Login
You're viewing the functional.cafe public feed.
  • Nov 3, 2026, 1:23 AM

    The most realistic aspect of the Hitchhiker's Guide is Deep Thought taking a silly question, failing to interrogate its premises, and running with it for 7.5 million years until finally regurgitating a nonsense answer that means nothing to anybody at the end after burning through unimaginable resources

    💬 16🔄 510⭐ 785

Replies

  • Nov 4, 2026, 11:12 PM

    I was doing a bit more thinking about this today and I reckon the thing that irks me the most is that the tools already exist to reliability and tractably write correct software for some pretty damn strong definitions of 'correct'. But, the industry has for years dragged its feet and has systematically decided that safety and privacy are worth throwing under the bus in order to prioritise extracting value at pace. In short, there's a lot of laziness, and it's been left to researchers and hobbyists to push the dial on a lot of this stuff.

    What AI seems to offer to industry is a shortcut: skip investing in the time and skills and workers required to get stuff right, and instead crank the magic post-hoc proof machine to get a nice rubber stamp for free. That whatever is spat out isn't nearly a strong or useful enough guarantee for the application is an irrelevance: the company gets to claim that it's been diligent and the responsibility for failure gets shifted into the machine instead.

    💬 3🔄 8⭐ 12
  • Nov 4, 2026, 11:25 PM

    It's very clearly an accountability sink, and it's one that rides on the reputational coattails of ecosystems like Rust & Lean (and others) that were built up slowly by thousands of volunteers labouring at the task of showing that a better world was possible.

    These ecosystems acquired that reputation because an obsessive amount of care was put into marrying up the capabilities of automated provers with the fuzzy, underspecified real world that systems have to actually operate in, ensuring that the two were weaved together into a whole that did not simply generate new outputs, but created entirely new concepts and language with which to talk about correctness.

    By jumping on the bandwagon with an automated rewrite without engaging with the culture that made it possible, companies muddy the trust and respect that surrounds those ecosystems. It is *rude*.

    💬 0🔄 7⭐ 15
  • 💬 1🔄 4⭐ 31
  • 💬 0🔄 0⭐ 6
  • 💬 0🔄 0⭐ 4
  • 💬 0🔄 0⭐ 4
  • Nov 3, 2026, 7:19 PM

    @jsbarretto Douglas Adams was interested in computing from pretty early on! You can even see hints of OOP in Mostly Harmless, grim a book as that is

    💬 0🔄 1⭐ 4
  • 💬 4🔄 5⭐ 21
  • 💬 1🔄 2⭐ 9
  • 💬 0🔄 1⭐ 3
  • 💬 0🔄 0⭐ 2
  • 💬 0🔄 0⭐ 0
  • 💬 1🔄 2⭐ 7
  • 💬 1🔄 0⭐ 0
  • 💬 0🔄 0⭐ 5
  • Nov 4, 2026, 7:25 AM

    @Wyatt_H_Knott @jsbarretto 👍 Arthur telling the Nutri-matic Food dispenser all what he knows about tea and still gets something almost but not quite like tea is a pretty spot on description of vibe coding.

    💬 0🔄 1⭐ 9
  • Nov 3, 2026, 9:02 PM

    @jsbarretto I wonder why Musk didn't name Grok Deep Thought. Can't have been copyright issues, he doesn't care about those.

    💬 2🔄 1⭐ 3
  • 💬 0🔄 0⭐ 5
  • 💬 0🔄 0⭐ 1
  • Nov 4, 2026, 12:17 AM

    @gbsills @jsbarretto Because "grok" is a programmer meme, and Deep Mind isn't.

    Seriously, I don't believe for a second that Musk actually read Stranger in a Strange Land, or Hitchhiker's.

    💬 0🔄 0⭐ 4
  • 💬 1🔄 0⭐ 1
  • 💬 0🔄 0⭐ 1
  • Nov 3, 2026, 9:59 PM

    @jsbarretto the answer was not very usable but quite telling : 42 = * in ascii code, meaning "anything" in computing slang, a "joker" for the poor machine having that burden to answer such a stupid question

    💬 0🔄 0⭐ 5
  • 💬 1🔄 1⭐ 3
  • 💬 1🔄 0⭐ 2
  • Nov 4, 2026, 12:04 AM

    @twasink @steely_glint @jsbarretto sort of

    IIRC In one of the books a cartel of psychologists or philosophers or something were scared that knowing the Ultimate Question would put them out of a job. So they paid off a couple officials and had a backwater planet nobody cared about destroyed to maintain their position.

    💬 1🔄 0⭐ 1
  • Nov 4, 2026, 12:07 AM

    @beeoproblem @steely_glint @jsbarretto Yeah, that’s part of the story in the third book. They arranged the destruction of the Earth to stop finding out the Ultimate Question.

