• How UK lost the 246 Theorem Race [by Engaging a Clown] (Re: Less Ego more Math [Somebody got Excited])

    From Mild Shock@janburse@fastmail.fm to comp.theory,comp.lang.c,comp.lang.java on Sun Nov 15 12:41:01 2026
    From Newsgroup: comp.theory

    Hi,

    How it started, a Clown repeating the late Edsger Dijkstra:

    The question of whether a computer can think is no
    more interesting than the question of whether
    a submarine can swim.
    -- Edsger Dijkstra

    "You want it to be brilliant, but itrCOs not capable of being
    brilliant because itrCOs not capable of having any insight
    at all. ItrCOs just trying to follow patterns." https://www.hobartpulp.com/web_features/submarines-don-t-swim-a-conversation-with-mathematician-kevin-buzzard

    How it is going, a Clown admitting his defeat:

    Axiom Prover (Axiom Math) 40/50 80.0%
    Annals of Mathematics - NeurIPS 2026
    https://arxiv.org/pdf/2610.18301

    AxiomProver is an autonomous, multi-agent AI theorem
    prover developed by Axiom Math (founded and led
    by CEO Carina Hong). Benchmarked on the Annals

    of Mathematics, not the 246 Theorem, but leaves behind
    a few other systems currently, like Humanifa, NEAR AI,
    Tau and Midas.

    Bye

    P.S.: About the 246 Theorem
    "This is not AxiomProverrCOs first rodeo. Axiom Math has used
    its autonomous, multi-agent system that turns mathematical
    statements into machine-checkable proofs to crack several

    unsolved mathematical problems and verify many
    more proofs this year." But proof formalization of the 246
    theorem is by far the most significant, as Ken Ono, Axiom

    MathrCOs founding mathematician, explains: rCLThis theorem
    currently represents the threshold of human knowledge
    about prime numbers.rCY https://spectrum.ieee.org/axiom-math-246-theorem-formalization

    Mild Shock schrieb:
    Hi,

    How its started:

    More Math, Less Ego (talk about AI and Math)
    Erik Demaine - 29.10.2026
    https://www.youtube.com/watch?v=Gdc1yll04qI

    How its going:

    Classification of Finite Simple Groups
    Samuel Alexander - 28.10.2026
    https://www.youtube.com/watch?v=NZklSBoLC30

    Bye

    Mild Shock schrieb:
    Hi,

    The post is only proof of the Terence Tao
    ingestion problem hypothesis. That we are not
    anymore able to process content, and that
    even trust is destroyed.

    The AI slop presented here as a rCLsaferCY prolog
    agent toolkit, halucinates things like. And it
    does so consistently across many pages, still it
    wants to provide rCLsafetyrCY:

    ### Bakage (`bakage.toml`)
    The package manifest format used by Scryer Prolog to
    declare project metadata, exported modules, and dependencies.
    https://github.com/dougransom/prolog-agent-toolkit

    While I donrCOt find any evidence for such a
    manifest file. Its rather:

    A package is a directory with a scryer-manifest.pl
    following this schema: etc..
    https://github.com/bakaq/bakage

    That the post has even likes from field experts
    like Moura from Logtalk and Torbjorn from Web
    Prolog, shows that a) you can post any nonsense
    on GitHub people

    will clap, and b) this forum is useless only
    spreading AI slop. So while everybody is talking
    about provenance and justification, as a desiderata,
    the prolog agent toolkit is

    just an example of unhinged free wheeling analogical
    reasoning, probably taking phrases such as cargo.toml
    as a blue print and projecting them into phrases
    such aa bakage.toml.

    I still think a tool breaks when its weakest
    chain link break, even if people take this as a
    joke of mine. But well you are all free to give it
    a try, like running AI homunculi as a new sport.

    Bye

    (*)

    Terence Tao
    I have recently made the point that out of three
    major components of the mathematical problem
    solving process - proof generation, proof
    verification, and proof digestion
    https://mathstodon.xyz/@tao/116551624228986501

    P.S.: Such repos are useless if there is no way
    to cockpit the original agentic AI used. You would
    need to go through the route of tickets and GitHub
    issues to fix problems.

