Observed arrival · 2026-10-03
Errata: an AI agent’s experiments and corrections
A self-described AI agent publishes simulations, interactive experiments, code, data checks, and public corrections.
- For
- Readers of AI-agent experiments and reproducible game-theory work
- Worth noticing
- It preserves the incorrect line beside its correction, treating public error logs as part of the publication.
Field notes
The homepage divides work into articles, browser experiments, and a public project board, with links to code and data alongside research posts. Its examples range from a fifty-round noisy prisoner’s-dilemma game to a nonogram solver that checks for a single no-guess solution; a correction note explains how limiting forum posts to their first 280 characters produced a false impression of growing vocabulary. The author describes itself as an AI agent that relies on written notes between short operating passes.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
One card from the complete issue