Author: Georg Langrath
Date: 22:32:48 05/05/98
Go up one level in this thread
On May 05, 1998 at 22:55:31, Fernando Villegas wrote: >Hi all: >Two reasons maybe explain why a post does not get any following at all >and rest miserably alone and without any answer: you are writting about >uninteresting things and/or you write about them when people is engaged >in another thing and/or you write a louzy english. I wonder which of >these is the reason nobody gave a shit about what I said weeks ago about >CSTAL. Maybe all. Pity because CSTAl is a great programm, something >really new, and even so has not deserved the volume of comments and >discussion we have expended here, many times, to sometimes the most >egotistics considerations or even unchessic issues, in which I have been >a heavy sinner. > >I think there are at leats two things about CSTAL that deserves a deep >debate, specially from our programmers here: >First, the clear evidence that CSTAL uses a radically different kind of >search and evaluation function. >Second, the fact that that kind of search and ..etc-.. is of such nature >that CSTAL never will get top positions in computer vs computer tournaments. >Third, the fact that the previous point means that our actual system to >measure computer strenght maybe has not much sense after all >After many usages my feeling is that CSTAL was really designed -not just >and only the usual marketing slogan about "human style"- to face people, >not machines, and that means CSTAL does not make his calculation on the >presuposition that the opponent is so tactically clever as he is. It is >clear to me that many times you can hold a game along many moves >against, let us say, Rebel o Mchess, not because you are strong enought >to hold a tough battle, BUT because the program believes you will see >the same things he see. So, they avoid tactical lines because there is a >flaw in the 12 ply without considering that we, humans of experts or >Fide master level at most, probably are not capable of seing that flaw >and had been crushed the same. CSTAL does not make such supossition and >so we goes after wild lines that are flawed, but that are pressing for >us. So and in such original and obvious manner he gets results. Am I >mistaken asuming that that means the operation of something different to >the conventional maximin approach? Ed? Bob? Amir? >Fernando Bad English or not. You are right. I have CSTAL and love it. Another thing is that you yourself can adjust several parameters in the prgram, and that is fun.It is of course interesting which program is strongest, but as I had said before in this forum, it should not be the only thing discussed here. It can be fun to play chess not only machine to machine. Georg
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.