Topics · Concept
Self-play A program improving by playing itself: generating its own training data, learning from the outcomes, and testing each generation against the last.
Pages 1 On the map 1 page and the topics around it Tap a dot to select it, tap again to open it. Open in the full map →
Form Reviews Notes Talks Books Novels Papers Reports Articles Essays Journals
Topic Aerospace AI-assisted engineering AI for operations AI platforms Antiquity Applied security Architecture Art history Board games Chess Cryptocurrency Cryptography Cybernetics Distributed systems Edge systems Energy Enterprise architecture Finance Game AI Games Genetics Geography Go History Insurance Life sciences Linguistics Literature Machine learning Mathematics The open web Operating systems Palaeontology Philosophy Physics Platform automation Policy Programming languages Psychology Reliability Semiconductors Shogi Simulation Sovereignty Space StarCraft Synthetic biology
Technology Arcanist bhc Durable execution Elixir Go GPUs Haskell Kubernetes Linux OpenBSD OpenTelemetry Pingora Rust TypeScript WebAssembly Zola
Length Under 10 min 10 to 20 min Over 20 min
2026 Oct 28 Self-play as reinforcement learning: the engine makes its own training data, a network learns to predict the outcome, and one unfitted constant cost 176 Elo.
10 min read Game AI Self-play Reinforcement learning Training data