Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: ponder_on ponder_off comparision

Author: Peter Kappler

Date: 17:38:19 07/18/00

Go up one level in this thread


On July 18, 2000 at 18:59:14, Mogens Larsen wrote:

>On July 18, 2000 at 18:04:44, Robert Hyatt wrote:
>
>>If everyone plays equally worse, then
>>no harm is done.  But adding the extra degree of freedom is simply adding more
>>noise to an already complex and unstable system of comparing chess engines.
>
>Just out of curiosity. What makes you think that all programs implement ponder
>equally well? Are you suggesting that a system of comparing chess engines is
>unstable when your programs isn't tuned for a certain parameter and stable when
>it is?
>
>Best wishes...
>Mogens


Well said.  This is one issue where I really disagree with Bob.

Imagine if the following statements were made by a programmer:

1) "My program is designed to play against humans, not computers.  Do not play
my program against computers - the results are meaningless."   (This was a
favorite of Chris Whittington, author of CSTal).

2) "My program isn't designed to play at fast time controls.  Matches played at
a fast time control with my program have no scientific value."  (Another CW
favorite)


Now, two more extreme examples, just to make my point:

3) "My program doesn't use endgame tablebases.  Do not match it against a
program that uses tablebases - this would not be fair"

4) "My program doesn't ponder.  When playing it against other engines, make sure
they aren't pondering, either."


Anybody who releases a winboard-compatible engine should realize that it is
going to be used in single-computer, non-ponder matches.  Knowing this, it seems
a bit careless to ignore ponder=off time management.

--Peter










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.