Discussion:
Noncontact bearin play
(too old to reply)
Tim Chow
2016-09-09 22:54:32 UTC
Permalink
XGID=-BDBABB--AA--a--a--abcbcb-:0:0:1:42:0:0:0:0:10

X:Player 1 O:Player 2
Score is X:0 O:0. Unlimited Game
+13-14-15-16-17-18------19-20-21-22-23-24-+
| O O | | O O O O O O |
| | | O O O O O |
| | | O O |
| | | |
| | | |
| |BAR| |
| | | |
| | | X |
| | | X |
| | | X X X X X |
| X X | | X X X X X X |
+12-11-10--9--8--7-------6--5--4--3--2--1-+
Pip count X: 61 O: 63 X-O: 0-0
Cube: 1
X to play 42

---
Tim Chow
Bradley K. Sherman
2016-09-10 00:00:26 UTC
Permalink
Post by Tim Chow
Score is X:0 O:0. Unlimited Game
+13-14-15-16-17-18------19-20-21-22-23-24-+
| O O | | O O O O O O |
| | | O O O O O |
| | | O O |
| | | |
| | | |
| |BAR| |
| | | |
| | | X |
| | | X |
| | | X X X X X |
| X X | | X X X X X X |
+12-11-10--9--8--7-------6--5--4--3--2--1-+
X:61 O:63, Cube:1, X to play 42
10/6,9/7

(Are these positions played perfectly by the bots?)

--bks
Paul
2016-09-10 08:08:04 UTC
Permalink
Post by Bradley K. Sherman
Post by Tim Chow
Score is X:0 O:0. Unlimited Game
+13-14-15-16-17-18------19-20-21-22-23-24-+
| O O | | O O O O O O |
| | | O O O O O |
| | | O O |
| | | |
| | | |
| |BAR| |
| | | |
| | | X |
| | | X |
| | | X X X X X |
| X X | | X X X X X X |
+12-11-10--9--8--7-------6--5--4--3--2--1-+
X:61 O:63, Cube:1, X to play 42
10/6,9/7
(Are these positions played perfectly by the bots?)
Tim is trying to explain the importance of filling in/preventing gaps, and also the importance of not wasting pips on a crossover. Hence 10/4 is correct, and I'm sure you know this. Are you saying 10/6 9/7 because you feel honour-bound to always answer by what you would play OTB? It's hard to know because you don't give justification for your plays. Tim has given several problems about the dangers of playing to the 7 point and being forced to waste pips by bearing in deep.

Here, the equity-loss with your play is considerable because it has two big problems, playing to the 7 and having a potential gap on the 4.

A bot could play this position perfectly, I would think, because we are very few plies away from a Sconyers-database position. Sconyers-database positions can be played perfectly but the databases use too much memory for a typical home PC.

Other than via a database, it's implausible that we could get perfection. In particular, if one play is considered 0.001 better than another, there's around a 50% likelihood that the play considered weaker is actually better.

However, this is a very easy problem so I wouldn't worry about bot strength here. It's a cautionary tale about good bearoff technique rather than a challenging problem where good players are likely to have different opinions. It would be easy to lapse OTB.

Paul
Tim Chow
2016-09-10 20:16:08 UTC
Permalink
Post by Bradley K. Sherman
(Are these positions played perfectly by the bots?)
There do exist perfect-play databases for noncontact positions, but as far as
I know, GNU and XG only use databases for positions where all the checkers are
in the home boards.

But even without a perfect-play database, bot rollouts are reliable when there
are only a few checkers left to bear in. (The bots do have a few bizarre
weak spots when it comes to bearin plays with lots of checkers still far from
home.)

---
Tim Chow
BlueDice
2016-09-10 13:11:26 UTC
Permalink
Post by Tim Chow
XGID=-BDBABB--AA--a--a--abcbcb-:0:0:1:42:0:0:0:0:10
X:Player 1 O:Player 2
Score is X:0 O:0. Unlimited Game
+13-14-15-16-17-18------19-20-21-22-23-24-+
| O O | | O O O O O O |
| | | O O O O O |
| | | O O |
| | | |
| | | |
| |BAR| |
| | | |
| | | X |
| | | X |
| | | X X X X X |
| X X | | X X X X X X |
+12-11-10--9--8--7-------6--5--4--3--2--1-+
Pip count X: 61 O: 63 X-O: 0-0
Cube: 1
X to play 42
---
Tim Chow
10/4 seems the "obvious" choice. However, the cynic in me is suspicious of the QF and wonders if it is "too" obvious
--
BD
Paul
2016-09-11 07:43:36 UTC
Permalink
Post by BlueDice
Post by Tim Chow
XGID=-BDBABB--AA--a--a--abcbcb-:0:0:1:42:0:0:0:0:10
X:Player 1 O:Player 2
Score is X:0 O:0. Unlimited Game
+13-14-15-16-17-18------19-20-21-22-23-24-+
| O O | | O O O O O O |
| | | O O O O O |
| | | O O |
| | | |
| | | |
| |BAR| |
| | | |
| | | X |
| | | X |
| | | X X X X X |
| X X | | X X X X X X |
+12-11-10--9--8--7-------6--5--4--3--2--1-+
Pip count X: 61 O: 63 X-O: 0-0
Cube: 1
X to play 42
---
Tim Chow
10/4 seems the "obvious" choice. However, the cynic in me is suspicious of the QF and wonders if it is "too" obvious
The convention here seems to be that discussions of every position are in the form of a quiz. IMO, it would be better to say "In positions like this, be careful about the distribution and don't rush the play." A good warning because it's easy to get careless OTB particularly in a chouette.

