Author: Peter Hegger
Date: 07:08:26 08/18/00
Go up one level in this thread
On August 18, 2000 at 06:00:47, Stefano Gemma wrote: >I've seen some thread, in WinBoard forum, talking about some clone of chess >program, this make me think about an interesting idea. The sources of some >program are available to public domain, i think about GnuChess, Crafty and so >on. Any programmer can take that sources, modify them and releases a new >program, that is partially made from the original source. Someone take piece of >code (or algorithm) from various author, making that way some "Frankenstein >program". Someone (a few ones, i think) starts from scratch. So we have three >kind of "new" programs: > >1) started from scratch programs >2) variation of some base program >3) frankenstein programs ;-) > >This situation make me think about the way genetic algorithm works... or better: They've done something very similar to an evolving program in the game of checkers. The stronger programs are allowed to re-produce while the weaker ones are allowed to die off. So far, an expert strength checkers program has evolved from one which knew only the rules. The same thing done in chess would be very interesting. Regards, Peter >how evolution itself works. The base programs are the father/mother of child >programs. Child programs can be clones of a single parent (type 2) with some >variation (as occurs in DNA copying errors). Other child programs can be the >result of a crossing the "DNA" of various programs (type 3). All the programs >will compete with its parents and brothers, in the virtual world of >chess-software matches. the stronger programs (Crafty, for sample) are more >interesting for new programmers. This will simulate a natural competion and >selection. The Crafty's DNA will be transferred to new programs, with some >modification. I think this is good, beacause it will give us stronger programs >but... it is even dangerous beacause doing so we will explore just a way to >solve chess-software probles. Luckly, we have programs of the first type. Maybe >they are weeker than the stronger ones, but they give us new interesting ideas. > >I think that we need all the 3 kind of programs, to raise the complexity of this >strange virtual world. > >Ciao!!!
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.