I mentioned recently that I put all the Othello quizzes through XGR+ and found some interesting cases, mostly backgames. On the whole, the positions have survived the test of time extremely well. The examples below are the ones where XGR+ raises the most doubts, though in most cases XGR+ still agrees with the top play and just casts doubt on how far behind the second play is. Of course, a full XG rollout might yield a different story altogether. If someone is willing to do full XG2 rollouts of these positions, I'd be curious to see the results.
By the way, some of the positions below are included not because of a bot disagreement but because I found them interesting for a different reason, as you'll see.
The very first Othello quiz position, Position 1 from 2000. XGR+ still chooses bar/23 22/17* but bar/20 22/20 is only 0.026 behind rather than 0.116 behind.
| | White is Player 1
score: 0 pip: 186 | 7 point match | pip: 160 score: 4
Blue is Player 2 | |
XGID=--aaBcD-Bb---Ea--a-c-cA--A:0:0:1:52:4:0:0:7:10 |
Blue to play 52 |
1. | XG Roller+ | Bar/23 22/17* | eq: +0.268 |
| Player: Opponent: | 60.67% (G:21.90% B:1.28%) 39.33% (G:8.22% B:0.29%) | |
|
2. | XG Roller+ | Bar/20 22/20 | eq: +0.242 (-0.026) |
| Player: Opponent: | 60.72% (G:16.58% B:0.64%) 39.28% (G:7.60% B:0.26%) | |
|
eXtreme Gammon Version: 1.21, MET: Rockwell-Kazaross
Position 9 from 2000. Same story, i.e., same choice for the top play, but XGR+ doesn't think the runners-up are all whoppers.
| | White is Player 2
score: 2 pip: 184 | 5 point match | pip: 149 score: 0
Blue is Player 1 | |
XGID=aa-BcAC-C-C-ba---a-bAbbB--:1:-1:1:43:0:2:0:5:10 |
Blue to play 43 |
1. | XG Roller+ | 20/17* 17/13* | eq: +0.628 |
| Player: Opponent: | 61.85% (G:34.13% B:3.50%) 38.15% (G:8.55% B:0.40%) | |
|
2. | XG Roller+ | 20/17* 5/1* | eq: +0.616 (-0.012) |
| Player: Opponent: | 61.40% (G:34.98% B:2.73%) 38.60% (G:9.97% B:0.45%) | |
|
3. | XG Roller+ | 10/6 8/5 | eq: +0.588 (-0.040) |
| Player: Opponent: | 61.42% (G:32.73% B:2.78%) 38.58% (G:9.32% B:0.43%) | |
|
4. | XG Roller+ | 20/16 8/5 | eq: +0.586 (-0.042) |
| Player: Opponent: | 62.52% (G:30.91% B:2.52%) 37.48% (G:8.86% B:0.41%) | |
|
eXtreme Gammon Version: 1.21, MET: Rockwell-Kazaross
Position 7 from 2003. Should have been D/P before the roll, though this doesn't seem to affect the checker play much.
| | White is Player 2
score: 0 pip: 76 | Money session | pip: 131 score: 0
Blue is Player 1 | |
XGID=--aA-BDCB------Ac----Bfcb-:0:0:1:00:0:0:0:0:10 |
Blue on roll, cube action? |
Analyzed in XG Roller+ |
Player Winning Chances: | 78.32% (G: 16.49% B: 0.77%) |
Opponent Winning Chances: | 21.68% (G: 5.02% B: 0.04%) |
|
Cubeless Equities |
No Double: | +0.688 |
Double: | +1.350 |
|
Cubeful Equities |
No Double: | +0.900 (-0.100) |
Double/Take: | +1.215 (+0.215) |
Double/Drop: | +1.000 |
|
Best Cube action: Double / Drop |
|
eXtreme Gammon Version: 1.21
Position 2 from 2005. I'm including this one not because the bots disagree (they don't) but because, surprisingly, none of the top five human participants got it right. I think opening theory has progressed since then, since many players today would see at once that the back checkers need to stay frozen, and two down is almost automatic.
