Author: blass uri
Date: 15:28:31 08/11/99
Go up one level in this thread
On August 11, 1999 at 09:49:25, Robert Hyatt wrote: <snipped> >Actually, when you think about the SSDF rating list, this does make a lot of >sense to do. > >I don't do it. But I do see oddball underpromotions for the reason I gave >earlier. But when you are interested in topping the SSDF list, every little >thing helps. I do not believe they do it in order to top the ssdf list because it is not clear if it is productive. It should take less time to prove that promotion to queen is the best relative to promotion to rook if you assume the right reply is taking the promoted pawn because when you promote to a queen it is more easy to see that the opponent must take the queen because the alternatives are bad. I also found that programmers could do other things that they do not do in order to goto the top of the ssdf. For example Fritz's engines way of learning is stupid in the ssdf games. one example: I found that Junior5 won nimzo3.5 and after repeating the same opening failed to repeat the same moves again because nimzo3.5 used different time for its moves and Junior could not use the permanent brain for the same time. Even without learning the result it is logical to repeat the move you used more time for it so Junior should repeat the move it played in the first game because of this reason but Fritz's engines way of learning and logic are 2 things that do not go together. I do not believe that prograsmmmers care for little tricks to get better results in the ssdf list because of this stupid mistake and other stupid mistakes that I see in programs. For example Fritz engines are using tablebases in a wrong way. Chessbase know it for a long time and does not care to fix it to be without mistakes. 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.