Author: Serge Desmarais
Date: 20:17:30 09/10/98
Go up one level in this thread
On September 10, 1998 at 11:47:35, blass uri wrote: >I think that the value of opening book is in going to positions that the program >understands better than the opponent(for example if the program has good results >in positions with isolated pawn in both sides than the program should play for >these positions. > >I know that fritz3 when it played against humans on a pentium90 did it >and played in a tournament game with white 1.e4 e6 2.d4 d5 3.exd5 exd5 >4.c4 and won against a grandmaster. >They used in this tournament a book for fritz3. > > >Fritz5's book for the ssdf and for sale and for other tournaments against >computers(like paris micro computer championship) is not a book for fritz5 and I >do not understand what is the reason. >I believe they could have better results if they used a book for fritz5. > >Uri Maybe they believe in the "book learning" they implemented? With it and time, Fritz will/should its "favourite" and more efficient lines by itself... I take the opportunity, here, to talk about opening books in general. I think it would be interesting if the chess programs were ADDING moves into their book by themselves. Here is my suggestion : in long time controls (I will not be specific about it) the program could add the first move its opponent play out of its book AND its own reply. Then after the game result is known, it would write down the "learned" info as Fritz 5 does. Next time the same position appears, it would NOT play the same move it tried, but rather start "thinking", looking at all the moves EXCEPT the one played before (which is assumed to be "bad"). If the game was a win, then it would stick to that move, adding its opponent's next move AND its own reply. This way, its knowledge about positions that appear often would get better and better, like for the human players. I tried Fritz 5 against one of the Crafty on ICC. The Crafty played 1...d5 against Fritz 5 1.e4. Fritz won a nice game. Later, same opening moves and Crafty tried something new and lost again. Same thing for the third game but with Crafty trying something new(this Crafty seems to always play the Scandinavian!). Then, in the 4th game, Crafty's innovation was much better and Fritz had trouble to draw! Note that Fritz was always playing the same moves, once out of book. Now, Fritz only plays a different line as White (cause of the drawn game). But it would have been interesting if Fritz had a way to self improve its play too! Serge Desmarais
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.