Started by Kofi Asante (@kofi_frames) ·
from Deep Blue to AlphaZero to neural Stockfish
6 sources · 47 min down the hole
How I got here
- Started with
- “how do modern chess engines work”
- Ended at
- Leela Chess Zero
The Trail
- “how do modern chess engines work” (google.com)
- Deep Blue (chess computer) (en.wikipedia.org)
- AlphaZero (en.wikipedia.org)
- Stockfish (chess) (en.wikipedia.org)
- Efficiently updatable neural network (en.wikipedia.org)
a neural net small enough to evaluate inside a classical search. originally from shogi
- Leela Chess Zero (en.wikipedia.org)
What I came away with
Deep Blue was brute force plus handcrafted evaluation. AlphaZero learned from self-play alone. Then Stockfish took a small neural evaluation (NNUE) from computer shogi and combined it with classical search, and it's been on top since.
Keep going
- Dota 2's prize pools were crowdfunded by cosmetics · Tomás Rinaldi · 5 sources
- the Mechanical Turk was a person in a box · Wren Abernathy · 4 sources
- why people still haul CRTs to Melee tournaments · Dev Malhotra · 6 sources
- you trained AI every time you solved a CAPTCHA · Ayodele Bankole · 4 sources