σ Orionis retweeted
This is Nobel Laureate Geoffrey Hinton showing precisely where the entire AI industry went wrong.
He went along with the trend: “If you have any theory of how the brain works, you can simulate it on a digital computer” unless it’s quantum effects.
This is false.
The brain is a physical learning system that uses the laws of physics to calculate. It *cannot* be reproduced in digital form, nor can it be simulated. The moment you digitize information, you lose the part that makes it real. By converting it to bits—no matter how many—you lose the infinite resolution of reality.
All forms of biological intelligence are 100% ANALOG. This is not an incidental feature of intelligence. It is fundamental to it. We have never simulated intelligence or consciousness on a digital computer because you can’t.
LLMs simulate the OUTPUT of intelligence. They do not simulate intelligence itself. They do not learn. They are static piles of human data compressed in high-dimensional math. The whole idea they are intelligent is absurd.
You will never get an intelligent digital computer, much less a conscious one. The future of AI is analog computing. When computers use the universe to calculate, when they adapt based on input from the physical world, we’ll be getting closer.
Until then, AI is a dead end.
σ Orionis retweeted
Replying to @molecularmusing
AI has this leveling effect, and many are excited because they never had the required time or skills to make stuff, but now are fully empowered and feeling the magic of creation for the first time. That is good.
But on the other hand, I think that building something new, still requires imagination, taste, stamina (all three trainable), and either lots of luck or being plugged in the right circles. That's a smaller subset of folks.
I imagine the empowerment will wear off eventually, and most will return to buying/consuming what a few create? We'll see.
σ Orionis retweeted
BREAKING: OpenAI might have stolen another major proof.
In a detailed Mastodon post, which I report in full in the comments, Andreas Thom presents several pieces of evidence suggesting that OpenAI may have trained Astra on conversations in which he and Gábor Kun were working on Gromov’s soficity conjecture, one of the ten problems OpenAI later announced Astra had solved.
I know Andreas. We met several times early in our careers. He is an exceptional mathematician, a leading expert on sofic and hyperlinear groups, and one of the most respected scholars in the field. He has spent two decades working on this problem.
If his account is correct, this is not a minor dispute over attribution. It would mean that unpublished human work was absorbed into a model and then presented to the world as a breakthrough by the model itself.
And if the allegations raised by Levent Alpöge, Tristan Buckmaster, and now Andreas Thom are all substantiated, we are no longer looking at isolated incidents.
We may be looking at one of the greatest intellectual scandals in the history of science.
AI is not discovering new mathematics.
AI is stealing human discovery.
General intelligence: Fable 5.1 ≈ 3–5% better
Hard technical/scientific agent work: Astra ≈ 5–20% better, highly task-dependent
Coding: essentially same tier; Fable wins the aggregate agent score, Astra wins some important individual coding benchmarks
Computer/tool use: Astra clearly better
Overall: tie, with different specializations
Getting tired of hyping.
1
37
What is the only thing you cannot exhaust?
Chronos with his scythe is time's taking face - dust to dust.
But beneath clock time, beneath proper time, is the giving face: the source that births every next moment.
Sources are mothers.
Her name is prototime. We owe her everything.
CT 4.0.2 is out.
1
19
The Jacobian guarantees a coherent local world. It does not guarantee that the global world has only one branch.
hello there the jacobian conjecture is false thanx to my close friend akhil for asking about it and my other close friend fable for working during the world cup final
((1+xy)^3 z + y^2 (1+xy) (4+3xy), y + 3 x (1+xy)^2 z + 3 x y^2 (4+3xy), 2 x - 3 x^2 y - x^3 z): \C^3\to \C^3, has jacobian determinant -2, and sends (0, 0, -1/4), (1, -3/2, 13/2), and (-1, 3/2, 13/2) to (-1/4, 0, 0)
1
40
CT 4.0 is public. Don’t believe it - try to break it. Find a hidden assumption, circular step, failed derivation or conflicting observation. If its Lorentz/Minkowski and Born/Malus chains survive independent audit, CT provides a common foundation for relativity and QM.
