Computer Chess Club Archives


Search

Terms

Messages

Subject: Normal Low Proportion Of Draws

Author: Jay Scott

Date: 13:11:40 11/25/03

Go up one level in this thread


I noticed years ago that computers make far fewer draws than humans. I think
it's because humans are more afraid to lose than eager to win. It's a well-known
human cognitive bias; see the classic 1974  Tversky and Kahneman paper
"Judgement Under Uncertainty". Wimpy humans. Computers don't have that flaw, and
always fight.

Other possible explanations:

- Programs are tuned to be aggressive, to win against humans and to play more
entertaining games, at the cost of making more errors.

- Humans understand drawing better. Humans have far more knowledge about when a
position is drawn and how to draw it, whereas programs often calculate "I am
0.05 pawn better, I am still playing for a win."

If the game of chess is a draw, then theoretically as players become close to
optimal the number of losing mistakes per game should approach zero and most
games will be drawn. Alas, we are still at the stage where the expected number
of mistakes per game is greater than one. :-) My personal guess is that top play
is still very far from game-theoretically optimal, witness the surprising
discoveries made in endgame databases.



This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.