A 26 billion‑parameter mixture-of-experts LLM loses its mind, burning three War & Peace’s of tokens uselessly as it searches for memories it never formed…As I continue to cosplay as a SysAdmin, we have:⏺ Bash(python3 -c “ import json…) ⎿ {“
The Strait of Hormuz is closed, oil is climbing, and both sides insist the other will blink first. Iran’s leaders think they have found a structural advantage over Washington: they can outlast a US administration that fears swing‑state vote
Jason’s subheadline: “The exploit shows the extreme risk of offloading technical support to AI”. Yes indeedee, it does. The Vergecast <https://www.theverge.com/podcast/944058/ai-laptop-nvidia-build-gemini-spark-vergecast>—the Flagship Podca
Friedrich Engels saw socialized production clearly—but missed the ethnonational maps that would colonize the workers’ heads. Ernst Gellner’s theory of nationalism and Charlie Maier’s theory of applied-science economy zero-sum interest-group
A platform built for writer–reader relationships now has to survive venture capital, discovery algorithms, and AI intermediaries in a world of frictionless payments, neo-Nazi edge cases, TikTok brain-melt and ad-driven outrage brain-hacking
Reading, arguing, & prompting as different ways of making absent minds speak, from Machiavelli’s study to the fan‑whir of an M5 Max., for it is a fact that black squiggles on the page and linear algebra behind the screen both become voices
Soaring semiconductor orders are supposed to signal a bright future for investment and jobs; but I cannot help but suspect the AI unicorns and those donning AI-unicorn clothing sprinting to IPOs are sending a very different message…We have
“Not much of the kinds that we are used to”, says Noah Smith. His subhead: probably a lot, but not necessarily the kinds people have made money on so far. It is not that scaling laws are nearly exhausted for the machines. It is that the sca
Digital Gods, real costs: why a rational world would see the doom of the foundation‑model-builder IPO, because the AI labs are highly unlikely to ever get profits, let alone hyperprofits. Inference never becomes sufficiently cheap, AI-entit
On logs, Latin, and linear algebra: learning from a stochastic parrot; somewhat awkward questions about agency and pedagogy arising from working through one ridiculously knotty sentence of In Catilinam with an LLM…OpenClaw (now powered by o
Ah. Now I remember this! Mike Beggs rejecting the arguments of those jeering leftist morons who claimed Keynesian policies would not work in 2008-2010 but keeping the jeer. Still, I do want to have a few passages from this to hand should th
How a bad book, a bogus formula, and a hunger for favor led Kevin Hassett to discover that he could lie his way up the ladder as a career, and it would pay. Republican politicians and journalists call it “economic advising”. But that really
We have built an economic régime where GDP climbs, markets soar, and the typical worker stands still unless the labor market is white‑hot. And if inflation fears keep us from ever running the economy hot again, real median wages may be stuc
Marx stuck to his guns on the theoretical adequacy of the labor theory of value. Plus he recognized the role of competition in equalizing profit rates. This contradiction was not because he was unaware of the “transformation problem”: he wa
This picture of a Ukrainian president in front of a Swedish fighter jet may be a sign of a significant change in the wind: if opinion, industry, and force structure align, a northern VIking Ráðagerð could become the mid‑continent super‑regi
Aghion, Bergeaud, & Garicano claim America’s tech productivity-growth lead is materially widening a US vs. EU real wage and standard-of-living gap. Product wage yes. User wage no. This is—or ought to be—elementary. And I at least, am horrif
The Trumpists say “victory”; Tehran calls it “reconstruction payments”; the meter on the Strait of Hormuz is running either way; from “maximum pressure” to war indemnities and investment funds, Trump’s Iran adventure is ending in a very exp
We scaled “attention is all you need” into an industrial‑scale stochastic parrot farm, then bolted on agents and tools until it started to look somewhat more like thought. Now the engineering reality—fabs, power, and eye‑watering token bill
Dan’s subhead: These are not very bright people and things got out of hand. My take: when your grand strategy is based on hoping the other side lets you pretend you’ve won, things do indeed easily get out of hand when you confront an advers
In which a frontier LLM model tells me to configure a new chatbot to use model “/anthropic/claude-sonint-4-6”; yes, this only means that somewhere in the training data some poor SOB mistyped “sonint” for “sonnet”; but nothing with even spar