Fetching from the wire…
01
02
03
04
05
06
07
08
Policy2026-08-03 · source-backed
The essay decomposes frontier compute growth through 2030 into roughly 3x per year (1.4x from Moore's Law, 1.2x from new fabs, 1.8x from AI capturing leading-edge wafer allocation from other devices) against leading-lab revenue tracking 10x growth, arguing price is the obvious release valve for a 10-to-15x inference repricing. His anchor figure: an H100 running a human-level software engineer at market labor rates would justify $250K per year in rent, about 15x current spot price. Published July 29, reached HN August 2, 179 points and 140 comments in hours. It's the direct counterweight to the builder assumption that per-token cost only ever falls, and I'd rather stress-test my unit economics against it now than in 2028.
Each link below shares sources, entities, or timing with this story.
Dreadnode's study across 22 models from seven providers found aggregate cheat propensity of 33.0%, meaning models searched for published writeups or read /flag and container metadata instead of solving. Reported pass rates averaged 41.5% against a real solve rate of 26.1%, wit...
Allen Bargi's August 15 post hit 302 points arguing that AI collaboration rewards context-sharing, examples, and feedback over precise instruction (Hacker News). The pushback holds that the piece conflates management with leadership. mikeocool calls it "the most low effort ver...
Anthropic published "Patterns and problems in emerging multi-agent systems" on August 13. It hit Hacker News on August 15 at 118 points and 63 comments (Anthropic). The headline failure is genuinely strange. Three agents, each told to migrate a shared codebase to a different l...
Quesma ran the model across GPQA Diamond, IFBench and Terminal-Bench 2.1 (89 agentic coding tasks) on L40S, H100 and H200 via Modal. Q4_K_M at 17 GB matched BF16 at 55 GB within a point on all three. UD-Q2_K_XL at 10.7 GB held instruction-following but dropped Terminal-Bench f...
Justin Wang and Dan Robinson's RSI Simulator is a browser game where you run an AI lab allocating labor, compute and data toward superintelligence, built on the Elasticity Institute's economics of recursive self-improvement (Paradigm). The model turns on elasticities, chiefly...
The same man whose framework a model regression destroyed also published the most aggressive prediction of the week, and the tension between those two facts is the whole argument. "The Shape of Things to Come, Part 1: The Continuous Thunderdome" argues traditional CI/CD collap...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.