Weak chess engines
For a long time, I’ve thought that there’s been a hole in the computer chess world—there are no chess engines that play weak games with human-like mistakes (that I’m aware of). It’s easy to make a chess engine that plays a 2000+ strength game, but weakening these engines by limiting their search depth or time always seems to produce games with many strong moves and a few blunders. I can’t be completely sure—it could just be paranoia—but it seems like real weak players (i.e. most casual players) make mistakes that are quite different, and more fun to compete against. Are there any engines that play good weak chess, that accurately mimic the style of a human weak player? Am I crazy for thinking that existing engines don’t?
June 13th, 2007 at 4:07
All chess engines, at their core, search for moves according to the MiniMax algorithm. This is why they play cold, calculated, and conservative games, even at low search depths.
I’m developing my own chess engine. You can find it at my website. Please feel free to contact me with further questions about computer playing style. Thanks.
June 23rd, 2007 at 16:21
Yes, I understand a little about chess algorithms. (I need to write a tic-tac-toe solver one of these days just so I can say I did.) It looks like your chess engine uses a minimax search too. Are you planning to change that? Do you have some other way to make your engine more human-like?
Why am I responding two weeks after the comment? Why not?