Author: Russell Reagan
Date: 11:21:35 04/20/02
Go up one level in this thread
On April 20, 2002 at 08:47:26, Gian-Carlo Pascutto wrote: >On April 20, 2002 at 08:36:39, Mike Hood wrote: > >>Everything is based on positional evaluation and >>search depth. If the search depth is deep enough, a computer may make a series >>of moves that simulate a strategy, but that's all it is: a simulation; a fake. > >What allows you do conclude it's a simulation, a fake? > >>Strategy is all about looking at the board and planning a series of moves to >>achieve a goal, whether it's a positional improvement or material gain. Computer >>programs don't do this. All they do is look at the current position and choose >>the next move. That's all. > >By your definition, computers are all about strategy. > >What they do is plan a series of moves (the PV) to archieve positional >improvement or material gain (represented by the evaluation). > >Sometimes they discover a new, better strategy (a fail low), or realize >the strategy they are following at that moment is flawed (a fail low). > >Moreover, my program (and others too I guess) can influence decisions in the >evaluation (which directly influences what move is played) by looking at the >position and determining what manoeuvres will be possible later on. > >If that isn't strategical planning, I don't know what is. > >The problem is that programming knowledge like this is hard. If there were >an easy way to program in 'check whether he will be able to shift all pieces >to our kingside in a while without us being able to do something about it or >launch a counterattack on time', then my program would handle the stonewall >like a GrandMaster. But there isn't, so it has to do with some simpler rules >that aren't always correct. So it will mistakes in the planning, and play >a losing strategy. > >-- >GCP That's an interesting idea. Giving your program some kind of knowledge about whether the opponent (or itself) can shift it's pieces to another sector of the board quickly. I think that a lot of these kinds of things are not so much "hard", but just a great quantity of work to develop some sort of algorithm and hammer out the details in it, finding special cases, etc. And most people aren't willing to do that. Not that it's a bad thing, most people have jobs and families which should certainly come before chess programming :) I just think that a lot of what we consider to be so "impossible" for a computer to "know" is really not so hard if we chose to really give it a try. I think success in a computer chess program is really a matter of effort (like most things in life are). It just depends where you choose to apply your effort, and most people don't choose to apply theirs to computer chess. The ones that do are few but they have very strong chess programs. Russell
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.