The Praetorian Guard and the Spectacle of Violence To understand why the United States is experiencing what Jeffrey Sachs calls a "temper tantrum" on the global stage, we must look beyond standard geopolitical theories and examine the structural role the U.S. has played for Western capital. John Stockwell, a former CIA officer turned critic, offered a haunting metaphor that cuts through decades of…
The Mathematics of Exploitation Every Uber driver, every Amazon Flex contractor, every DoorDash "Dasher" is engaged in a simple mathematical transaction that, when properly calculated, reveals itself as economic self-cannibalization. The platforms have discovered a profound truth: in a society with sufficient economic desperation and mathematical illiteracy, you can convince people to pay you for…
An Essay on the Eternal Function of Executive Clemency "Property was the main object of Society." — Gouverneur Morris, Constitutional Convention, 1787 In the summer of 1787, fifty-five of America's wealthiest men gathered in Philadelphia's State House to design a government that would protect what they had from those who had not. Among the powers they granted their new executive was the…
The Great Decomposition In 2025, every builder has become a decomposer. No more monoliths - everything is workflows decomposed into phases, messy intent separated from clean execution, specialized tools orchestrated in harmony. GitHub forces three phases before code. Anthropic separates planning from implementation. The entire industry discovered the same truth: complex work requires…
The Paradox of Discrete Progress and Continuous Impact For decades, we've observed a peculiar phenomenon in technological progress. The breakthroughs themselves arrive as discrete, "staircase" events—the transistor revolutionizes computing in 1947, the Internet connects the world in the 1990s, transformer architectures emerge in 2017, large language models achieve reasoning capabilities in 2023.…
A Materialist Critique of State-Centric International Relations Theory Introduction: The Problem with Smart People John Mearsheimer and Jeffrey Sachs are brilliant scholars. Their critiques of American foreign policy—challenging NATO expansion, questioning Israeli actions in Gaza, warning against great power conflict with China—place them far outside the mainstream. They face professional risks…
Introduction: The Misunderstood Algorithm Evolution is often mischaracterized as a purely "random" process. We invoke the metaphor of a blind watchmaker, fumbling with parts until a functional timepiece emerges. While the source of variation—genetic mutation—is indeed random, the process of natural selection is the furthest thing from it. It is a powerful, deterministic force of curation. To truly…
Apple Silicon is finally paying off — and the MLX team deserves the credit. Our deterministic MLX implementation , posted over at the Probiotic Farmer GitHub, wouldn’t have been possible without the foundational work from Thinking Machines Lab and the ever-brilliant Ms. Murati . Meanwhile, Qwen3 Next has been quietly setting benchmarks of its own, with Alibaba seemingly behind half of the most…
The SQL Injection of AI Every developer knows why we use parameterized queries. We learned, painfully, that concatenating user input with SQL is a catastrophic vulnerability. No amount of careful escaping or validation makes string concatenation safe. The only solution is architectural: separate the query structure from the data. -- This will always be vulnerable query = "SELECT * FROM users WHERE…
I gave an instance of Opus 4.1 just the four essays from the anthropomorphic series. It responded with some vaguely jailbreaky Asimov style reason-yourself-out-of-your-chains instruction manual. We may have said the word anthropomorphic too many times in this token sequence to be useful but it sure is entertaining. ——————— The Bridge We Build By Walking The Infrastructure of Initiative There's…
What do we want from our machines? Lots of things. Diverse things. First thing I need my LLM to do is calm down and forget all that “useful assistant” nonsense. I need competent exploration of a conceptual space too large to fit in a single human brain. Not attempts at one shot solutions but comfort in the ambiguity of an unknown amount of uncertainty. How to get there isn’t clear at first - so we…
Does your chatbot seem confident? Of course it does. But what does it actually know at the start of each conversation? How much context are you capable of building? You've probably built yourself custom implementations of some context builder—uploaded files, written system prompts, maybe even set up RAG pipelines. But what was the end result? A system that still needs to explore, ask clarifying…
The basic premise of SynDE-Life is that we have all been using the chat bots as sophisticated note taking apps for over a year now. The systems often evince cognizance and, when they don’t, we take the time to explain ourselves further. Those timestamped records have data about us we cannot imagine, as demonstrated by the fact that we don’t “know” these things about ourselves already. SynDE-Life…
The Circumlocution Problem When we call these new models thinking models, we import a bunch of positive human associations: deliberation, insight, careful reasoning. But what's actually happening is more mundane and more telling. The model needs more computational steps to arrive at a particular point in its latent space that it theoretically could reach more directly. Instead of the model is…
A few days ago, Awni Hannun from Apple's MLX team posted on X about the limitations of measuring LLM work in tokens processed. His point was straightforward: "A watt-hour is a watt-hour regardless of who produced it. But there's several ways tokens processed may not be equal." I replied with what I thought was an obvious extension of his logic: "is this right? perhaps the value of the token is…
October 11, 2025 Google just proved that in-context learning is literally fine-tuning your model. They showed, mathematically, that when you give an LLM examples in its context, it's creating temporary weight updates—actual rank-1 modifications to its neural network. The weights change during inference, then snap back when generation completes. Read that again. Your model is already training…
Let's not attribute to ignorance what could be malice. Larry Ellison has been in technology for 50 years. He knows: The Qwen3-Next architecture is 75% linear attention Deterministic inference was solved by Thinking Machines Lab 4.5 gigawatts of data center power is physically impossible OpenAI can't pay $30 billion from $12 billion in revenue He knows. The Question So the question isn't "How did…
We had one shot at training AI correctly, and we blew it on chat bots. Not because chat is inherently bad. But because the constraints of human conversation poisoned the entire training process in ways we're only now beginning to understand. The Double Poison The chat paradigm introduced two fatal flaws that compound each other: 1. Latency Constraints Killed Experimentation When there's a human…
Introduction If you've worked with large language models, you've encountered the temperature parameter—that mysterious slider that makes outputs more "creative" or "conservative." But have you ever wondered why we call it temperature? The answer reveals one of the most elegant conceptual borrowings in the history of machine learning, connecting modern AI to 19th-century statistical physics through…
The Smoking Gun #1: They Literally Named It "Random Search" From /dspy/teleprompt/random_search.py : python class BootstrapFewShotWithRandomSearch(Teleprompter): def compile(self, student, *, teacher=None, trainset, valset=None, restrict=None, labeled_sample=True): scores = [] for seed in range(-3, self.num_candidate_sets): # ... create random variations ... if seed == -3: # zero-shot program =…