I reread my comments in the previous post. I'm concerned they might have come across as a criticism of Chris, for whom I have the utmost respect. Rather, I believe I was venting some pent up irritation over the years at seeing so many online shoot-from-the-hip analyses based on already knowing the "answer," and I can only imagine some of the virtues that might be getting circulated about the 6/3 5/1 play in the position below.
With that out of the way...
          
 
         
 | | White is Player 2
score: 0
pip: 109 | Unlimited Game
Jacoby Beaver | pip: 96
score: 0
Blue is Player 1 |
|
| XGID=--ACBBBbB----C-a-aadbbb---:0:0:1:43:0:0:3:0:10 |
| Blue to play 43 |
| 1. | Rollout1 | 6/3 5/1 | eq: +0.536 |
| Player:
Opponent: | 67.00% (G:1.51% B:0.02%)
33.00% (G:2.23% B:0.09%) | Conf.: � 0.009 (+0.527...+0.545) - [100.0%]
Duration: 18 minutes 37 seconds |
|
| 2. | Rollout1 | 5/1 4/1 | eq: +0.474 (-0.061) |
| Player:
Opponent: | 65.76% (G:0.77% B:0.02%)
34.24% (G:1.41% B:0.04%) | Conf.: � 0.007 (+0.467...+0.481) - [0.0%]
Duration: 14 minutes 38 seconds |
|
| 3. | Rollout1 | 5/2 5/1 | eq: +0.472 (-0.064) |
| Player:
Opponent: | 65.75% (G:0.75% B:0.02%)
34.25% (G:1.41% B:0.04%) | Conf.: � 0.007 (+0.465...+0.479) - [0.0%]
Duration: 40 minutes 18 seconds |
|
| 4. | Rollout1 | 6/2 4/1 | eq: +0.467 (-0.069) |
| Player:
Opponent: | 65.54% (G:0.84% B:0.02%)
34.46% (G:1.58% B:0.05%) | Conf.: � 0.007 (+0.460...+0.474) - [0.0%]
Duration: 12 minutes 23 seconds |
|
| 5. | Rollout1 | 8/5 8/4 | eq: +0.466 (-0.070) |
| Player:
Opponent: | 64.97% (G:0.87% B:0.02%)
35.03% (G:1.00% B:0.03%) | Conf.: � 0.007 (+0.459...+0.473) - [0.0%]
Duration: 11 minutes 58 seconds |
|
| 6. | Rollout1 | 13/6 | eq: +0.463 (-0.073) |
| Player:
Opponent: | 64.74% (G:1.23% B:0.07%)
35.26% (G:1.36% B:0.03%) | Conf.: � 0.006 (+0.457...+0.469) - [0.0%]
Duration: 13 minutes 23 seconds |
|
| 7. | Rollout1 | 6/2 5/2 | eq: +0.458 (-0.078) |
| Player:
Opponent: | 65.25% (G:0.77% B:0.02%)
34.75% (G:1.61% B:0.05%) | Conf.: � 0.007 (+0.451...+0.465) - [0.0%]
Duration: 23 minutes 07 seconds |
|
| 8. | Rollout1 | 6/3 6/2 | eq: +0.439 (-0.097) |
| Player:
Opponent: | 63.92% (G:0.81% B:0.01%)
36.08% (G:1.67% B:0.04%) | Conf.: � 0.007 (+0.432...+0.446) - [0.0%]
Duration: 20 minutes 09 seconds |
|
| |
1 1296 Games rolled with Variance Reduction.
Dice Seed: 5
Moves: 3-ply, cube decisions: XG Roller
|
Above is the rollout Chris posted, the lopsided result of which I was and am convinced is due to an XG bug (and apparently a Gnubg bug as well).
I experimented with one-pip offsetting moves on White's side of the board, checking XGR++ evaluations, but Black's 6/1 4/3 continued to crush the competition.
I hit paydirt when I broke White's 5pt (with 6/5, and offset it with 8/7). This time, we see a much more believable result: Seven plays roll out very close to each other, as shown below.
| | White is Player 2
score: 0
pip: 109 | Unlimited Game
Jacoby Beaver | pip: 96
score: 0
Blue is Player 1 |
|
| XGID=--ACBBBbB----C-a--beabb---:0:0:1:43:0:0:3:0:10 |
| Blue to play 43 |
| 1. | Rollout1 | 13/6 | eq: +0.491 |
| Player:
Opponent: | 65.76% (G:1.72% B:0.10%)
34.24% (G:1.24% B:0.03%) | Conf.: ± 0.006 (+0.485...+0.497) - [38.8%]
Duration: 3 minutes 06 seconds |
|
| 2. | Rollout1 | 5/2 5/1 | eq: +0.490 (-0.001) |
| Player:
Opponent: | 66.17% (G:0.94% B:0.04%)
33.83% (G:1.36% B:0.03%) | Conf.: ± 0.007 (+0.483...+0.497) - [32.4%]
Duration: 2 minutes 28 seconds |
|
| 3. | Rollout1 | 6/2 4/1 | eq: +0.488 (-0.003) |
| Player:
Opponent: | 66.15% (G:0.95% B:0.02%)
33.85% (G:1.44% B:0.05%) | Conf.: ± 0.007 (+0.482...+0.495) - [14.2%]
Duration: 2 minutes 40 seconds |
|
| 4. | Rollout1 | 5/1 4/1 | eq: +0.488 (-0.003) |
| Player:
Opponent: | 66.15% (G:0.94% B:0.03%)
33.85% (G:1.31% B:0.03%) | Conf.: ± 0.007 (+0.481...+0.495) - [14.4%]
Duration: 2 minutes 46 seconds |
|
| 5. | Rollout1 | 8/5 8/4 | eq: +0.480 (-0.011) |
| Player:
Opponent: | 65.65% (G:1.00% B:0.02%)
34.35% (G:0.87% B:0.02%) | Conf.: ± 0.007 (+0.473...+0.487) - [0.1%]
Duration: 2 minutes 52 seconds |
|
| 6. | Rollout1 | 6/2 5/2 | eq: +0.479 (-0.012) |
| Player:
Opponent: | 65.75% (G:0.86% B:0.02%)
34.25% (G:1.53% B:0.05%) | Conf.: ± 0.007 (+0.472...+0.486) - [0.0%]
Duration: 2 minutes 14 seconds |
|
| 7. | Rollout1 | 6/3 5/1 | eq: +0.478 (-0.013) |
| Player:
Opponent: | 65.78% (G:0.96% B:0.02%)
34.22% (G:1.67% B:0.04%) | Conf.: ± 0.008 (+0.471...+0.486) - [0.0%]
Duration: 2 minutes 56 seconds |
|
| |
1 1296 Games rolled with Variance Reduction.
Dice Seed: 57309652
Moves: 3-ply, cube decisions: XG Roller
|
The comparison of these two rollouts seems like powerful evidence that one of two positions is generating a corrupt result, and I know which one I think it is. :)
White is most likely going to cover her 5pt before Blue's next roll, but even if she never does, that aspect cannot begin to explain the large swing in margins. But the checker-placement difference between the two positions might help point to a feature that will lead to further clues.
Nack