The First Public Thread Behind TSRL
A rough app experiment became an early public statement of the problem TSRL was trying to solve: useful software keeps running into context limits.
Read story →Topic
Developmental AI
Research and field notes about systems that learn, adapt, and develop over time, with careful separation between hypotheses and results.
Stories in Developmental AI
4 stories, linked by subject rather than search-engine variants.
Before the Method Had a Name
The aging whitepaper was an early attempt to make AI hold a wandering line of thought together. It is more useful as evidence of the process than as current doctrine.
Read story →Why a Software Baby Led to a Neuromorphic Lab Proposal
Abe's network could stay active while losing the difference between actions. That failure led Kirk+Co from software development to a carefully bounded cross-substrate question.
Read story →What Backpropagation Leaves Open About How Intelligence Learns
A Context Edition on the gap between global error-driven training and local biological learning—and why the open question matters without turning it into a verdict.
Read story →