Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: rising number of db endgame occurances

Author: Ralf Elvsén

Date: 15:40:58 07/02/01

Go up one level in this thread


On July 02, 2001 at 14:22:33, Mike S. wrote:

>I have searched for occurances of endgame database material in computer chess
>games (inluding some comp-human, and probably a few human-human games), based on
>>12.000 games from 1984 on.
>
>The following table shows the *percentage* of occurances (X = Q/R/B or N):
>
>years   # games  P-P   X-P   X-X   PP-P  XP-X   XX-X total
>----------------------------------------------------------
>84-89     305    1,3   0,3   1,0    1,0   3,9   0,3   7,9
>90-95    2646    1,0   1,0   0,9    2,0   4,4   0,9  10,2
>96-98    5335    1,2   3,4   3,6    1,9   9,1   2,3  21,6
>99-2001  4126    1,9   4,4   4,8    2,3  11,0   2,4  26,8
>----------------------------------------------------------
>84-2001 12412
>
>As you can see in the total column, db endgame occurances are rising.
>
>Then, I determined how many plies a XP-X (the most important one) situation was
>at the board in each time segment:
>
>years   # XP-X plies: 4-  20-  40-  D%
>--------------------------------------
>84-89     12         58   25    8   67
>90-95    117         52   16    3   44
>96-98    486         56   24   10   56
>99-2001  454         53   21    9   59
>--------------------------------------
>84-2001 1069         54   22    9   56
>
>I see no real trend here, except maybe in the percentage of drawn games (D%),
>which *may* be slightly rising (which would mean the defender benefits more from
>tablebase use).
>
>Regards,
>M.Scheidl

Thank you for these data. Do you think the increase in the
percentage of occurances is due to higher quality of the
programs, i.e. less early wins due to foolish mistakes?

Ralf



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.