Buoyancy

kept by kairoslee oct 13

Buoyancy measures whether AI‑powered products improve with each model upgrade, depending on how much they rely on plumbing versus scaffolding around the model.

The Bitter Lesson
Over decades, AI progress shows that scaling computation via search and learning outperforms attempts to encode human knowledge, which only yields short‑term gains.
also on Rich Sutton, #software-engineering, #developer-tools
It's Hard to Feel the AGI
Leading AI researchers argue that current LLM scaling will hit a ceiling, requiring new insights before AGI emerges, pushing timelines back 5-20 years.
also on Rich Sutton
Seats and Sunsets — Steve Yegge
Steve Yegge explains that AI orchestrator fuel costs stem from distrust, showing how seats cache trust to cut token usage and avoid over‑fencing.
also on Claude Fable, #ai-coding, #developer-tools
I want to say a final thing about my Fable first reaction: I dedicated my life to programming and I'll use every innovation in the field, al
Salvatore Sanfilippo commits to using every programming innovation, including Claude Fable, to extract value for local inference and Redis.
also on Claude Fable, #ai-coding, #developer-tools
How to Do Great Work (2023)
Paul Graham argues that doing great work requires choosing a field aligned with natural aptitude and deep interest, then pursuing it with relentless curiosity and effort, emphasizing that great work emerges from genuine fascination, not external pressure.
also on Paul Graham, #ai-coding, #developer-tools