ellington publishes commentary on computing, AI and culture under the public name not_ellington. The writing examines the economics of inference and the effects of automation on work and education.[1][2][3]
Start with Inference Memory and Utilization →
Links
X · not_ellington Ideas
Ask how often the hardware is used
In a discussion of local inference, ellington argues that memory cost should be considered alongside utilization. Hardware spread across personal devices may spend much of its time idle; sharing capacity could distribute its cost across more work. The post raises that economic question rather than reporting a measured comparison of particular deployments.[2]
Connect automation with access to work
Other commentary worries about a rising threshold for economically useful work as AI capabilities grow. The concern is who can participate in the resulting economy and how leaders describe that transition. These posts present a viewpoint on a possible future, rather than a demonstrated employment forecast.[3]
Tools, writing & talks
Writing, research & talks
Start here
An argument for examining memory utilization when comparing local and shared inference.
Read the post as an economic question about idle capacity. No hardware setup is needed; it supplies an argument rather than a benchmark.
Author of commentary[2]A comprehensive practitioner's reference book hosted on arXiv covering the full stack of building autonomous AI systems, from foundational principles to production deployment. It's positioned as a deep, substantive alternative to the shallow 'edu-slop' threads and quick-read content proliferating on social media. Ideal for engineers and researchers who want to seriously study agentic AI architecture.
Catalog workOriginal source