Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: The Draw phenomenon in Computer- human encounters

Author: Mike S.

Date: 12:47:41 05/07/00

Go up one level in this thread


On May 07, 2000 at 14:46:27, Jerry Adams wrote:

> Am I just imagining this or is there a high number of draws between strong
>humans and computers? Computers are appearing to be the classic draw master,
>perhaps because they lack the creativity of the human player??

The number of draws will decrease sooner or later, but until then I think that
GM's may more often go for a draw by simplifying the position (for example: if
queens are exchanged in an equal position, this reduces the chances to win, but
also those to lose in general). One could have this impression from today's game
Piket vs. Fritz if you are referring to that, but Piket was in time trouble
(btw. should Fritz have played the strong maneuvre ...Nd3 earlier? I didn't
analyse yet).

Sometimes, when a computer program cannot reach more than a draw, it's more a
lack of technique, than of creativity (although both term could be used for the
same situation sometimes). See the following example from a computer game; the
variation after 49.Qd5+ shows the simple winning idea Gnu didn't find:

[D]3q2k1/3P4/1p1P4/p4Qp1/P6p/1P6/1K6/8 w - - 0 49

[Event "1 min per move, P2-333 MHz"]
[Site "Vienna"]
[Date "1998.??.??"]
[White "GNU Chess 3.21"]
[Black "Mephisto Lyon 16Bit 12 MHz"]
[Result "1/2-1/2"]
[BlackElo "2104"]
[SetUp "1"]
[FEN "3q2k1/3P4/1p1P4/p4Qp1/P6p/1P6/1K6/8 w - - 0 49"]

49. Kc2 (49. Qd5+ {White's queen has to go to either e8 or c8 to make use of the
d-pawns; e8 is difficult to reach because of black's king - so the queen goes
for c8:} 49... Kg7 50. Qc6 Kh6 (50... Qf6+ 51. Qc3) 51. Kc2 {Black cannot give
check now} 51... Qf8 52. d8=Q Qxd8 53. d7+ Kh5 54. Qc8 {White has won}) 49...
Kg7 50. Qe6 Kf8 51. Qh6+ Kf7 52.Kb2 {Draw! Gnu was still convinced to have an
advantage here (+1055).} (52. Qh5+ {suggested by Fritz5} 52... Ke6 53. Qe8+
{...and white would have won}) 52... Qxd7 53. Qh7+ Ke6 54. Qg6+ Ke5 55. Qxg5+
Kxd6 56. Qxh4 Qg7+ 57. Ka2 {aborted} 1/2-1/2

I'm not sure that every new top program finds 49.Qd5+ within tournament time.

Regards,
M.Scheidl

Permanent Brain: http://members.surfeu.at/MScheidl



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.