miniblog.

← Back to all posts
Jan 6, 2019 at 22:00
The Stockfish chess engine requires patches to pass a test: it must beat the old version a sufficient proportion of the time. This introduces an interesting problem: what if a patch set makes it stronger, but applied individually they make it worse?
Photo
Creating my first test
Creating my first test
The Stockfish testing framework. Contribute to official-stockfish/fishtest development by creating an account on GitHub.
X / ?

Related Posts

Aug 26, 2018 at 22:26
@cstanhope runkit: https://runkit.com/ is a notebook-style UI that has stronger reproducibility guarantees AIUI. Jupyter is still exploring this space and I don't think there's a standard tool for enforcing reproducibility there.
503 Service Temporarily Unavailable
Aug 24, 2018 at 21:22
Computers are stronger than humans at chess and go. They're even winning at arimaa, which was specifically intended to be hard for machines. Watching AI play Quake-style CTF recently, I've realised I no longer have intuitions on which games are hard for computers.
Jun 14, 2018 at 22:24
Really cute video demo of the hypertext documentation in Symbolics Lisp Machines: https://youtu.be/7DxYj32cvoE The speaker carefully explains why links are a good thing! It also lists history in a pane (rather than a back button) and has a stronger notion of navigating hierarchies.