| | White is Player 2
score: 1 pip: 147 | 5 point match | pip: 164 score: 2
Blue is Player 1 | |
XGID=-b---AD-C--AcD---c-e--b-B-:0:0:1:45:2:1:0:5:10 |
Blue to play 45 |
1. | XG Roller+ | 13/9 13/8 | eq: -0.437 |
| Player: Opponent: | 45.34% (G:12.31% B:0.56%) 54.66% (G:20.10% B:2.09%) | |
|
2. | XG Roller+ | 13/9 11/6 | eq: -0.513 (-0.076) |
| Player: Opponent: | 43.62% (G:11.63% B:0.50%) 56.38% (G:19.34% B:1.67%) | |
|
3. | XG Roller+ | 13/4 | eq: -0.536 (-0.099) |
| Player: Opponent: | 43.94% (G:11.24% B:0.46%) 56.06% (G:22.08% B:2.38%) | |
|
4. | XG Roller+ | 13/8 11/7 | eq: -0.556 (-0.119) |
| Player: Opponent: | 42.80% (G:10.61% B:0.44%) 57.20% (G:20.38% B:1.77%) | |
|
5. | XG Roller+ | 24/20 13/8 | eq: -0.649 (-0.212) |
| Player: Opponent: | 43.00% (G:10.65% B:0.53%) 57.00% (G:23.12% B:1.77%) | |
|
eXtreme Gammon Version: 1.21, MET: Rockwell-Kazaross
Position 9 from 2007. The original rollout put 24/22 8/7 on top, 0.070 ahead of 24/22 7/6 and 0.096 ahead of 24/22 13/12.
| | White is Player 2
score: 0 pip: 121 | 5 point match | pip: 220 score: 0
Blue is Player 1 | |
XGID=----aaBBA-B--B--bbbbbbBaD-:1:1:1:21:0:0:0:5:10 |
Blue to play 21 |
1. | XG Roller+ | 24/22 7/6 | eq: -0.440 |
| Player: Opponent: | 44.02% (G:1.71% B:0.05%) 55.98% (G:39.35% B:10.62%) | |
|
2. | XG Roller+ | 24/22 13/12 | eq: -0.441 (-0.001) |
| Player: Opponent: | 43.85% (G:2.04% B:0.07%) 56.15% (G:39.43% B:10.37%) | |
|
3. | XG Roller+ | 24/22 8/7 | eq: -0.444 (-0.005) |
| Player: Opponent: | 42.53% (G:2.29% B:0.07%) 57.47% (G:38.15% B:8.52%) | |
|
4. | XG Roller+ | 10/9 10/8 | eq: -0.486 (-0.046) |
| Player: Opponent: | 42.30% (G:1.93% B:0.06%) 57.70% (G:39.73% B:10.26%) | |
|
5. | XG Roller+ | 24/22 10/9 | eq: -0.490 (-0.051) |
| Player: Opponent: | 43.07% (G:1.38% B:0.05%) 56.93% (G:40.65% B:11.32%) | |
|
6. | XG Roller+ | 13/10 | eq: -0.498 (-0.058) |
| Player: Opponent: | 42.27% (G:1.69% B:0.05%) 57.73% (G:40.36% B:10.25%) | |
|
7. | XG Roller+ | 24/23* 7/5* | eq: -0.519 (-0.080) |
| Player: Opponent: | 42.42% (G:0.80% B:0.02%) 57.58% (G:41.82% B:10.92%) | |
|
8. | XG Roller+ | 8/6 7/6 | eq: -0.523 (-0.083) |
| Player: Opponent: | 41.57% (G:1.45% B:0.04%) 58.43% (G:40.77% B:10.24%) | |
|
9. | XG Roller+ | 24/23* 10/8 | eq: -0.536 (-0.097) |
| Player: Opponent: | 40.09% (G:1.06% B:0.02%) 59.91% (G:39.35% B:8.78%) | |
|
10. | XG Roller+ | 10/8 7/6 | eq: -0.542 (-0.102) |
| Player: Opponent: | 41.72% (G:1.32% B:0.05%) 58.28% (G:41.48% B:11.81%) | |
|
eXtreme Gammon Version: 1.21, MET: Rockwell-Kazaross
Position 10 from 2007. Here XGR+ does actually agree with the Snowie rollout, though I wonder how reliable the bots really are here. In particular I don't know if I totally believe that 4/3* 4/2* is really as far back as the numbers seem to indicate.
