Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: The Most Difficult Tactical Combination Ever Shocking Computer Analysis

Author: John Merlino

Date: 11:09:35 11/30/01

Go up one level in this thread


On November 30, 2001 at 13:34:19, ALI MIRAFZALI wrote:

>The following Position is Taken from a Game Played by IGM Yuri Shulman
>in a recent tournament in the USA:
>[D] r1b2r2/pp1nqn1k/2pp2pb/4Pp1p/2PP1p1N/2N3BP/PPQ1B1P1/3RR1K1 w  - 1 0
>In this position Amazingly 19.Nxg6!!! Wins by Force .However Fritz 6 does not
>find this move however Chessmaster 8000 does! However there is a slight problem
>Here is the line given by Chessmaster 8000 when run after about 3.5 minutes.
>Time:3:32 Depth 10   Score:0.09  Positions :13062068
>Line:19.Nxg6 Kxg6 20.Qxf5+ Kg7 21.Bd3 Rh8 22.Qg6+ Kf8 23.exd6 Qg5 24.Qxg5 Bxg5
>25.Bh2
>
>The Problem Here is that Chessmaster 8000 changes its mind at higher plys!
>to the inferior 19.exd6. Here is the Analysis
>Time:40:02 Depth 12    Score:0.00  Positions :149376486
>Line:19.exd6 Qxd6 20.c5 Qf6 21.Bf2 Ng5 22.Nf3 b5 23.d5 Nxf3+ 24.gxf3 a6
>25.dxc6 Qxc6.
>
>Now we come to the Most intertesting part of this article;that is why 19.Nxg6
>wins.In the actual game there was 19.Nxg6 Kxg6 20.Qxf5+ Kg7 21.Bd3 Rh8 22.Qg6+
>Kf8 .and now 23.e6!! instead of 23.exd6 as given by Chessmaster 8000.
>So Therefore one has to feed 23.e6!! in the line starting with
>19.Nxg6 to see that 19.Nxg6 actually wins. This is confirmed by chessmaster
>8000 as follows:
>Time:10:54  Depth 11  Score:1.17  Positions : 43387513
> Line:23.e6 fxg3 24.exf7 Be3+ 25.Kh1 Ne5 26.dxe5 Qxe5 27.Ne4 Bf5 28.Qxd6+ Qxd6
>29.Nxd6 Bg4
>In this line if Black tries 24... Qxf7 he will be crushed by 25.Qd6+ Kg8
>26.Re6! Ne5 27.Rf6 Qd7 28.Qxe5 Bg7 29.Rdf1 Rh6 (with a 3 pawn advatage to white
>according to CM 8000).So 19.Nxg6!!! is actually winning.
>In the actual game Black blundered with 23...Rg8? and lost after 24.Bh4 Rxg6
>25.Bxe7+ Kxe7 26.exd7+ Re6 27.dxc8=Q  1-0
>No program other than CM8000 even comes close to suggesting 19.Nxg6!!! .And
>even CM8000 changes its mind and does not see the winning lines.
>It is My opinion that this position should serve as a BENCHMARK for the
>progress of future programs/Computers.It is only when a program can clearly
>see why 19.Nxg6!!! wins can one write chess as pratically solved by Computers.
>I am howeverinterssted in results by other programs.(at this point I donot
>think anyprogram will see this:but I am still curious).

Confirmed. This is what I get on my PIII-600 with CM8000:

Time	Depth	Score	Positions	Moves
0:00	1/3	-0.39	1557		19.exd6 Qf6 20.Bf2 Nxd6
0:00	1/3	0.55	2837		19.Bxh5 fxg3 20.Bxg6+ Kg8 21.Nxf5
0:00	1/4	0.55	5765		19.Bxh5 fxg3 20.Bxg6+ Kg8 21.Nxf5
0:00	2/5	0.72	30855		19.Bxh5 gxh5 20.Nxf5 Qg5 21.Nxd6+
					Kg8 22.Nce4 Qg6
0:01	2/6	-0.54	102128		19.Bxh5 Nh8 20.exd6 Qxd6 21.c5
					Qf6 22.Nxf5 Qxf5
0:02	2/6	-0.04	167301		19.Nxg6 Kxg6 20.Qxf5+ Kg7 21.Bd3
					Rh8 22.exd6 Qxd6 23.Re6 Qxd4+ 24.Bf2

0:05	3/7	0.00	339839		19.Nxg6 Kxg6 20.Qxf5+ Kg7 21.Bd3
					Rh8 22.Qg6+ Kf8 23.exd6 Qxd6 24.Qxd6+
					Nxd6 25.Bh4 Kg8 26.Re7
0:13	3/8	-0.13	921435		19.Nxg6 Kxg6 20.Qxf5+ Kg7 21.Bd3
					Rh8 22.exd6 Qg5 23.Bh2 Nxd6 24.Qe6
					Qf6
0:44	4/9	-0.24	3321800		19.Nxg6 Kxg6 20.Qxf5+ Kg7 21.Bd3
					Rh8 22.exd6 Qg5 23.Bh2 Qxf5 24.Bxf5
					Nf6 25.Bxc8 Raxc8
2:41	4/10	-0.09	12002079	19.Nxg6 Kxg6 20.Qxf5+ Kg7 21.Bd3
					Rh8 22.Qg6+ Kf8 23.exd6 Qg5 24.Qxg5
					Bxg5 25.Bh2 Nxd6 26.c5 Nf7 27.Bg6
4:02	4/10	-0.03	17662730	19.exd6 Qxd6 20.c5 Qc7 21.Bh2 Nf6
					22.Bc4 b5 23.cxb6ep axb6 24.Nf3
5:35	5/11	-0.06	24338766	19.exd6 Qxd6 20.c5 Qc7 21.Bh2 Nf6
					22.Bc4 b5 23.cxb6ep axb6 24.Nf3
					Ra5

From my tests, it appears that CM would need to get to at least depth 19 (from
the original position) to see that 23.e6 is the best move. I'm not 100% certain
of this, though. Just shows how deep even IM's can think.

Note that you can easily get the above output format by right-clicking in the
appropriate "Thinking Lines" window and selecting "Copy Contents". This will
copy the above type of text into the Windows Clipboard, which you can then paste
into your posts.

jm



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.