Author: Christophe Theron
Date: 17:51:34 05/30/99
Go up one level in this thread
On May 30, 1999 at 19:31:02, Anatoli wrote:
>Hi !
>Yeasterday I got in an interesting situation when I wanted to show Fritz 5.32 in
>action. One of my friends proposed a very easy task to solve. I said that Fritz
>5.32 played at IG level and could solve most of the problems in seconds. But
>then we gave Fritz 5.32 a well known task by P.Morphy, which we knew
>since our childhood.
>Here it is :
>W. Kc8, Ra1, b6
>B. Ka8, Bb8, a7,b7
>Checkmate in 2. 1.Ra6 ba 2.b7X 1.Ra6.B--- 2.Rxa7X
>Everybody knows this chess puzzle. I set up this position and smiled quite
>loudly. Then I press the Watch button on the main tool bar and proposed
>Fritz to Play. To my big confuse Fritz 5.32 gave 1.Kd7 !!!??? It didn't find
>1.Ra6. Why ? I repeated this task many times but Fritz proposed only 1.Kd7.
>Then I gave this position to all available engines and they solved it in 1
>sec.So, can anybody try this super difficult position on your computer ? Fritz
>5.32 can checkmate now with a knight and a bishop. But will it be able to find
>athis checkmate in 2 ?
>Thanks
>Anatoli
Once again you have just found a "null move" problem. Questions like yours pop
up from time to time. One given program looks stupid in a given endgame
position, and everybody is surprised.
Fritz, like many programs, uses the "null move" algorithm. The authors
themselves recognizes that Fritz uses "null move" (see Fritz description in the
list of participants to the 9th World Computer Chess Championship).
The idea is that if by not playing a move your opponent can do nothing to you,
it is likely that your position is safe enough, because generally there is at
least one move that is better than doing nothing (OK, this is a rough
simplification).
This idea works in many many chess positions, and saves a lot of work (you don't
have to search positions after the given position). But it does not work in some
cases, the most known are "zugzwang" positions.
In your example, after 1.Ra6, if black is allowed to pass his turn, well nothing
can happen to him. So black assumes he has at least one move that will be good
enough, and unfortunately in this position there is no such move. Black is in
zugzwang!
This is a well known problem of null-movers, so generally programmers try to
identify positions where the null move idea could fail. Generally a simplistic
rule is used: "if I have only my king, some pawns, and only one piece, I should
not use the null move idea" (a zugzwang is highly possible).
It sounds like Frans Morsch has tried a more risky rule, because if he had used
the simple rule above Fritz would not have overlooked the mate in 2.
This is NOT a bug. This is the programmer's choice. Given Fritz very good level
of play, we can only say that the choice looks reasonnable, even if sometimes
the program can miss simple things.
Christophe
This page took 0.01 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.