There is no prize - except discovering that existence itself may be immortal.
You may discover that you were never merely the temporary record you took yourself to be.
28
σ Orionis retweeted
Average Codex users:
> keeps getting rate-limit resets
> gets 10% more usage
> 5-hour cap temporarily disappears
> codex keeps working until the task is finished
> Sleeps peacefully
Average Claude Code users:
> Watches the weekly limit disappear in real time
> Hits the cap halfway through the task
> Claude stops immediately
> Begs Anthropic for another rate-limit reset
> Anthropic announces another “final” extension
> Starts panic-coding before access disappears again > Ends up in the hospital
σ Orionis retweeted
An IBM mathematician spent 3 years convinced he was the worst programmer at his company at work.
He built to escape that embarrassment became the first high-level programming language in history. Every line of code running on Earth today traces back to that one act of shame.
His name was John Backus.
He was born in 1924 in Philadelphia, the son of a wealthy stockbroker who expected him to follow the same path. He failed out of the University of Virginia. He dropped out of Haverford College. He enrolled in a medical program in the Army and decided he hated medicine. He spent years doing exactly nothing the conventional way.
Then one afternoon in 1945 he walked past a radio repair shop in New York and got talking to the owner and ended up building a radio from scratch in the shop's back room. Surprising thing is he had never done it before. He stayed for hours. When he left he knew what he wanted to study.
He taught himself mathematics and got into Columbia. From Columbia he walked into IBM in 1950 with a degree and no idea what he was doing.
He learned to program on machines that had no business being programmed. IBM computers in 1950 spoke in machine code. Raw binary. Every instruction written as a string of ones and zeros that told the hardware exactly which switches to flip. There were no shortcuts. No syntax. No vocabulary a human brain could hold in its head.
The programmers who were good at it held the entire machine inside their minds. They saw the binary and felt the logic. Backus could not do this. He wrote programs that were slow, tangled, and embarrassing next to what his colleagues were producing. He was not the worst programmer at IBM. But he believed he was, which amounted to the same thing.
He started building a tool to help himself. Not out of ambition. Out of humiliation.
The idea was simple to the point of seeming naive. He wanted to write mathematical expressions in something that looked like mathematics, not machine code, and have the computer translate them automatically into the binary the hardware needed. He called the project a "formula translation" system. His colleagues thought it was a nice idea that would never work.
The problem everyone could see was speed. Machine code written by a skilled human would always run faster than code generated by an automatic translator. The translator had to make guesses. Guesses meant inefficiency. Inefficiency meant the whole project was a toy.
Backus spent three years proving them wrong.
In 1957 IBM released FORTRAN to its customers. The first compiled programming language in history. The translator Backus built was so efficient that the code it generated ran at speeds within 20 percent of hand-written machine code. Not a toy. Not a curiosity. A working tool that let scientists and engineers write programs in expressions their own minds had generated, and watch the machine execute them.
The adoption was immediate and total. Scientists who had spent careers translating their equations into machine code by hand were suddenly writing programs in hours instead of weeks. Labs that had used IBM machines for narrow tasks started using them for everything. The market for computing changed overnight.
Then something happened that nobody predicted. Other people started building other languages using the same idea. COBOL. LISP. ALGOL. BASIC. Every language built its own translator using the architectural logic FORTRAN had demonstrated. The idea that a computer could read something resembling human thought, rather than the other way around, was now a proof of concept that anyone could extend.
Every programming language that has ever existed was built on the answer to the question Backus asked because he was ashamed of the code he was writing.
He won the Turing Award in 1977. The committee citation said his work had made it possible for more people to use computers for more things than any other single development in the history of computing.
He said in the acceptance speech that he had not set out to change computing. He had set out to stop writing bad code.
The gap between what you are bad at and what you are trying to fix is usually where the real invention lives.


















