Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Another win...

Author: Robert Hyatt

Date: 08:03:41 11/26/02

Go up one level in this thread


On November 26, 2002 at 10:17:12, Joachim Rang wrote:

>well another shootout between DF7 and Crafty 18.14:
>
>
>
>Crafty 18.14 - Deep Fritz 7
>for bob, 120'/20+120'/20+60' JOACHIM (1), 26.11.2002
>
>W=16.4 ply; 183kN/s; 163.882 TBAs
>B=15.9 ply; 747kN/s; 187.425 TBAs
> 1.Qxe7+ Ka2 2.Qe6+  0.00/17  6:19  Qb3  -0.25/16  7:31  3.Qa6+  0.00/17  6:19
>Kb2 (Kb1)  -0.16/17  7:27  4.Qf6  0.00/17  6:19  Kb1 (Ka2)  -0.25/16  2:57  5.h4
> 0.00/19  6:21  c2  -0.59/16  2:41  6.Qf5  -0.01/18  12:37  Kb2  -0.44/15  2:34
>7.Qe5+  -0.01/18  35:10  Qc3  -0.09/14  4:52  8.Qb5+  0.00/16  21:38  Ka3
>-0.28/15  5:02  9.Qa6+  -0.50/16  2:06  Kb3  -0.53/15  5:06  10.Qb6+  -0.43/16
>2:06  Kc4  -0.53/14  2:59  11.Qe6+  -0.43/17  2:06  Kb5 (Kc5)  -0.81/15  4:35
>12.Qd5+  -0.41/16  2:06  Ka4 (Kb6)  -0.69/15  2:54  13.Qa8+ (Qd7+)  -0.28/15
>8:24  13...Kb3  -1.16/15  5:28  14.Qb7+  -0.22/15  1:12  Qb4  -1.97/16  13:41
>15.Qc7  -0.69/16  1:11  Qd4  -4.84/14  3:41  16.Kg2  -4.87/13  1:14  Qd5+
>-5.44/14  4:00  17.Kh2  -5.39/16  1:11  Kb2  -5.34/15  1:46  18.Qb6+  -5.77/15
>1:11  Kc1  -5.94/15  1:38  19.f4  -5.79/12  1:11  Kd2 (Kd1)  -7.03/16  4:39
>20.Qf2+  -5.94/13  35  Kc3  -7.09/16  3:49  21.Qe3+  -7.26/15  6:01  Qd3
>-7.22/15  1:57  22.Qc5+  -7.56/14  36:10  Kd2  -8.44/15  4:11  23.Qa5+  -8.44/14
> 4:21  Ke2 -8.50/15  4:25  0-1
>
>
>
>it seems to me, that the win, although forced, takes many moves. But we know
>that queenendgames can take very many moves for the winning side.


That is all well and good, but it is not answering "the question".

A different question:  Can a carpenter drive in 100 nails without having one
go "sproing" and shooting off somewhere?  You get the best carpenter you know
and try it and at nail 45, "sproing".  You do this a few times and each time
you get a "sproing".  And you conclude it can't be done.  That's not a proof
of any sort.  For example, several KQPKQ and KNNKP ending positions are probably
impossible for a compute to win.  This kind of testing methodology would suggest
those are all draws.  And it would be wrong.

Playing a game out proves exactly nothing about the ultimate outcome from the
root position.

This can be proven as a win by either a human or a program, all they need to do
is give analysis with the best black move for each possible white move, to show
that _all_ white move at _all_ positions in the tree lead to a loss.  Anything
else is a "vincent proof".



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.