The research notebook of Aaron Bosloper
Signal Kiln is where I publish my research, useful rabbit holes, unfinished questions, and the niche ideas I cannot stop thinking about.
Published research & open threads
The notebook can move from AI to hidden systems, internet culture, strange histories, and any subject that rewards a closer look.
Aaron Bosloper’s critique of Skill-library scaling—and evidence-based case for specialized, quantized models inside agent loops—and why skills alone stop one layer too early.
Measured forgetting, degraded recall, contaminated memories, and recursive model collapse—with the limits stated clearly.
A long-form note on subculture, curation, algorithms, and why scarcity may be moving from production to attention.
The notebook has layers
Signal Kiln distinguishes quick discoveries, complete investigations, and ideas that remain open to revision.
A sharp observation, source, object, tool, or connection worth recording before the trail goes cold.
A researched essay that traces an idea across sources, examples, context, and competing explanations.
A developing argument with its uncertainty left intact—designed to change when stronger evidence appears.
Aaron Bosloper · From the notebook
Signal Kiln is a place to take a question seriously—even when that question is too narrow, strange, early, or interdisciplinary for a conventional publication.
The notes may begin with AI, a forgotten protocol, an obscure interface, a scientific disagreement, or one peculiar detail that reveals a larger system. The subject can change. The commitment to curiosity does not.
“The best research begins with a detail everyone else decided was too small to notice.”
Notes from the kiln
Get new essays, short field notes, sources worth keeping, and the questions still being worked through.
Follow via RSS ↗Email subscriptions aren’t open yet. The notebook currently has private access.