Author: James Swafford
Date: 16:28:17 02/17/01
Go up one level in this thread
On February 17, 2001 at 18:59:14, Dana Turnmire wrote: > When a programmer creates a book for a chess program does he only put in lines >that would not give either side a huge advantage? For example if the Sicilian >Defense is programmed into computer A to play only the best lines, what happens >if an opponent (computer B) decides to play the Sicilian against 1.e4? Would >not computer A play along the lines that would give Black the advantage? > The same goes for the King's Gambit for example. If computer A has only the >best lines for White wouldn't it when playing Black be at at huge disadvantage? > What do programmers look for in openings books? I'll answer for myself here, as there are probably dozens of implementations. I don't enter "lines." I enter "positions." Side to move is one of many parts of a position. At any root position that still qualifies as an opening, my move generator creates the move list. Each one of those moves are made, and the resulting position looked up in the book. Along with each position is a score. The challenge with this approach is assigning a score to the position, which is where book learning comes into play. Another challenge is determining whether to play a book move, or one of the moves (or actually positions) that aren't in the book. I used code in one of my programs to add positions to the book after every game. The book grew steadily over the course of a few months. I wanted to get around to developing a gui front end to the book so I could manually "tweak" some scores, but never got around to it. Hope this helps. -- James
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.