ru ding ware
@nvyx.bsky.social
Something I took from the signs at the NoKings protests was that resist libs are totally obsessed with the size of Trump’s penis, which is gross and bizarre, and also strikes me as sort of the opposite conclusion: if anything, Trump seems like someone with too much T, not too little
December 16, 2025 at 10:45 PM
Something I took from the signs at the NoKings protests was that resist libs are totally obsessed with the size of Trump’s penis, which is gross and bizarre, and also strikes me as sort of the opposite conclusion: if anything, Trump seems like someone with too much T, not too little
the serif vs sans-serif split, in cats
December 16, 2025 at 2:57 PM
the serif vs sans-serif split, in cats
I also loved NieR: Automata (2017)
December 13, 2025 at 8:25 AM
I also loved NieR: Automata (2017)
Can’t express how happy I am that the new show is independent and I can just pay you for it.
December 13, 2025 at 1:10 AM
Can’t express how happy I am that the new show is independent and I can just pay you for it.
We already brought back whaling (unregulated crypto- and sports- betting)
December 10, 2025 at 10:24 PM
We already brought back whaling (unregulated crypto- and sports- betting)
You had me in the first half
December 10, 2025 at 10:23 PM
You had me in the first half
Also, tragically, there are lots of people actually using LLMs to do novel mathematical research and they *are* getting results. Can they get "all of mathematics" this way? No, and the question is nonsensical. Can they replace mathematicians? I don't think so. Are they all over the lean zulip? yes.
December 10, 2025 at 4:33 AM
Also, tragically, there are lots of people actually using LLMs to do novel mathematical research and they *are* getting results. Can they get "all of mathematics" this way? No, and the question is nonsensical. Can they replace mathematicians? I don't think so. Are they all over the lean zulip? yes.
...that is, we cannot /program/ them, or define their behavior unambiguously in some well-defined language with a well-defined and unambiguous semantics, like we can for all other "good" software systems)
December 10, 2025 at 4:30 AM
...that is, we cannot /program/ them, or define their behavior unambiguously in some well-defined language with a well-defined and unambiguous semantics, like we can for all other "good" software systems)
(and, earlier, when I said "LLMs are not capable of...", what I meant fully was "we are not capable of formally directing an LLM to either respect these classifications in its input and behave accordingly, or to produce these classifications correctly and consistently".
December 10, 2025 at 4:30 AM
(and, earlier, when I said "LLMs are not capable of...", what I meant fully was "we are not capable of formally directing an LLM to either respect these classifications in its input and behave accordingly, or to produce these classifications correctly and consistently".
Overall, it gives me the same ick as like the string theory or "the universe is a simulation because of planck metrics" guys.
December 10, 2025 at 4:26 AM
Overall, it gives me the same ick as like the string theory or "the universe is a simulation because of planck metrics" guys.
There is a suggestion that LLMs represent some sort of philosophical project, which to me is reminiscent of the sort of "euphoric rationalism" of people like Leibniz, but... you also wouldn't attack that like this. Actually, referencing the incompleteness theorems in this way is a pet peeve of mine.
December 10, 2025 at 4:25 AM
There is a suggestion that LLMs represent some sort of philosophical project, which to me is reminiscent of the sort of "euphoric rationalism" of people like Leibniz, but... you also wouldn't attack that like this. Actually, referencing the incompleteness theorems in this way is a pet peeve of mine.
The very real problems we see with LLMs today have nothing to do with Gödel's incompleteness theorems or the Church-Turing hypothesis. They are the consequence of specific technical limitations, not of bounds on the expressiveness of the underlying logic or computational substrate.
December 10, 2025 at 4:25 AM
The very real problems we see with LLMs today have nothing to do with Gödel's incompleteness theorems or the Church-Turing hypothesis. They are the consequence of specific technical limitations, not of bounds on the expressiveness of the underlying logic or computational substrate.
This merits a longer response, which I'll maybe put together sometime, but I think the original post is confused enough that there's really nothing productive you can get out of it. It's pure vibes, and I jive with the vibes, but that's... not what Turing and Godel and Church et al.'s work says.
December 10, 2025 at 4:21 AM
This merits a longer response, which I'll maybe put together sometime, but I think the original post is confused enough that there's really nothing productive you can get out of it. It's pure vibes, and I jive with the vibes, but that's... not what Turing and Godel and Church et al.'s work says.
you cannot make a formal distinction between good and bad outputs from LLMs in the same way as separation of code and data lets you fix every other variety of "injection" attacks
December 10, 2025 at 12:40 AM
The better argument is that LLMs are not capable of making distinctions of their input and output, like “correct or incorrect” or “configuration data or executable code”. The ability to consistently make these distinctions is the defining feature of modern software engineering, and why LLMs suck.
December 10, 2025 at 12:39 AM
The better argument is that LLMs are not capable of making distinctions of their input and output, like “correct or incorrect” or “configuration data or executable code”. The ability to consistently make these distinctions is the defining feature of modern software engineering, and why LLMs suck.
He didn’t even say that, because the incompleteness theorems apply to a particular logic. There are useful logics where it doesn’t apply.
The real point is that there is no absolute, canonical, or universal sense of “mathematical truth” or mode of “mathematical reasoning.“
The real point is that there is no absolute, canonical, or universal sense of “mathematical truth” or mode of “mathematical reasoning.“
December 10, 2025 at 12:36 AM
He didn’t even say that, because the incompleteness theorems apply to a particular logic. There are useful logics where it doesn’t apply.
The real point is that there is no absolute, canonical, or universal sense of “mathematical truth” or mode of “mathematical reasoning.“
The real point is that there is no absolute, canonical, or universal sense of “mathematical truth” or mode of “mathematical reasoning.“
I‘m doing a PhD in mathematical logic and use proof assistants, and while I agree with the sentiment of the post, and while I’m a total “today’s GenAI has nothing to offer” hater, this post is just wrong.
There’s a pop-sci type understanding of church, turing, and godel’s results which is wrong.
There’s a pop-sci type understanding of church, turing, and godel’s results which is wrong.
December 10, 2025 at 12:33 AM
I‘m doing a PhD in mathematical logic and use proof assistants, and while I agree with the sentiment of the post, and while I’m a total “today’s GenAI has nothing to offer” hater, this post is just wrong.
There’s a pop-sci type understanding of church, turing, and godel’s results which is wrong.
There’s a pop-sci type understanding of church, turing, and godel’s results which is wrong.
That makes sense. My mom actually only uses tiny Japanese toothbrushes for this same reason
December 8, 2025 at 6:21 PM
That makes sense. My mom actually only uses tiny Japanese toothbrushes for this same reason
I don’t understand how these products exist given that all the “old-school” electric toothbrushes are all still available (phillips sonicare, braun oral-b, panasonic, etc.) AND they all have generic, inexpensive replacement heads. Are people really that won over by millennial chic marketing..?
December 8, 2025 at 9:27 AM
I don’t understand how these products exist given that all the “old-school” electric toothbrushes are all still available (phillips sonicare, braun oral-b, panasonic, etc.) AND they all have generic, inexpensive replacement heads. Are people really that won over by millennial chic marketing..?
did you see that old reddit post where some guy thought the cup on the front part of most urinals was where you rested your balls
December 4, 2025 at 7:40 AM
did you see that old reddit post where some guy thought the cup on the front part of most urinals was where you rested your balls
too-restive and malleable
December 2, 2025 at 10:55 PM
too-restive and malleable
we’re making it out of the Middle Kingdom with this one boys
December 2, 2025 at 10:53 PM
we’re making it out of the Middle Kingdom with this one boys