#12 GNUBG 2-ply or BG GIANT

[Zorba]

A pretty straightforward problem:

GNU Backgammon Position ID: sM/CgCHgc8MBUA
Match ID : cIkmAkAAYAAA
+24-23-22-21-20-19------18-17-16-15-14-13-+ O: White O
| X O O | | O O X | 4 points
| O O | | O X |
| O | | X |
| O | | |
| O | | |
| |BAR| |v 17 pt match (Cube: 1)
| X | | |
| X | | |
| X | | X |
| O X | | X X O | Rolled 51
| O O X | X | X X O | 12 points
+-1--2--3--4--5--6-------7--8--9-10-11-12-+ X: Blue X

Pips: X 160 O 155

Game 11, X to play 1-5 (from the bar)

[Peter Grotrian]

I don't see enough immediate threats of O which would justify hitting
loose on the ace. I like to see O play something with his stacked and
stripped position and choose 13/8. My motto (from Robertie): Don't hit
against stacks!

There is a strong argument for hitting however. It doesn't strip our
own midpoint and develops the checkers on our 6pt. If X is hit, he has
more possibilties to form an advanced anchor, eg 43 and 32, without
splitting his ace point anchor which is somewhat dangerous here. I have
convinced myself now and hit.

Peter

[paulde...@att.net]

Strongly disagree! After entering, X should play 13/8. 0 is
developing priming threats and X doesn't want to be hit back after
hitting loose with 6/1* 6/1* is a very bad play -- bad for X if 0
hits but no real gains if 0 misses.

Paul Epstein

[Philippe Michel]

X's position stinks. 13/8 is frustrating in that it does nothing useful
and significantly weakens X's position. It merely completes the play
because one just has to play both dice.

6/1* is more active. It creates a few nasty rolls for O (6-2, 6-3, 5-3)
but loses quite a lot if the blot is hit (only 9 hits I guess ; I don't
think O should hit with 5-1). The balance is negative as well.

I have no strong convictions for one play or the other, but I'm pretty
sure that on the board I would have played Bar/24 13/8.

[Raccoon]

Peter Grotrian wrote:
> I don't see enough immediate threats of O which would justify hitting
> loose on the ace.

O's immediate threat, which I'm sure Peter does see, is to improve his
prime before X can improve his anchor. Stopping that threat is another
reason for playing 6/1*, in addition to the ones Peter mentions.

See how much better hitting is if O's next roll is 66, 62, 63, 52, 54,
42 or 53.

Looking just at O's distribution of spares (the stack on the 6 point,
no spare on the 8 point or midpoint), it's not all that good and may
not seem very threatening. But the made 5 point is already a big
advantage for 0. It's likely that just one more positional improvement
will be enough for O to double. X should try to prevent that.

> There is a strong argument for hitting however. It doesn't strip our
> own midpoint and develops the checkers on our 6pt. If X is hit, he has
> more possibilties to form an advanced anchor, eg 43 and 32, without
> splitting his ace point anchor which is somewhat dangerous here.

If getting hit means making the 22 or 21 point and also having a spare
back man to play, getting hit isn't bad at all.

[Raccoon]

paulde...@att.net wrote:

> 0 is developing priming threats

That's the best argument for 6/1* -- to preempt O's priming threats.
Look at the swing on O's 66 62 63 54 53 42.

> and X doesn't want to be hit back after hitting loose with 6/1*

If it helps O establish a better anchor, with a spare back checker to
play to boot, getting hit isn't bad at all.

[Walt]

Zorba wrote:

Well, Bar/1 is forced. What to do wit the 5? The choices are 6/1*,
13/8, 8/3 and 9/4, the last two of which don't make any sense at all.

Hitting loose on the one point is likely the better play, but since I'm
not convinced I'll take the safe play:

Bar/1 13/8.

//Walt

[Grunty]

At that score, X has to fear O's double, so the passive play 13/8
doesn't seem to be indicated.
X should play 6/1*, depriving O half his roll. I think a subsequent
double by O is more afordable this way than would be giving him
complete freedom to make the bar or another inner point.

Worst possibility is getting hit on the ace, well, he'll then have a
third checker back to maneuver trying to establish a better anchor
without having to give up his rear anchor.

[Zorba]

X played B/24 13/8
GNUBG 2-ply picks B/24 6/1*
Rollout prefers B/24 6/1*

Score X -0.0729 (-0.244%)
Subtotal: X -0.6414 (-2.691%)

1. Rollout bar/24 6/1* Eq.: -0.5244
40.09% 9.22% 0.43% - 59.91% 20.21% 0.81% CL -0.3606 CF
-0.5244
[ 0.11% 0.13% 0.04% - 0.11% 0.13% 0.05% CL 0.0033 CF
0.0064]
Full cubeful rollout with var.redn.
3888 games, Mersenne Twister dice gen. with seed 984532736 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 0.13
Skip pruning for 1-ply moves.
Cube: 2-ply cubeful prune [world class]
2. Rollout bar/24 13/8 Eq.: -0.5972 (
-0.0729)
38.66% 8.31% 0.41% - 61.34% 20.87% 0.81% CL -0.4029 CF
-0.5972
[ 0.10% 0.12% 0.05% - 0.10% 0.13% 0.04% CL 0.0031 CF
0.0058]
Full cubeful rollout with var.redn.
3888 games, Mersenne Twister dice gen. with seed 984532736 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 0.13
Skip pruning for 1-ply moves.
Cube: 2-ply cubeful prune [world class]

Thanks for the replies, everyone!

My post-rollout ideas: X's position indeed stinks, with the acepoint
anchor versus O's fivepoint anchor, plus O's offensive fivepoint. With
the race about equal, X's timing is therefore terrible.

13/8 makes X's position very fragile. How to play the next roll? X
can't afford to give up the midpoint with the two backcheckers still
completely undeveloped. Losing the last spare from the mid looks very
costly here. X is likely forced to dump a checker and create blot in
his homeboard next turn, if he doesn't roll well.

Also, O is threatening to strenghten his position even further by
making his nine-, bar- four- or even threepoint. Unlike X, O can
afford to give up his midpoint if need be, because O has the advanced
anchor and O has serious chances to succeed at priming X.

So how bad is getting return-hit after 6/1*? Well, X has an equal race
which would normally make it costly to get hit like that. Here however,
X has no serious chances to get in a race for at least several turns,
so losing in the race isn't so bad. Also, X would very much like to
advance his anchor which is difficult from the 24pt versus O's offense.
A third checker would really help here.

How bad is the checker on the acepoint if it does NOT get hit? Well
it's a bit ugly (as usual), but with O anchored high it doesn't look
like a big deal. X won't get a chance to close-out or prime O, so
slotting/making a deep point isn't terrible here.

The rollout shows that plusses and minuses combined, 6/1* is clearly
worth it.

--
_
/
_