Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: About Kramnik Anti-computer play.. :-((((

Author: Robert Hyatt

Date: 06:48:52 04/25/01

Go up one level in this thread


On April 25, 2001 at 02:42:16, Tanya Deborah wrote:

>
>
>Here is a little interview about * Kramnik-Junior 6 game * in the past Dortmund
>GM tournament in where Kramnik beat Junior in a very easy way.
>
>Please read this :
>
>
>ChessBase: Did you prepare for today?s game?
>
>Kramnik: Yes, it was generally my plan to get the computer into time trouble so
>it would blunder something...
>
>CB: And seriously?
>Kramnik: Okay, I played some training games with Fritz, and I tried all kinds of
>setups, because to go for the main lines against the computer makes no sense at
>all - you simply forget something and the computer never does this at all. That
>is why you have to find an interesting setup where the computer can go wrong. I
>tried several different things and finally decided that this setup is the best:
>Stonewall, 1.d4 d5 2.e3 and then the computer always plays 2...Nf6 and after
>3.Bd3 it is very good that the computer plays 3...e6. It is okay but it gives
>you very pleasant type of play against the computer. Then you go 4.f4 and
>finally you get what you want. I tried several setups, as I already mentioned,
>but in this setup the computer was doing extremely badly. In one training game I
>mated the computer very similar to this, even faster. I think in 25 moves.
>
>
>CB: So actually you got what you wanted in the first four moves.
>
>Kramnik: That?s not the end of the story. I didn?t get any advantage out of the
>opening, maybe my position was even slightly worse, but I was happy with it,
>because it is exactly the position you need to get against the computer. The
>objective evaluation doesn?t really matter so much. I expected the computer to
>go wrong at some point and it did so. 13...g6 and 15...Nxd2 were very bad. But
>it was very natural. In fact when I was backstage during the game I mentioned to
>Piket that I think the computer will play Nxd2 at some point, because this is an
>aweful positional move. And finally in two or three move the computer took on
>d2. I simply understand the mentality of the computer and that is why I am so
>successful.
>
>********************
>
>The Interview is very CLEAR.
>
>I think, that the program that play against Kramnik in October will need to have
>more than a Super fast computer. I think that it will be very good to make a
>Special opening book to avoid some unknown openings that Kramnik will have
>prepare for the machine. A new opening book-with some help by GrandMasters to
>know how to play better against the best Anti computer  technique by Kramnik.
>
>Kramnik will know very good the program and he will prepare for a secure win.
>
>I wish that the program can show to us a very good chess in October, and also  I
>will not like to see Kramnik winning almost all the games.
>
>Kramnik is very dangerous!!! with a copy of the program three months before!!
>Please, we need to make something to avoid the disaster!
>
>Tanya Deborah.


A "small book" won't cut it.  The first thing the program is going to have to
know is how to not lose when white plays a stonewall.  The computer loses these
games for several reasons:  (1) the black pawn on e6;  (2) the black bishop
trapped behind that pawn and cut off from the kingside;  (3) white's space on
the kingside with pawns at d4, e3 and f4.

When white gets ready for g4 and h4, black has the queen-side to himself,
but he has probably castled kingside and gets smothered.  There are defenses
to this.  They are _not_ "book defenses" however.  A GM might well try
f4 and if black plays d5 then d4 and there you go again.  I tried to cure this
for a couple of years with book lines.  Even had a GM helping.  It was hopeless
to fix, until I modified the evaluation to handle the opening directly.



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.