XGID=a-Bb-BC-BB---B-----bbbAbdA:1:1:1:00:0:0:0:0:10

Score is X:0 O:0. Unlimited Game
+13-14-15-16-17-18------19-20-21-22-23-24-+
| X | | O O O X O O |
| X | | O O O O O |
| | | O |
| | | O |
| | X | |
| |BAR| |
| | O | |
| | | |
| | | X | +---+
| X X | | X X O X | | 2 |
| X X | | X X O X | +---+
+12-11-10--9--8--7-------6--5--4--3--2--1-+
Pip count X: 139 O: 107 X-O: 0-0
Cube: 2, X own cube
X on roll, cube action

---
Tim Chow

On Saturday, January 6, 2024 at 10:18:57 PM UTC, Timothy Chow wrote:
> XGID=a-Bb-BC-BB---B-----bbbAbdA:1:1:1:00:0:0:0:0:10
>
> Score is X:0 O:0. Unlimited Game
> +13-14-15-16-17-18------19-20-21-22-23-24-+
> | X | | O O O X O O |
> | X | | O O O O O |
> | | | O |
> | | | O |
> | | X | |
> | |BAR| |
> | | O | |
> | | | |
> | | | X | +---+
> | X X | | X X O X | | 2 |
> | X X | | X X O X | +---+
> +12-11-10--9--8--7-------6--5--4--3--2--1-+
> Pip count X: 139 O: 107 X-O: 0-0
> Cube: 2, X own cube
> X on roll, cube action
>
> ---
> Tim Chow

Easy take. Only the double is the question.
I'd also like to take a guess at the context:
Tim doubled based on the dual threats of entering while the opponent's on the
bar and the opponent rolling an anti-joker of entering and cracking the inner board.
Tim's double was then badly dinged by XG and he's curious as to what others think.

I think it's a hold. I'd put doubling from the bar into "When in doubt don't" territory.
The roller might be more gammon vulnerable than the opponent.
Also I don't think the opponent's anti-joking cracking numbers are all that fatal.
Nothing wrong with holding until that happens.

Paul

XGID=a-Bb-BC-BB---B-----bbbAbdA:1:1:1:00:0:0:0:0:10

Score is X:0 O:0. Unlimited Game
+13-14-15-16-17-18------19-20-21-22-23-24-+
| X | | O O O X O O |
| X | | O O O O O |
| | | O |
| | | O |
| | X | |
| |BAR| |
| | O | |
| | | |
| | | X | +---+
| X X | | X X O X | | 2 |
| X X | | X X O X | +---+
+12-11-10--9--8--7-------6--5--4--3--2--1-+
Pip count X: 139 O: 107 X-O: 0-0
Cube: 2, X own cube
X on roll, cube action

Paul's "guess at the context" unfortunately is wrong. I did not
double here, but XG thinks it's a big double, and depending on
what one means by an "easy take," the take is maybe not so easy.
But Paul correctly spotted something that I overlooked OTB, which
is that even though O has a five-point board, it is brittle, and
even if X dances, O could enter and crack immediately.

Analyzed in Rollout
No redouble
Player Winning Chances: 70.97% (G:21.87% B:0.95%)
Opponent Winning Chances: 29.03% (G:10.57% B:0.09%)
Redouble/Take
Player Winning Chances: 71.41% (G:22.41% B:1.01%)
Opponent Winning Chances: 28.59% (G:10.52% B:0.09%)

Cubeful Equities:
No redouble: +0.839 (-0.088)
Redouble/Take: +0.927
Redouble/Pass: +1.000 (+0.073)

Best Cube action: Redouble / Take

Rollout:
1296 Games rolled with Variance Reduction.
Dice Seed: 271828
Moves: 3-ply, cube decisions: XG Roller
Confidence No Double: ± 0.013 (+0.826..+0.852)
Confidence Double: ± 0.019 (+0.908..+0.946)

eXtreme Gammon Version: 2.19.211.pre-release

---
Tim Chow

On Tuesday, January 9, 2024 at 1:55:48 PM UTC, Timothy Chow wrote:
> XGID=a-Bb-BC-BB---B-----bbbAbdA:1:1:1:00:0:0:0:0:10
>
> Score is X:0 O:0. Unlimited Game
> +13-14-15-16-17-18------19-20-21-22-23-24-+
> | X | | O O O X O O |
> | X | | O O O O O |
> | | | O |
> | | | O |
> | | X | |
> | |BAR| |
> | | O | |
> | | | |
> | | | X | +---+
> | X X | | X X O X | | 2 |
> | X X | | X X O X | +---+
> +12-11-10--9--8--7-------6--5--4--3--2--1-+
> Pip count X: 139 O: 107 X-O: 0-0
> Cube: 2, X own cube
> X on roll, cube action
> Paul's "guess at the context" unfortunately is wrong. I did not
> double here, but XG thinks it's a big double, and depending on
> what one means by an "easy take," the take is maybe not so easy.
> But Paul correctly spotted something that I overlooked OTB, which
> is that even though O has a five-point board, it is brittle, and
> even if X dances, O could enter and crack immediately.
>
> Analyzed in Rollout
> No redouble
> Player Winning Chances: 70.97% (G:21.87% B:0.95%)
> Opponent Winning Chances: 29.03% (G:10.57% B:0.09%)
> Redouble/Take
> Player Winning Chances: 71.41% (G:22.41% B:1.01%)
> Opponent Winning Chances: 28.59% (G:10.52% B:0.09%)
>
> Cubeful Equities:
> No redouble: +0.839 (-0.088)
> Redouble/Take: +0.927
> Redouble/Pass: +1.000 (+0.073)
>
> Best Cube action: Redouble / Take
>
> Rollout:
> 1296 Games rolled with Variance Reduction.
> Dice Seed: 271828
> Moves: 3-ply, cube decisions: XG Roller
> Confidence No Double: ± 0.013 (+0.826..+0.852)
> Confidence Double: ± 0.019 (+0.908..+0.946)
>
> eXtreme Gammon Version: 2.19.211.pre-release

It's interesting (to me) to clarify what "easy" means in the
phrase "easy take", although I had never thought of this before.

