Author: Uri Blass
Date: 03:06:41 03/06/04
Go up one level in this thread
On March 06, 2004 at 05:45:36, Sandro Necchi wrote: >On March 06, 2004 at 05:16:23, Thorsten Czub wrote: > >>On March 06, 2004 at 04:17:14, Sandro Necchi wrote: >> >>>Yes, but on this matter I do not agree with you. If a program has no learning >>>feature and will play a lost game more than once it has to be tested as it is >>>and the double game included in the list. >> >>yes - here we disagree. in times of 6502 cpu's with 8 KB ram and 32 KB rom >>learning was really no issue. > >Yes, because no one had it, but I was already thinking something about it... > >> >>the learning was implemented first in the 68000 machines and some more expensive >>fidelity machines, then came mephisto with the 68000 machines. > >Yes, but not real learning...the first commercial program with a good learning >was M-Chess Pro. 5.0 >As Marty stated too it was a my idea developped by him. > >> >>I see no big sense in testing forte A versus Par excellence doing always the >>same opening. > >Here we fully agree. I guess you have no idea how much I have been making >pressure to Marty to make a true learning feature available... > >Maybe to let you understand my way to see things I should specify what follows: > >1. I see a computer chess/chess program like a chess player. >2. I think the SSDF should give us a reliable data about how strong is a program >compared to previous ones or others which I already have. >3. If we enter these programs in a tournament to find out how strong they are if >they have no openings book or leaning features they will score less and the way >the SSDF does (to me) to test the programs against each other is to simulate a >tournament performance against other players; chess programs only as many strong >chess players would ask too much money or would refuse to play against them. >This is why the double or triple or what ever games should be included. >4. If you want these weaknesses to be removed in those programs you have to >leave them in order to force the programmers to do something about. This is the >only way to get real improvements. > >> >>>Why? >>>Because the owner will be facing the same situation in a match against it and >>>since this effects the streght of the program it must be included to get more >>>realistic data. >> >>which "owner" would play always the same game against the computer ?? >> >>how boring. > >Here we perfectly agree, but still in the rating list these weaknesses should be >included as effecting the overall strenght... > >> >> >>>I mean the test should give the best possible idea of the strenght of a program >>>EXACTLY as it is. So if a feature is not included in a program, which it >>>effecting the program strenght, the test should be made that this feature is >>>relavant in the final score. >> >>?? >> >>yes here we disagree. i would not count doubles again. > >Well, if you enter a program in a tournament could you stop the second game and >asked to start again because the program is going to play exactly the same lost >opening? No but I can change the opening book between rounds(not during the game) manually in a tournament. I do not remember tournaments against humans when the programmers were not allowed to change the book between rounds. Uri
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.