Position # 37

[PersianLord]

Position information:

Pip-count: Red 138-141 White
Score:       Red     6-4     White
Match:                   9-pointer

Red on roll. Proper cub action?

Spoiler

I watched this match along with some friends. I was so sure that this must be a drop that I didn't feel obliged to run the analysis, but when white took, re-doubled, gammoned and won the match, I decided to analysis the position. To my surprise, GNUBG's 2-ply agrees that it's a take:

Cube analysis

2-ply cubeless equity  +0.693 (Money:  +0.635)

   71.8%  25.5%   1.2% -  28.2%   6.7%   0.1%

Cubeful equities:

1. Double, take         +0.853
2. Double, pass         +1.000  ( +0.147)
3. No double            +0.800  ( -0.053)

Proper cube action: Double, take

But Snowie at 1-ply (eqaul to GNUBG 0-ply) thinks that it's a terrible mistake to accept the cube:

White Double

11. Green Take

Cube action equity Alert: Wrong take
1-Ply Money equity: 0.718
   1.3%  27.6%  74.6%    25.4%   6.0%   0.2%

1. Double, pass 1.000 
2. No double 0.909 (-0.091)
3. Double, take 1.209 (+0.209)
Proper cube action:   Double, pass

Thus, more analysis is needed.

[close]

[playBunny]

Spoiler

My simplistic approach is this. If there were only one red blot back then it would probably get home with ease given all the space between white's chimney stacks but with two blots back that's enough for one at least to get into all sorts of trouble. Add to that the open red home table and the chance of making a decent anchor and you get enough hope for white to take. I don't think the danger to white's blot on the 2-point matters that much.

There's also the question of the score but I'm not the expert on that side on things so I'll just give this (somewhat temporary) link to the g11 MET and maybe someone else can say how the numbers figure into it.

http://www.gellie.plus.com/Misc/MET-g11.htm?matchlen=9&player=6&opp=4&cube=1

Playing around in GnuBg...
If red has only the piece on the bar and the backrunner other is on 18, ie. still in danger, then it's a take.
If red has only the piece on the bar and the other is on 14, ie. the least possible danger, then it's a big drop.
If red has only the piece on the bar and the other is on 12 then it's a big drop.

If the white blot on the 2-point is safe on 12 then it's a doubtful double.
If white is safe on 12 and red has only the piece on the bar and the other is on 15 then it's a take.
If red has only the piece on the bar and the other is on 14, ie. the least possible danger, then it's a doubtful take.
If white is safe on 12 and red has only the piece on the bar and the other is on 12 then it's a clear drop.

So it is the fact that there are two men back and the white home blot matters more than I thought.

[close]

[PersianLord]

Spoiler

My simplistic approach is this. If there were only one red blot back then it would probably get home with ease given all the space between white's chimney stacks but with two blots back that's enough for one at least to get into all sorts of trouble. Add to that the open red home table and the chance of making a decent anchor and you get enough hope for white to take. I don't think the danger to white's blot on the 2-point matters that much.

There's also the question of the score but I'm not the expert on that side on things so I'll just give this (somewhat temporary) link to the g11 MET and maybe someone else can say how the numbers figure into it.

http://www.gellie.plus.com/Misc/MET-g11.htm?matchlen=9&player=6&opp=4&cube=1

Playing around in GnuBg...
If red has only the piece on the bar and the backrunner other is on 18, ie. still in danger, then it's a take.
If red has only the piece on the bar and the other is on 14, ie. the least possible danger, then it's a big drop.
If red has only the piece on the bar and the other is on 12 then it's a big drop.

If the white blot on the 2-point is safe on 12 then it's a doubtful double.
If white is safe on 12 and red has only the piece on the bar and the other is on 15 then it's a take.
If red has only the piece on the bar and the other is on 14, ie. the least possible danger, then it's a doubtful take.
If white is safe on 12 and red has only the piece on the bar and the other is on 12 then it's a clear drop.

So it is the fact that there are two men back and the white home blot matters more than I thought.

[close]

Spoiler

thnx for the analysis. here's snowie 3-ply analysis btw:

Cube action equity Alert: Wrong take
3-Ply Money equity: 0.694
   1.3%  26.4%  74.0%    26.0%   5.9%   0.3%

1. Double, pass 1.000 
2. No double 0.929 (-0.071)
3. Double, take 1.159 (+0.159)

Proper cube action:   Double, pass

i'm convinced now that gnubg is much more aggressive then snowie, both in checker play and cube decisions. which is better, i don't know.

[close]

[blitzxz]

Spoiler

This looks like a another weird take by gnu. This time the rollout shows too that pass is correct but the margin is very small. The score does the trick again and also white will get very effective redouble if he turns the game around. But still take looks for me desperate. And in real game white moves are also a lot harder then red's.

Here's gnu 2-ply rollout with narrow filter.

Cube analysis
Rollout cubeless equity  +0,688 (Money:  +0,687)

Cubeful equities:
1. Double, pass         +1,000
2. Double, take         +1,035  ( +0,035)
3. No double            +0,901  ( -0,099)
Proper cube action: Double, pass
Rollout details:
Centered 1-cube:
  0,735 0,252 0,024 - 0,265 0,058 0,002 CL  +0,688 CF  +0,901
[0,002 0,002 0,002 - 0,002 0,001 0,000 CL   0,005 CF   0,011]
Player gnubg owns 2-cube:
  0,745 0,268 0,050 - 0,255 0,054 0,007 CL  +1,435 CF  +1,035
[0,002 0,005 0,003 - 0,002 0,001 0,001 CL   0,011 CF   0,016]
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 824138749 and quasi-random dice
Play:  2-ply cubeful prune [world class]
keep the first 0 0-ply moves and up to 8 more moves within equity 0,12
Skip pruning for 1-ply moves.
Cube: 2-ply cubeful prune [world class]

[close]