Arimaa Forum (http://arimaa.com/arimaa/forum/cgi/YaBB.cgi)
Arimaa >> Game Analysis >> Who's ahead purely from material evaluation?
(Message started by: Swynndla on Nov 26th, 2011, 3:00am)

Title: Who's ahead purely from material evaluation?
Post by Swynndla on Nov 26th, 2011, 3:00am
At 39g (before the move is played) in this game:
http://arimaa.com/arimaa/games/jsShowGame.cgi?gid=199480
... who is winning?  Harlog and Fame both say silver is winning, until another silver rabbit is captured on 40g, when they both change their mind and say gold is winning.  I think gold is winning at the beginning of 39g, and not only that, if someone told me what pieces were on the board without showing me the position (so I don't know where the pieces are, just like the material evaluators), I'd say that on average gold is ahead in material evaluation score.  So I agree it's a close thing, and so biases will tend to show through, but I disagree with both Harlog and Fame, and I'm wanting to know what others think.

At 39g:
DAPE thinks silver is ahead by 2.00.
FAME thinks silver is ahead by 0.61.
HarLog thinks silver is ahead by 1.98.

At 40s (when another silver rabbit has been captured):
DAPE thinks gold is ahead by 2.83.
FAME thinks gold is ahead by 1.36.
HarLog thinks gold is ahead by 0.07.

Title: Re: Who's ahead purely from material evaluation?
Post by chessandgo on Nov 26th, 2011, 7:45am
pfoooo ... that's a tough one. I suppose if the game was played with both armies on their backrows, I'd take Gold, but I have no idea whether this is correct. In practice, I guess it should heavily depend on the position.

Title: Re: Who's ahead purely from material evaluation?
Post by Swynndla on Nov 26th, 2011, 3:16pm
lol I was thinking that if there was only going to be one person who agreed with me, it'd be you c&go!  :D  (if only because you value rabbits so highly. ;))  I was hoping you'd respond to this post - cheers c&go. :)

Title: Re: Who's ahead purely from material evaluation?
Post by Fritzlein on Nov 27th, 2011, 9:11am
This is difficult because my intuitions about material balance have been built mostly by openings and middlegames, not by endgames like this.  My endgame is weak, not only compared to the other phases of my game, but compared to Jean's endgame.  Nevertheless, I will venture to take the opposite opinion from Jean: apart from the position (i.e. if we set up the material on the home rows), I would rather play Silver.

The ideal piece alignment for Silver is obvious: make a beeline for the gold horse with the silver elephant.  Gold can't afford to lose the horse, or suddenly Silver has four(!) deputies.  Gold also can't afford to have his elephant deadlocked defending the gold horse or Silver has four(!) strongest free pieces.  So the gold horse simply has to hide with the help of rabbits, so as to leave the gold elephant free.

What can an elephant and rabbits do with no supporting pieces?  On such a thinned out board, maybe a lot.  But I could also see the battle working out badly for gold.  Silver has two little pieces on each side, enough to hold back an ERRR attack and maybe even get a gold rabbit framed.  Gold has no pieces to break frames.  And don't forget that, as thin as Silver's goal defense is, Gold's goal defense is thinner.  A single advanced silver rabbit could be a nightmare for Gold, requiring four or five gold rabbits to stay home to block it.

I submit that your evaluation may be colored by the actual progress of the game, in which the silver elephant never got close to the gold horse.  Instead the silver elephant was doing a cat's job, trying to help a rabbit goal against rabbit defense, or trying to capture a rabbit.  Let the elephant do an elephant's job and a cat do a cat's job, and things might soon look very promising for Silver.

Title: Re: Who's ahead purely from material evaluation?
Post by pago on Dec 2nd, 2011, 5:10am

on 11/26/11 at 03:00:21, Swynndla wrote:
At 39g (before the move is played) in this game:
http://arimaa.com/arimaa/games/jsShowGame.cgi?gid=199480
Harlog and Fame both say silver is winning, until another silver rabbit is captured on 40g, when they both change their mind and say gold is winning.  


I have used this situation as a test for the evaluators I proposed in other threads. For memory the scale goes from 0% to 100%.

HERD @ 39 g
HERD(EH8R/eddcc4r) = 46,45% => silver advantage

HERD @ 40g
REM(EH8R/eddcc3r) = 50,22% => slight gold advantage (almost balanced)


For REM the differences are larger

REM @ 39 g
REM(EH8R/eddcc4r) = 45,85% => silver advantage

REM @ 39 g
REM(EH8R/eddcc3r) = 51,88% => Gold advantage

Title: Re: Who's ahead purely from material evaluation?
Post by Nombril on Dec 6th, 2011, 10:02pm
My gut reaction is that silver is winning.  Silver has a NRP in all four quadrants... Seems the gold elephant can be easily overloaded.  If either the gold elephant or horse tries to attack, there is an entire wing undefended...and the silver elephant can help defend...

Of course, silver doesn't have time to just shuffle the enemy rabbits around...



Arimaa Forum » Powered by YaBB 1 Gold - SP 1.3.1!
YaBB © 2000-2003. All Rights Reserved.