Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: correct shootout results

Author: Stephen A. Boak

Date: 00:20:05 01/06/06

Go up one level in this thread


On January 05, 2006 at 15:32:59, Joseph Ciarrochi wrote:

>ooops. there are two shootouts at the bottom of the email. the first shootout is
>something i was doing from a scandanavian position. you can ignore that. the
>second one, ending with "31%", is correct. I past the correct one  below.
>
>( i should note that Rybka gives one of the scandanavian lines I use a +.84,
>thus a much worse evaluation than the one it gives to this position missing the
>f pawn. Yet the scandanvian position has white wining about 55% percent of the
>time, which is not great for black but not as imbalanced as the missing f pawn
>position (given a -.29 evaluation).
>
>
>Shoot out:
>----- Ply = 5 -----
>Fritz 9   0-1
>Fruit 2.2.1   0-1
>Rybka 1.0 Beta 32-bit   0-1
>Toga II 1.1   ½-½
>Spike 1.0a Mainz   0-1
>Ruffian 2.1.0   0-1
>Aristarch 4.50   1-0
>Fritz 8   0-1
>List 512   ½-½
>Dragon 4.6   0-1
>Crafty 19.19   -0.58
>----- Ply = 7 -----
>Fritz 9   1-0
>Fruit 2.2.1   ½-½
>Rybka 1.0 Beta 32-bit   ½-½
>Toga II 1.1   0-1
>Spike 1.0a Mainz   ½-½
>Ruffian 2.1.0   0-1
>Aristarch 4.50   ½-½
>Fritz 8   0-1
>List 512   1-0
>Dragon 4.6   1-0
>Crafty 19.19   1-0
>----- Ply = 9 -----
>Fritz 9   ½-½
>Fruit 2.2.1   ½-½
>Rybka 1.0 Beta 32-bit   ½-½
>Toga II 1.1   ½-½
>Spike 1.0a Mainz   0-1
>Ruffian 2.1.0   0-1
>Aristarch 4.50   ½-½
>Fritz 8   0-1
>List 512   ½-½
>Dragon 4.6   0-1
>Crafty 19.19   ½-½
>----- Ply = 11 -----
>Fritz 9   ½-½
>Fruit 2.2.1   0-1
>Rybka 1.0 Beta 32-bit   0-1
>Toga II 1.1   0-1
>Spike 1.0a Mainz   0-1
>Ruffian 2.1.0   1-0
>Aristarch 4.50   0-1
>Fritz 8   0-1
>List 512   ½-½
>Dragon 4.6   0-1
>Crafty 19.19   0-1
>
>Score White: 31%, 43 Games
>Done

I suspect that evaluation balancing of Pawn (material) vs Dynamism (piece
mobility and attacking potential) requires a deep enough search for the eval
tradeoff to be accurately followed up.

Here's a poor analogy (may be backward, in fact!) to ponder:

What if Rybka Eval is the foresighted equivalent of saying "My (your) fort is
weak, all my (your) gunpowder is stored in a single, potentially attackable &
ignitable location, and enemy invaders (me or you) are on the horizon and
closing in."

But then the fort commander says, "Move out.  Get this gunpowder magazine away
from the front of my tent, it's liable to go up at anytime during a fight and
but we need our central command post intact at all times."

So the sargeant and his troops say, "We can't do much overnight, so let's
accomplish what we can in a short (5-ply, 7-ply, 5-man, 7-man) brigade to move
the gunpowder as swiftly as possible away from the front of the commander's
tent."

So the troops haul the gunpowder around quickly for a few hours (short-sighted,
limited ply work, you see), and end up shuffling the gunpowder to the rear of
the commander's tent.

Eval again says: "My (your) fort is weak, all my (your) gunpowder is stored in a
single, potentially attackable & ignitable location, and enemy invaders (me or
you) are on the horizon and closing in."

If the ply limit isn't enough to permit necessary, long term strategic planning
improvements to one's relative position, the whole defensive (offensive) effort
of one side or the other may go up in smoke!

This applies to both the dynamic (I'm ahead in deployment, I'm attacking!) side,
and the static (I'm ahead, I got a pawn!) side.  Both have to "make progress"
based on their relative strengths and weaknesses, compared to the opponent's
side.

If either side has mere seconds to shuffle (search & move), they may not achieve
the true long term value of their strengths, nor adequately mitigate the true
long term negative value of their weaknesses--due to short-sightedness, even if
the 'eval' is properly weighted for the long haul.

--Steve








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.