Author: Dann Corbit
Date: 10:58:38 07/20/99
Go up one level in this thread
On July 20, 1999 at 13:49:04, Dave Gomboc wrote: [snip] >There are a couple of groups doing stuff like this already. There is a "Simple >Game Bench", though when I checked it out it was quite undeveloped except for >the GUI-engine interface. I am sure it will move forward with time. Also, >maybe I misremembed, but I actually thought that Dan Andersson, who posts here, >was doing some work like this. There is a third project I know of in this area, >but they haven't gone public with it yet. > >Anyway, with all those people working on it, I'd imagine you'll get your wish >sooner or later. :) This really is an important idea. Imagine all the chess programmers who just pick some technique at random. Then they sit there and shine their apple for hours. But it turns out that if they had just picked the right apple, it would have been chrome plated to start with. What I mean by all of this is that it seems that chess algorithms are not benched against each other very much. So people just pick some technique and then try to perfect it. But what if you chose bubble-sort and your opponent has chosen radix sort? No matter what your skill as a programmer, you are going to get your butt waxed. A systematic and scientific investigation of the algorithms needed could greatly enhance the ability of the programmers to make wise choices. Do we even know the big O efficiency of these algorithms, so that we know the behavior when the problem scales? Do we know the strengths and weaknesses when the board is sparce, when formations are open, or closed? I think this is the most important idea in a long time.
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.