    Of course, the real answer is that they themselves were manipulated into doing that by the Universe. The Universe protects itself, because if both the Answer and the Question are known in the Universe simultaneously, the Universe destroys itself and is recreated as something even less understandable.

    This has happened repeatedly.

    💬 0🔄 0⭐ 4
  • 💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 1
  • 💬 0🔄 0⭐ 2
  • Nov 4, 2026, 12:53 PM

    @jsbarretto I think the 2nd book (but I might misremember) had a scene where Arthur asked the replicator to create a tea. The replicator was unsure how to do that, and consulted the HOG's main computer. Together, they took a few hours to figure out what tea is. This took all the ship's compute resources, making it unable to manoeuvre when they were attacked during that time.

    💬 0🔄 0⭐ 1
  • Nov 4, 2026, 11:29 PM

    @jsbarretto a functioning regulatory state exists to curb the impulses of capital extraction, which dovetail with the goals of government intelligence agencies to illegally surveil their own and other country's citizens. it's quite notable that funding for operating system development in the US completely dried up after the 90s, dovetailing with last week's NIST document finally deprecating diffie-hellman key exchange based upon a cartoonishly flawed model of "quantum" computation constructed over decades to legitimize this political decision.

    this is not mere laziness, but malice. LLMs are a tool to hide backdoors by destroying the vestiges of code review processes that threaten to expose very blatantly broken code, which is subsequently propagated to end user computing devices. it is extremely important for the legitimacy of this apparatus that LLMs be seen as "intelligent", and not "obfuscatory". and this is why we see the bowdlerization of formal theory everywhere, as a means to the same end.

    💬 2🔄 2⭐ 2
  • Nov 4, 2026, 11:29 PM

    @jsbarretto it is a tremendously radical thing to provide a strong guarantee of behavior. it's important to understand how threatening that is to systems of centralized authoritarian control

    💬 1🔄 0⭐ 1
  • Nov 4, 2026, 11:31 PM

    @jsbarretto i would urge you to consider this as well. it is not intended to make you feel better--it makes me feel much worse. however, it is more explanatory of behavior than laziness and greed alone (arguably, authoritarian coercion is prone to laziness and greed, but indirectly so)

    💬 0🔄 0⭐ 1
  • Nov 4, 2026, 11:41 PM

    @hipsterelectron To be honest, I'm skeptical that there's much of a grand plan here. I think it's quite easily explained by orgs making locally reasonable (if self-interested) moves that culminate in a poisoning of the well, because all of this is happening in the context of technology that's increasingly good at producing externalities. The physical, economic, and information processing barriers that used to demarcate the economy and society and - to some extent - force orgs to account for the consequences of their decisions are withering away. The things that used to require sustained effort, and hence acted as a disincentive for exploitation, are sliding away. It should not be surprising: technology is a force multiplier almost by definition.

    💬 1🔄 0⭐ 0
  • Nov 5, 2026, 12:23 AM

    @jsbarretto "grand plan" is missing the point. backdoors are not just done for the purpose of direct wartime attacks, but to serve the purpose of normalizing incorrect and broken semantics so that correctness and robustness seems rare and far-between—and this opens the path for particular imaginings of "correctness" and formal semantics which serve the purposes of coercion rather than liberation to be platformed, funded, and marketed. this has now happened to me twice, first from bazel and then from uv. if your work hasn't been treated this way yet, try building something a little more radical for once. i formalized the behavior of python packaging

    💬 1🔄 0⭐ 1
  • Nov 5, 2026, 12:25 AM

    @jsbarretto if you have not seen a commit which is unmistakeably broken in a way which cannot be explained by laziness then i envy your ignorance. your model of corporate behavior will remain incomplete until you can incorporate actual malice

    💬 0🔄 0⭐ 0
  • Rustybridgeenjoyer@gamepad.club
    Nov 6, 2026, 11:32 PM

    @jsbarretto you put this very well. Ive been saying how backwards ai use is. People are using it as a computer program. That is stupid. We have computer programs for a reason that are deterministic and not beholden to the whims of a Nazi (not to mention constant internet. What happened to offline first? I hate constant internet requirement). Its just annoying how many many people are using it (while destroying earth) for shit we could ALREADY DO. much of which we could do decades ago, like OCR.

    💬 0🔄 0⭐ 0
  • 💬 1🔄 0⭐ 0
  • 💬 1🔄 0⭐ 0
  • 💬 1🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0
  • Nov 7, 2026, 8:06 PM

    @jsbarretto
    We seem to be seeing that aspect of human nature play out in real time now: a failure to understand that its the process of asking the questions and learning how to do things the hard way is what brings realistic understanding. Taking a shortcut to an answer robs that answer of any context, value or wisdom.

    💬 0🔄 0⭐ 0