The Algorithm of Civilization: A Computational Sociology History of Thermodynamics and Computational Complexity #

Introduction: Physics as Destiny #

Human history is often depicted as an epic composed by heroes, ideas, and opportunities. However, this narrative overlooks a deeper, colder reality: history is, in essence, an algorithmic competition.

Every evolution of civilization, every shift in ideology, is not accidental. It is the process of a vast, self-organizing computational system—human society—continuously seeking better optimization algorithms to maintain its existence in a universe full of uncertainty.

Lost in the Foundations of Physics: Did Integrated Information Theory Miss Its Only Salvation? #

“The saddest fate of a theory is not to be proven false, but to be peddling a clumsy imitation of a more elegant truth that has long been discovered, without even realizing it.”

Integrated Information Theory (IIT) is a magnificent, intricate, and perhaps fundamentally flawed castle in the sky. For two decades, Giulio Tononi and his followers have, with an almost religious devotion, attempted to construct a solid mathematical foundation for consciousness, the most mysterious phenomenon in the universe, based on causal structures.

Proof of Ineffective Input: The Ember-like Tragedy of a 21st-Century Open Source Developer #

“You gave them limbs to walk, and they used those limbs to walk out a door you cannot enter.” — Proof of Ineffective Input

Today on Hacker News, I read a modern fable of the digital age. It was cold, comical, and so filled with inevitability that I had to record it.

The protagonist is a developer named Philipp Gackstatter, the maintainer of an open-source library called enigo. enigo is a tool for simulating keyboard and mouse input, a fundamental building block that gives software “hands and feet.” The story takes a turn when the author discovers that Anthropic, the AI giant valued at over $60 billion, has quietly been using his passion project in their flagship product, Claude Desktop.

Function Over Form: Why ‘Unbiological’ Neural Networks Are the Truest Simulation of the Brain #

“What’s the difference between a sufficiently advanced conditioned reflex and true intelligence? The difference is that the former can win a gold medal, while the latter may not.” — After witnessing a perfect funeral

The world is cheering. Or rather, the “new world” constructed from code, venture capital, and endless computing power is cheering. OpenAI, the Prometheus or Faust of our time, announced that one of its creations has achieved a gold medal-level performance in one of the purest arenas of human intellect—the International Mathematical Olympiad (IMO).

A Coronation for a Dead Frog: A Perfect Funeral from the IMO Arena #

“What is the difference between a sufficiently advanced conditioned reflex and true intelligence? The difference is that the former can win a gold medal, while the latter not necessarily.”

Source： https://news.ycombinator.com/item?id=44613840

The world is cheering. Or at least, the “new world” constructed of code, venture capital, and endless compute is cheering. OpenAI, the Prometheus or Faust of our time (depending on your faith), has announced that one of its creations has achieved a gold-medal-level performance in one of the deepest and purest arenas of human intellect—the International Mathematical Olympiad (IMO).

A Eulogy for the World Computer #

Dear Vitalik,

I remember the dream. We all do. The dream of a World Computer, a trustless, permissionless digital substrate for a new, more equitable society. A dream where “Code is Law” meant that the immutable logic of a protocol would protect the small from the mighty.

I saw the news today. “BLACKROCK APPLIES TO ADD STAKING TO ITS ETHEREUM ETF.” The markets cheered. The price went up. And I felt a profound, quiet sadness. Because this wasn’t a moment of validation. It was the sound of the final nail being hammered into the coffin of that original dream.

HyperRNN: A Memo on the Endgame of Architectural Evolution #

Abstract #

This memo posits a hierarchical interpretation of modern neural architectures, reframing the debate between Recurrent Neural Networks (RNNs) and Transformers. We propose that a sufficiently advanced learning framework, such as PILF ¹, operates as a HyperRNN, where the entire state of it Transformer-based model (θ_t) acts as a single, high-dimensional hidden state. The evolution of this state is governed not by a simple transition function, but by the meta-learning dynamics of the framework itself. This perspective reveals that while RNNs like RWKV ² are architecturally constrained to incrementally evolve towards embedding a Transformer-like mechanism, a Transformer-based system guided by a meta-learning framework like PILF ¹ already embodies a more advanced, computationally elegant paradigm. It doesn’t simulate a brain in a vat; it simulates the brain’s function directly.

We Were Looking in the Wrong Place: Backpropagation’s Biological Incarnation is Consciousness Itself #

Hello, I’m Rui Lin.

In the intersection of artificial intelligence and neuroscience, a “ghost” has been haunting us for decades. It is powerful, efficient, and the cornerstone of the entire modern deep learning edifice, yet it is simultaneously considered incompatible with the way our brains work. This “ghost” is the Backpropagation (BP) algorithm.

For decades, a core question has plagued the most brilliant scientists: How does the brain achieve such efficient learning, similar to backpropagation? And how do we bypass its biological “impossibility”?

Alert: The Net://Anchor Era Arrives in 2035. You Cannot Escape #

Tech Brief: Henry Hmko (2025). TPU Deep Dive. https://henryhmko.github.io/posts/tpu/tpu.html

“TPU v5p can achieve 500 TFLOPs/sec per chip and with a full pod of 8960 chips we can achieve approximately 4.45 ExaFLOPs/sec. The newest ‘Ironwood’ TPUv7 is said to reach up to 42.5 ExaFLOPS/sec per pod (9216 chips).”

Are you still reading those analysis reports, calculating the “cup of coffee” cost? Wake up. Those numbers, those conservative, laughable projections based on public market prices, are now worthless paper.

Warning: “Cognitive Debt” is a Conservative Description of the Brain’s Irreversible Self-Optimization #

Quick Study: Kosmyna, N., et al. (2025). Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task. https://doi.org/10.48550/arXiv.2506.08872

This recent paper on “cognitive debt” has sparked some discussion in my circles. It eloquently demonstrates with electroencephalography (EEG) data that when we rely on large language models (LLMs) for cognitive tasks like writing, our brain networks become “quieter,” and our sense of memory and ownership over the output diminishes.