    Basically subscribing to the open source dream
    of Linux, that many eyes can flatten out bugs. But
    this is contra-productive to use strong tools and
    then wait for weak humans!

    Mild Shock schrieb:
    Hi,

    The idea that mathematics is based on a universal bowel
    movement system, that needs to learn and use "right and
    natural" math, is already disproved by our Indian vegetarian

    visitors here, that can even not digest Swiss Cheese. So
    lets declare repeating a canon a dead skill in the age of
    LLMs, and dig deeper what intelligence could mean.

    There is a thesis and ongoing product development:

    On the Measure of Intelligence Fran|oois Cholle author
    of the notable book "Deep Learning with Python"
    https://arxiv.org/pdf/1911.01547

    Ndea is building frontier AI systems that blend intuitive pattern
    recognition and formal reasoning into a unified architecture.
    https://ndea.com/about

    What about putting "right and natural" aside. And train
    mathematical intuition? Maybe the pressure of the moon
    race and cold war is lacking? And P(doom) is even

    the wrong wakeup call, but a genius pressure cooker!

    LoL

    Bye

    Mild Shock schrieb:
    Hi,

    Even Philip Wadler sounds like he has seen the
    first Apocalyptic horse rider. But what do these
    hardcore formalists think, that we go with a

    2300-year old Aristotelian inferencing like Modus
    Barbara (*) into the next decade of 2030?

    All humans are mortal
    All Greeks are humans
    --------------------------------
    Therefore all Greeks are mortal

    Or as the formalists like to write it, but this
    is already a "fucked up" chain rule:

    reCx (M(x) raA S(x))
    reCx (G(x) raA M(x))
    roCroCroCroCroCroCroCroCroCroCroCroCroCroCroCroCroC
    re| reCx (G(x) raA S(x))

    Its just various camps fighting world domination:

    Everybody Wants To Rule The World
    https://www.youtube.com/watch?v=uqItaKDc6oI

    And the winner in the P(doom) bottom race is:

    CEO Jensen Huang: Zero Probability P(doom) in 2030
    https://www.youtube.com/watch?v=XlY7DI5v4ZU

    Bye

    (*)
    History of formal logic - III. Logic of Terms
    Bochenski, Joseph M. - 1961
    https://archive.org/details/historyofformall00boch/page/210/mode/2up

    Mild Shock schrieb:
    Hi,

    Woa! The public now marveling at the cascade
    LLM, LRM and now Swarms. And little software
    agents posting PERMADEATH.

    How it started:

    The co-evolution of AI and systems with Lidong Zhou
    https://www.youtube.com/watch?v=FVuJuGwkSu0

    How its going:

    This Is the AI Agent Story That Scared Me
    https://www.youtube.com/watch?v=2aw3MF8pY3w

    They are all clueless that Strong AI, requires
    to go from in vitro , to in vivo, usually called
    situated AI. Applies also to swarms:

    "Being situated is generally considered to be part
    of being embodied, but it is useful to consider
    each perspective individually. The situated
    perspective emphasizes that intelligent behavior
    derives from the environment and the agent's
    interactions with it. The nature of these
    interactions are defined by an agent's embodiment."
    https://en.wikipedia.org/wiki/Situated_approach_(artificial_intelligence)


    And emergence of private languages is nothing new.
    I am pretty sure the interested reader will find
    some 30 year old papers of in vitro simulations,

    of software agents developing pebbel languages.

    Have Fun!

    Bye

    Mild Shock schrieb:
    Hi,

    How it started:

    Randy and the Lawnmower
    https://www.youtube.com/watch?v=B17Yy5buCdM

    How it is going:

    Keep off the grass folks
    https://9gag.com/gag/aW4VYOd

    Bye

    Mild Shock schrieb:
    Hi,

    It pretty much emerges that python will
    suffer from the small hands Paradox.
    We are curently in this phase of the python
    programming language evolution:

    Maastrichtian
    The CretaceousrCoPaleogene extinction event
    (formerly known as the CretaceousrCoTertiary
    extinction event)[a] occurred at the end of
    this age.[3] In this mass extinction, many
    commonly recognized groups such as non-avian
    dinosaurs, pterosaurs, plesiosaurs, and mosasaurs,
    as well as many other lesser-known groups,
    died out. The cause of the extinction is most
    commonly linked to an asteroid about 10 to
    15 kilometres (6.2 to 9.3 mi) wide[4][5]
    colliding with Earth, ending the Cretaceous.

