 2025

    Jump

  • Thanks! You're getting better with your insults; that's a big step up from your trite classics like "sweet summer child". As long as you're here and not reading, let's not read from my third link:

    As a former musician, I know that there is no way to train a modern musician, or any other modern artist, without heavy amounts of copyright infringement. Copying pages at the library, copying CDs for practice, taking photos of sculptures and paintings, examining architectural blueprints of real buildings. The system simultaneously expects us to be well-cultured, and to not own our culture. I suggest that, of those two, the former is important and the latter is yet another attempt to coerce and control people via subversion of the public domain.

    Maybe you're a little busy with your Biblical work-or-starve mindset, but I encourage you to think about why we even have copyright if it must be flaunted in order to become a skilled artist. It's worth knowing that musicians don't expect to make a living from our craft; we expect to work a day job too.

  • Stubsack: weekly thread for sneers not worth an entire post, week ending 26th October 2025

    Jump

  • Previously, on Awful:

    [Copyright i]s not for you who love to make art and prize it for its cultural impact and expressive power, but for folks who want to trade art for money.

    Quoting Anarchism Triumphant, an extended sneer against copyright:

    I wanted to point out something else: that our world consists increasingly of nothing but large numbers (also known as bitstreams), and that - for reasons having nothing to do with emergent properties of the numbers themselves - the legal system is presently committed to treating similar numbers radically differently. No one can tell, simply by looking at a number that is 100 million digits long, whether that number is subject to patent, copyright, or trade secret protection, or indeed whether it is "owned" by anyone at all. So the legal system we have - blessed as we are by its consequences if we are copyright teachers, Congressmen, Gucci-gulchers or Big Rupert himself - is compelled to treat indistinguishable things in unlike ways.

    Or more politely, previously, on Lobsters:

    Another big problem is that it's not at all clear whether information, in the information-theoretic sense, is a medium through which expressive works can be created; that is, it's not clear whether bits qualify for copyright. Certainly, all around the world, legal systems have assumed that bits are a medium. But perhaps bits have no color. Perhaps homomorphic encryption implies that color is unmeasurable. It is well-accepted even to legal scholars that abstract systems and mathematics aren't patentable, although the application of this to computers clearly shows that the legal folks involved don't understand information theory well enough.

    Were we anti-copyright leftists really so invisible before, or have you been assuming that No True Leftist would be anti-copyright?

  • Stubsack: weekly thread for sneers not worth an entire post, week ending 26th October 2025

    Jump

  • Closely related is a thought I had after responding to yet another paper that says hallucinations can be fixed:

    I'm starting to suspect that mathematics is not an emergent skill of language models. Formally, given a fixed set of hard mathematical questions, it doesn't appear that increasing training data necessarily improves the model's ability to generate valid proofs answering those questions. There could be a sharp divide between memetically-trained models which only know cultural concepts and models like Gödel machines or genetic evolution which easily generate proofs but have no cultural awareness whatsoever.

  • Stubsack: weekly thread for sneers not worth an entire post, week ending 26th October 2025

    Jump

  • "Not Winston Smith?" So, O'Brien?

  • AI boyfriend users radicalise against OpenAI — and self-host their chatbot companions

    Jump

  • Boring unoriginal argument combined with a misunderstanding of addiction. On addiction, go read FOSB and stop thinking of it as a moral failing. On behavioral control, it's clear that you didn't actually read what I said. Let me emphasize it again:

    The problem isn’t people enjoying their fetishes; the problem is the financial incentives and resulting capitalization of humans leading to genuine harms.

    From your list, video games, TV, D&D, and group sex are not the problem. Rather, loot boxes, TV advertisements, churches, MLMs, and other means of psychological control are the problem. Your inability to tell the difference between a Tupperware party (somewhat harmful), D&D (almost never harmful), and joining churches (almost always harmful) suggests that you're thinking of behavioral control in terms of rugged individualist denial of any sort of community and sense of belonging, rather than in terms of the harms which people suffer. Oh, also, when you say:

    One cannot rescue such people by condemning what they do, much like one cannot stop self destruction by banning the things they use.

    Completely fucking wrong. Condemning drunk driving has reduced the overall amount of drunk driving, and it also works on an interpersonal level. Chemists have self-regulated to prevent the sale of massive quantities of many common chemicals, including regulation on the basis that anybody purchasing that much of a substance could not do anything non-self-destructive with it. What you mean to say is that polite words do not stop somebody from consuming an addictive substance, but it happens to be the case that words are only the beginning of possible intervention.