| | White is Player 2
score: 0 pip: 224 | 5 point match | pip: 134 score: 0
Blue is Player 1 | |
XGID=-baaBcBBBaBAAAB-cb-b------:0:0:1:21:0:0:0:5:10 |
Blue to play 21 |
1. | XG Roller+ | 11/9* 4/3* | eq: +0.763 |
| Player: Opponent: | 61.73% (G:31.23% B:4.27%) 38.27% (G:5.91% B:0.31%) | |
|
2. | XG Roller+ | 4/3* 4/2* | eq: +0.665 (-0.097) |
| Player: Opponent: | 60.05% (G:29.48% B:3.70%) 39.95% (G:7.49% B:0.42%) | |
|
3. | XG Roller+ | 11/9* 10/9 | eq: +0.607 (-0.156) |
| Player: Opponent: | 56.02% (G:33.01% B:5.26%) 43.98% (G:5.79% B:0.25%) | |
|
4. | XG Roller+ | 12/9* | eq: +0.598 (-0.165) |
| Player: Opponent: | 56.70% (G:31.95% B:4.55%) 43.30% (G:5.58% B:0.25%) | |
|
5. | XG Roller+ | 14/13 11/9* | eq: +0.567 (-0.196) |
| Player: Opponent: | 55.67% (G:32.09% B:4.69%) 44.33% (G:5.70% B:0.26%) | |
|
eXtreme Gammon Version: 1.21, MET: Rockwell-Kazaross
Position 9 from 2008. Same top play, but ahead by only 0.035 rather than 0.126.
| | White is Player 2
score: 1 pip: 189 | 5 point match | pip: 137 score: 2
Blue is Player 1 | |
XGID=--bbDbEB--a--aBa--a--babAA:0:0:1:66:2:1:0:5:10 |
Blue to play 66 |
1. | XG Roller+ | Bar/19 24/18* 7/1(2) | eq: +0.628 |
| Player: Opponent: | 58.67% (G:41.45% B:4.76%) 41.33% (G:5.09% B:0.10%) | |
|
2. | XG Roller+ | Bar/13* 24/18* 13/7 | eq: +0.593 (-0.035) |
| Player: Opponent: | 57.95% (G:38.69% B:4.69%) 42.05% (G:4.41% B:0.06%) | |
|
3. | XG Roller+ | Bar/13* 24/18* 7/1 | eq: +0.574 (-0.054) |
| Player: Opponent: | 58.40% (G:38.25% B:4.38%) 41.60% (G:6.08% B:0.15%) | |
|
4. | XG Roller+ | Bar/13* 24/18* 14/8 | eq: +0.572 (-0.056) |
| Player: Opponent: | 57.77% (G:37.75% B:4.52%) 42.23% (G:4.79% B:0.09%) | |
|
5. | XG Roller+ | Bar/13* 7/1(2) | eq: +0.524 (-0.104) |
| Player: Opponent: | 56.95% (G:38.16% B:3.99%) 43.05% (G:5.75% B:0.14%) | |
|
6. | XG Roller+ | Bar/13* 24/18* 18/12 | eq: +0.519 (-0.109) |
| Player: Opponent: | 56.91% (G:35.85% B:3.99%) 43.09% (G:4.98% B:0.08%) | |
|
eXtreme Gammon Version: 1.21, MET: Rockwell-Kazaross