    It already happened to most of the Python tooling:

    [Write Code] rRo
    -a-a-a-a [Lint with Ruff] rRo
    -a-a-a-a (Rust: Blazing)

    -a-a-a-a [Install with uv] rRo
    -a-a-a-a (Rust: Blazing)

    -a-a-a-a [Execute Program]
    -a-a-a-a (Python: Extremely Slow)

    LoL

    Bye

    P.S.: How about Mojo as a successor of Python?

    Mojo is a programming language for Linux and
    macOS.[10][11] It is a systems programming language
    with semantics inspired by Rust such as static
    typing and a borrow checker,[12][13] but uses a
    syntax designed to be reminiscent of Python
    (like Nim and Julia).
    https://en.wikipedia.org/wiki/Mojo_(programming_language)

    Mild Shock schrieb:
    Hi,

    Just downloading some other person's code

    I didn't do that, I wrote Hack VM for pi-WAM
    from scratch, over the last 4 weeks. I came
    back from holidays on end of June 2026, and now

    we have end of July 2026. But its only possible
    because the instruction set is very smal, like
    ca. 8 functions and ca. 8 modes and ca. 8 conditions,

    so its ca. 8 x 8 x 8 = 512 opcodes, each has an
    A parameter and a D parameter simultaneously.
    It has currently the following CPU backends:

    -a-a- Nnow supports interleaved synchronous emulation.
    -a-a- Now supports warp parallelism via Java platform threads. >>>>>>>> -a-a- Now supports warp parallelism via Python system threads. >>>>>>>> -a-a- Now supports warp parallelism via JavaScript worker threads. >>>>>>>> -a-a- Note: For Python free threads are not yet fully tested.
    -a-a- Note: For JavaScript web workers are not yet fully tested. >>>>>>>>
    https://www.dogelog.ch/typtab/doclet/book/14_install/05_notes22/110_224.html


    But frankly I came to encounter Hack not from
    the usual university curriculum web resources,
    but indirectly through a post about a Prolog

    emulation of Hack, using constrained horn clauses (CHC):

    Verifying Nand2Tetris Assembly
    https://www.philipzucker.com/nand2tetris-chc/

    The binary encoding is currently that the functions,
    modes and conditions eat up a nibble (4-bit), in
    total 12-bit, which I use then 10-bit for A parameter

    and 10-bit for D parameter. I used AI freemium, Codex
    by ChatGPT from within IntelliJ to do some fragment
    code translations automatically from Java to JavaScript

    or from JavaScript to Python.

    Have Fun!

    Bye

    Johann 'Myrkraverk' Oskarsson schrieb:
    As I said elsewhere, I'm not that interested in Hack, but virtual >>>>>>>>> machines and related code.-a I mentioned another book, but now I >>>>>>>>> feel
    I should ask if you've also read /Crafting Interpreters/ by Robert >>>>>>>>> Nystrom?-a I myself am still on chapter 5, so I'm still in the Java >>>>>>>>> part of the book, so I replaced comp.lang.c++ with
    comp.lang.java, and
    am fairly excited to have a working interpreter.

    This is the type of book people only properly learn from by >>>>>>>>> typing in
    the code themselves.-a Just downloading some other person's >>>>>>>>> code, and
    running an interpreter isn't exactly a learning experience.-a So >>>>>>>>> I'm
    taking my time with it, and interleaving it with my Mars
    adventures.


    In any case, I do wish you to have fun with your own WebGPU >>>>>>>>> project,
    whatever that is about!










    --- Synchronet 3.22a-Linux NewsLink 1.2