Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Question for chess experts

Author: Telmo Escobar

Date: 20:45:12 03/20/02

Go up one level in this thread


On March 20, 2002 at 15:13:58, Gian-Carlo Pascutto wrote:

>In the Dutch championship 2001, my program reached the following
>position with black against Rebel Century
>
>[D]r2q1rk1/pbp1bppp/1p2pn2/8/3P1B2/2PB1N2/PP2QPPP/R4RK1 b - - 0 1
>
>It played the losing move a5?, and proceeded with the 'plan' of
>pushing it further along the a-file.
>
>Nowadays, it would find the correct c5 after about a minute
>of search, but it is clear by looking at the variations that
>it is more due to luck and that it is still not really understanding
>what is going on.
>
>I am wondering if one of the strong players is able to explain
>why c5 is good and a5 is not. I would play c5 myself, but I
>have to admit I would not be able to explain why it is better
>either. I just know it is good to hit at the d4 pawn with the
>c pawn if possible, but I suspect there is more to it?
>
>Analysis by Sjeng 12.10:
>
>11...Qd7 12.Rfe1
>  ²  (0.33)   Depth: 2   00:00:00
>11...c5 12.dxc5 Bxc5
>  =  (0.23)   Depth: 2   00:00:00
>11...c5 12.dxc5 Bxc5
>  =  (0.23)   Depth: 2   00:00:00
>11...c5 12.dxc5 Bxc5 13.Rad1
>  ²  (0.39)   Depth: 3   00:00:00
>11...Qd7 12.Ne5 Qa4
>  ²  (0.34)   Depth: 3   00:00:00
>11...Nd5 12.Bd2 Qd7
>  =  (0.23)   Depth: 3   00:00:00
>11...Nd5 12.Bd2 Qd7
>  =  (0.23)   Depth: 3   00:00:00
>11...Nd5
>  ±  (0.73)   Depth: 4/16   00:00:00
>11...Qd7 12.Ne5 Qd5 13.f3
>  ²  (0.42)   Depth: 4/16   00:00:00
>11...c5 12.dxc5 Bxc5 13.Rad1 Qd7
>  =  (0.19)   Depth: 4/16   00:00:00
>11...c5 12.dxc5 Bxc5 13.Rad1 Qd7
>  =  (0.19)   Depth: 4/16   00:00:00
>11...c5 12.dxc5 Nd5 13.Be5 Bxc5
>  =  (0.25)   Depth: 5/16   00:00:00
>11...a6 12.Rad1 Nd5 13.Qe4 Nxf4 14.Qxb7
>  =  (0.22)   Depth: 5/16   00:00:01
>11...a6 12.Rad1 Nd5 13.Qe4 Nxf4 14.Qxb7
>  =  (0.22)   Depth: 5/21   00:00:01
>11...a6 12.Rfd1 Bd6 13.Ne5 c5 14.Bc2
>  ²  (0.26)   Depth: 6/21   00:00:01  391kN
>11...a6 12.Rfd1 Bd6 13.Ne5 c5 14.Bc2
>  ²  (0.26)   Depth: 6/21   00:00:02  616kN
>11...a6 12.Rad1 Qd7 13.Rfe1 Nh5 14.Bd2 Nf6
>  =  (0.20)   Depth: 7/21   00:00:03  1295kN
>11...a5 12.Rfe1 Bd6 13.Ne5 c5 14.Qe3 Qe7
>  =  (0.19)   Depth: 7/23   00:00:04  1995kN
>11...a5 12.Rfe1 Bd6 13.Ne5 c5 14.Qe3 Qe7
>  =  (0.19)   Depth: 7/24   00:00:05  2455kN
>11...a5 12.Rfd1 a4 13.Ne5 a3 14.b3 Qd5 15.f3
>  =  (0.20)   Depth: 8/24   00:00:08  4809kN
>11...a5 12.Rfd1 a4 13.Ne5 a3 14.b3 Qd5 15.f3
>  =  (0.20)   Depth: 8/27   00:00:09  5339kN
>11...a5 12.Rfe1 a4 13.Bc2 a3 14.b3 Nd5 15.Qe4 Nxf4 16.Qxb7
>  =  (0.18)   Depth: 9/27   00:00:26  16234kN
>11...a5 12.Rfe1 a4 13.Bc2 a3 14.b3 Nd5 15.Qe4 Nxf4 16.Qxb7
>  =  (0.18)   Depth: 9/31   00:00:34  20026kN
>11...a5 12.Rfe1 a4 13.b4 a3 14.Ne5 Qd5 15.f3 Nd7 16.Qf2 Nxe5 17.Rxe5
>  =  (0.21)   Depth: 10/31   00:00:52  33182kN
>11...c5 12.dxc5 Bxc5 13.Rad1 Qe7 14.Rd2 Rfd8 15.Rfd1 Rac8 16.Bc4 Rxd2 17.Rxd2
>  =  (0.13)   Depth: 10/31   00:01:35  63374kN
>11...c5 12.dxc5 Bxc5 13.Rad1 Qe7 14.Rd2 Rfd8 15.Rfd1 Rac8 16.Bc4 Rxd2 17.Rxd2
>  =  (0.13)   Depth: 10/31   00:01:40  64914kN
>
>--
>GCP

 This is a very interesting question. Somebody has stated his opinion that Sjeng
simply isn't good. That statement doesn't look as particularly clever for
several reasons. Somebody else has underlined the lack of kindness of that
person. I don't know, I often don't understand why human beings behave as they
do. For me, one point is simply that Sjeng is a very interesting engine, capable
of defeating even the currently regarded as top ones; so, trying to putting down
it doesn't make sense.

 But the main point is that, as you correctly state, Sjeng finally chooses
11...c5 but without apparently understanding why that's the best move.

 The point is not about Sjeng. The point is that, I guess, most chess programs
-maybe each one of them- face a particular difficulty here, as they hardly can
do the kind of reasoning a human player could do.

 Human beings don't prefer 11...c5 due to deep calculations or careful
examination of subtleties. They see instantly that that move is good, or at
least plausible, while 11...a5 is very bad, due to esthetic -or moral-reasons.
One move looks elegant, another looks ugly. In turn, these apparently vague
feelings are based upon considerations about different responsibilities. The "a"
pawn, it doesn't have much responsibility. It only wants to queen or to favor
its side by means of some tactical trick. Period. But the "c" pawn, it feels a
big burden over its shoulders, poor thing. It carries several positional
responsibilities. One, putting under fire White's centre. Due, having -as soon
as possible- to make place for its rook to enter the scene. Another one,
watching all the time its companion pawns in order to care for the health of
their formation. Should Black play 11...a5, then the "c" pawn has to face the
ridiculous: now it hardly could advance, to avoid creating obvious weaknesses
(on b5, say). It looks as if the "c" pawn could just stay in place for the rest
of the game. But no, a "c" pawn can't do that. An "a" pawn is a soldier, but the
"c" pawn is almost a general. Should it stay quiet, its side couldn't do
anything active.



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.