I think it means that "I would take and I would find it an easy decision."
Of course, confident or "easy" decisions can well be wrong.
I think my usage is a bit idiosyncratic and that the bg community
generally considers "big take", "easy take" and "clear take" to be
synonymous. So why use so many distinct phrases then?
Because bg writing would be more tedious otherwise.
People often need variety in their communication.
Poker players sometimes call eights "snowmen" for the same reason.

Why is my usage not the standard? Because I might have studied a
position beforehand and know that it's an extremely marginal take.
It's "easy" because I know it's a take but it is marginal. But others don't
call such takes easy.
Maybe they're not "easy" though because taking a highly marginal take
for money might cause psychological stress and players can even drop
these because of "money management".

Thanks for posting this. It's not often that someone on the bar
bar and needing to enter against a 5 point board almost has a 0.93 cube!
Holding and dropping seem to be about equal errors.

Paul

On Tuesday, January 9, 2024 at 1:55:48 PM UTC, Timothy Chow wrote:
> XGID=a-Bb-BC-BB---B-----bbbAbdA:1:1:1:00:0:0:0:0:10
>
> Score is X:0 O:0. Unlimited Game
> +13-14-15-16-17-18------19-20-21-22-23-24-+
> | X | | O O O X O O |
> | X | | O O O O O |
> | | | O |
> | | | O |
> | | X | |
> | |BAR| |
> | | O | |
> | | | |
> | | | X | +---+
> | X X | | X X O X | | 2 |
> | X X | | X X O X | +---+
> +12-11-10--9--8--7-------6--5--4--3--2--1-+
> Pip count X: 139 O: 107 X-O: 0-0
> Cube: 2, X own cube
> X on roll, cube action
> Paul's "guess at the context" unfortunately is wrong. I did not
> double here, but XG thinks it's a big double, and depending on
> what one means by an "easy take," the take is maybe not so easy.
> But Paul correctly spotted something that I overlooked OTB, which
> is that even though O has a five-point board, it is brittle, and
> even if X dances, O could enter and crack immediately.
>
> Analyzed in Rollout
> No redouble
> Player Winning Chances: 70.97% (G:21.87% B:0.95%)
> Opponent Winning Chances: 29.03% (G:10.57% B:0.09%)
> Redouble/Take
> Player Winning Chances: 71.41% (G:22.41% B:1.01%)
> Opponent Winning Chances: 28.59% (G:10.52% B:0.09%)
>
> Cubeful Equities:
> No redouble: +0.839 (-0.088)
> Redouble/Take: +0.927
> Redouble/Pass: +1.000 (+0.073)
>
> Best Cube action: Redouble / Take
>
> Rollout:
> 1296 Games rolled with Variance Reduction.
> Dice Seed: 271828
> Moves: 3-ply, cube decisions: XG Roller
> Confidence No Double: ± 0.013 (+0.826..+0.852)
> Confidence Double: ± 0.019 (+0.908..+0.946)

This position is a good example of why experts are better
judges of human play than a bot doing PR computations alone.

From a human standpoint, losing 0.088 equity on a cube decision
is consistent with being quite a strong player. I don't know the PR
formulas exactly but I think it's possible to play at a 5 PR rating
averaging one 0.09 error per game if other errors are minimized.

However, a world-class human, who is likely to be a StickLer for
accuracy and is likely to Stick to their views, would probably
deem someone who held that cube to be a weak player because
it shows quite a poor understanding to think that a 0.927 cube is
a hold.
The point is that, in cube play, huge errors in understanding can
go unpunished. For example, a player who isn't good enough to
double may be holding under the illusion of being too good to double.

Paul

On 1/9/2024 9:28 AM, peps...@gmail.com wrote:
> From a human standpoint, losing 0.088 equity on a cube decision
> is consistent with being quite a strong player. I don't know the PR
> formulas exactly but I think it's possible to play at a 5 PR rating
> averaging one 0.09 error per game if other errors are minimized.

It is absolutely true that even the best players will occasionally
make mistakes that, according to the bot, lose a lot of equity.
Sometimes it is because of loss of concentration or carelessness,
but sometimes it's simply because the player's heuristics don't work.

This should not be surprising. In chess, for example, we know that
even the best players will occasionally blunder in the eyes of the
chess engines. For example, there was an endgame in the Carlsen-Caruana
world championship match a few years ago in which both players missed
a win that the computer found. There was just no way a human could
find that win over the board (without being told that there is a win,
of course).

Despite this, I find that countless players make the mistake of
conflating an "easy decision for a human" with a "large equity
difference according to the computer." I corresponded with James Vogl
about his excellent book, "Backgammon Super Genius Quiz," and he was
of the opinion that the only way to discriminate between top players
in a quiz would be to include positions with small equity differences.
This simply isn't true, as the Othello Quiz demonstrates year after
year.

---
Tim Chow