Asking it in the form of a quiz renders it completely trivial, but (besides my own posts), that's the only way positions get discussed. I think the non-obvious thing will turn out to be how much equity is at stake, which will be far more than many will guess.

Paul
m***@gmail.com
2016-09-10 13:30:37 UTC
Permalink
Post by Tim Chow
XGID=-BDBABB--AA--a--a--abcbcb-:0:0:1:42:0:0:0:0:10
X:Player 1 O:Player 2
Score is X:0 O:0. Unlimited Game
+13-14-15-16-17-18------19-20-21-22-23-24-+
| O O | | O O O O O O |
| | | O O O O O |
| | | O O |
| | | |
| | | |
| |BAR| |
| | | |
| | | X |
| | | X |
| | | X X X X X |
| X X | | X X X X X X |
+12-11-10--9--8--7-------6--5--4--3--2--1-+
Pip count X: 61 O: 63 X-O: 0-0
Cube: 1
X to play 42
---
Tim Chow
I would have played 10/4 for better distribution. It's just one pip 'penalty" fee...
m***@gmail.com
2016-09-10 13:47:27 UTC
Permalink
Bradley can be excused (for not providing an explanation) for 3 reasons
a)he's been participating in rec.games for a very long time, as far as I can see from searching old topics.
b)He always replies first. ;-)
c)He participates in ALL quizes.

@Bradley: It's common sense and common knowledge to aim at good distribution.. But if you want to examine it in detail:
The bear off will last for about 6 rolls before clearing from the 6 to the 4pt. On the average each number will be rolled twice. With your play -on the average- you will waste one 4.
Paul
2016-09-10 17:07:23 UTC
Permalink
Post by m***@gmail.com
Post by Tim Chow
XGID=-BDBABB--AA--a--a--abcbcb-:0:0:1:42:0:0:0:0:10
X:Player 1 O:Player 2
Score is X:0 O:0. Unlimited Game
+13-14-15-16-17-18------19-20-21-22-23-24-+
| O O | | O O O O O O |
| | | O O O O O |
| | | O O |
| | | |
| | | |
| |BAR| |
| | | |
| | | X |
| | | X |
| | | X X X X X |
| X X | | X X X X X X |
+12-11-10--9--8--7-------6--5--4--3--2--1-+
Pip count X: 61 O: 63 X-O: 0-0
Cube: 1
X to play 42
---
Tim Chow
I would have played 10/4 for better distribution. It's just one pip 'penalty" fee...
Why is it one pip penalty fee? I don't understand that statement (even though I strongly agree with your play).

Thanks,

Paul
m***@gmail.com
2016-09-10 20:55:40 UTC
Permalink
Post by Paul
Why is it one pip penalty fee?
I called it one pip "penalty" fee because we do pay something. With 10/8 9/5 anything >=2 guarantees the bearing off of one checker. With the other play we need >=3. The difference would show in rolls like 12, 11,and 22. There are other cons to the move too. e.g with Bradley's move 55s and 66s are more efficient while only 44s are less efficient
m***@gmail.com
2016-09-10 21:08:49 UTC
Permalink
Post by m***@gmail.com
Post by Paul
Why is it one pip penalty fee?
I called it one pip "penalty" fee because we do pay something. With 10/8 9/5 anything >=2 guarantees the bearing off of one checker. With the other play we need >=3. The difference would show in rolls like 12, 11,and 22. There are other cons to the move too. e.g with Bradley's move 55s and 66s are more efficient while only 44s are less efficient
Woops sorry I just noticed Bradley's move was not 10/8 9/5.
Tim Chow
2016-09-11 17:28:15 UTC
Permalink
XGID=-BDBABB--AA--a--a--abcbcb-:0:0:1:42:0:0:0:0:10

X:Player 1 O:Player 2
Score is X:0 O:0. Unlimited Game
+13-14-15-16-17-18------19-20-21-22-23-24-+
| O O | | O O O O O O |
| | | O O O O O |
| | | O O |
| | | |
| | | |
| |BAR| |
| | | |
| | | X |
| | | X |
| | | X X X X X |
| X X | | X X X X X X |
+12-11-10--9--8--7-------6--5--4--3--2--1-+
Pip count X: 61 O: 63 X-O: 0-0
Cube: 1
X to play 42

I don't think I have any thing to add to the excellent comments already posted.

1. Rollout¹ 10/4 eq:+0.317
Player: 59.91% (G:0.00% B:0.00%)
Opponent: 40.09% (G:0.00% B:0.00%)
Confidence: ±0.004 (+0.313..+0.321) - [100.0%]

2. Rollout¹ 9/5 6/4 eq:+0.272 (-0.045)
Player: 58.68% (G:0.00% B:0.00%)
Opponent: 41.32% (G:0.00% B:0.00%)
Confidence: ±0.004 (+0.268..+0.276) - [0.0%]

3. Rollout¹ 10/8 9/5 eq:+0.264 (-0.053)
Player: 58.61% (G:0.00% B:0.00%)
Opponent: 41.39% (G:0.00% B:0.00%)
Confidence: ±0.004 (+0.260..+0.268) - [0.0%]

4. Rollout¹ 10/6 9/7 eq:+0.231 (-0.086)
Player: 57.46% (G:0.00% B:0.00%)
Opponent: 42.54% (G:0.00% B:0.00%)
Confidence: ±0.004 (+0.227..+0.235) - [0.0%]

¹ 1296 Games rolled with Variance Reduction.
Dice Seed: 271828
Moves: 3-ply, cube decisions: XG Roller

eXtreme Gammon Version: 2.19.208.pre-release

----
Tim Chow

Loading...