Splitting in a priming game
USRobots
Here's a position for your consideration:
24 23 22 21 20 19 18 17 16 15 14 13
-----------------------------------------------------
| O X X | | X X X O |
| O X X | | X O |
| X X | | X |
| X | | X |
| | | |
| |BAR| |
| | | |
| | | |
| O | | |
| X O O O | | O |
| X O O O | | O O O |
|_______________________|___|_______________________|
1 2 3 4 5 6 7 8 9 10 11 12
score -3:-3, cube=1, pip count: O-145, X-139
O to play 2-1
As O, I considered the following options over the board:
ALSO-RANS
=========
24/21 Comes within range of 4 attackers
24/22 any 1 Still within range of 4 attackers
6/4 any 1 Too aggressive: only 2 builders, and bar point is open
PLAUSIBLE
=========
10/7 Too aggressive: I can make it naturally
11/9 10/9 Too cautious: six away from made 3-pt, removes builder
CONTENDERS
==========
11/9 13/12
This maximizes builders for the 4- and 7-points, but leaves 4 blots and 9
immediate shots, and may be difficult to clean up later.
Safer moves involving two outfield checkers
Do not challenge X's prime building
13/11 24/23 (my actual move)
Maintains builders while challenging X's prime building.
After my actual move, X rolled 5-5, which was now very good for X instead
of very bad. X played 19/24*(2), 17/22 (2), I rolled badly, and finally
ended up with 4 checkers on the bar against a closed board. I continued
to roll badly, didn't even get a shot, and lost a backgammon and the match
(X was down to two checkers on the 24-pt and I still had two in the air
when I resigned.)
If I reached this position again, I would probably play one of the safer
outfield moves, mostly because I have a timing advantage that I had not
considered over the board.
I would like to pose the following questions:
1.) Was my actual move the best? If not, was it at least close, or was it
clearly inferior?
2.) What features of a priming game are most important when one is
deciding whether to split one's runners? Magriel's "Backgammon" merely
states that such decisions are very difficult, while Robertie's "Advanced
Backgammon" doesn't seem to cover the issue at all.
3.) (Bonus question) What are the backgammon chances with four checkers
closed out?
Thanks for any input,
USRobots
usro...@aol.com
David Montgomery
> 24 23 22 21 20 19 18 17 16 15 14 13
> -----------------------------------------------------
> | O X X | | X X X O |
> | O X X | | X O |
> | X X | | X |
> | X | | X |
> | | | |
> | | | |
> | X O O O | | O |
> | X O O O | | O O O |
> |_______________________|___|_______________________|
> 1 2 3 4 5 6 7 8 9 10 11 12
>
>score -3:-3, cube=1, pip count: O-145, X-139
>
>O to play 2-1
>1.) Was my actual move the best?
There are so many possibilities here that look plausible
that I would not expect to find the best move over the board.
In fact, sometimes in positions like this I can't find the
right move using rollouts until I've looked at if for a
couple of months, since the best move may not occur to me at first.
My first reaction is to play 24/23 10/8. I think this is
better than 24/23 13/11 because:
1) The 11 point is not such a valuable point, being 6 away
from the 5 point. The midpoint isn't vital either, but
I think since it provides better coverage of the outfield
and blocks 6-6, it is a slightly better point to own.
2) 10/8 leaves only 1 outfield blot, subject to 2 hits (6-4),
while 13/11 leaves 2 outfield blots, subject to 4 hits
(6-3, 6-6, 3-3) of which two are tremendous rolls for X.
3) The fact that 10/8 removes a builder 6 away from the 4
point is not that significant because O has a lot of uses
for sixes already, either running or making the bar point
or even making the two point.
>If not [best], was it at least close, or was it clearly inferior?
In my opinion, it is close. It looks a little worse to me, but
not by much.
>2.) What features of a priming game are most important when one is
>deciding whether to split one's runners? Magriel's "Backgammon" merely
>states that such decisions are very difficult, while Robertie's "Advanced
>Backgammon" doesn't seem to cover the issue at all.
Whenever you split your back checkers, a principle consideration
is how strong your opponent's attack is likely to be. This is
primarily determined by
1) your ability to defend yourself, which is determined by
A) whether you have an anchor
B) how many blots you have about the board and how vulnerable they are
C) how loose your opponent's position is (i.e., whether you are
likely to be able to return hit ).
D) how strong your offense is, since a strong offense may
deter your opponent from a full-fledged attack.
2) your opponent's ability to build a strong board, which is determined by
A) how strong the opponent's board is all ready
B) how much and how readily material is available for building
new points.
3) your opponent's ability to come home after a successful attack.
For example, if you have your opponent primed, then your opponent
will never be able to come home after attacking you. Even if
you have a 4 or 5 prime, the opponent's attack may temporarily
succeed, but then fall apart because of an inability to free
the back checkers.
The above I think applies quite generally. In priming games,
the additional consideration of note is timing. If you will have
great difficulty extracting your rear checkers, and you have significantly
less timing than your opponent, then splitting is very often called for
even in the face of a potentially strong attack by the opponent. By
splitting you hope to gain timing in three ways:
A) the actual movement of the back checkers are pips that you
avoid moving on the offensive side.
B) if you can escape a back checker, you usually pick up about
two rolls worth of timing.
C) if you are hit, you move backwards, and you may gain even
more timing due to fanning.
On the other hand, you may just get closed out and backgammoned :-).
But if you have a strong prime you hope that category 3) above will
give you good winnning chances even when attacked.
On the other hand, if you will probably not have such great difficulty
escaping the back checkers (and your opponent will), and you have more
timing than your opponent, then there is no need to split. Just contain
the opponent, wait for his or her prime and board to bust to splinters,
come around (maybe trapping the opponent off his or her anchor while
you're at it) and win a backgammon :-).
In between is a huge gray area.
In the given position, I don't think priming considerations give
any great motivation for splitting. O looks to have the better timing,
by just a bit. Since the opponent's bar point is open, O should have
escaping chances for some time. Even if X closes the bar, X will only
have a four prime, and O will probably be able to split to the 22 or
21 points if necessary. For all these reasons, I'm not 100% convinced
that splitting is correct, although I would do it.
For me the split is motivated by the fact that O's prime is not so
strong either. With the bar point open, X may be able run to a checker
out and gain both a timing advantage and the advantage of having one
less checker back. It may turn out that neither side ever fully
primes the opponent, and the advantage in this case is likely to be
with the player that can escape a checker. By splitting, O increases
the likelihood of being this player, as well as getting more fly
shots at X's outer board blots. X has a lot of material ready for
an attack, but currently has only a two point board, so I think that
the blitz is only a small danger.
>3.) (Bonus question) What are the backgammon chances with four checkers
>closed out?
I don't know. I'm sure I've seen this rolled out somewhere though --
I think in _Fascinating_Backgammon_. If no one else posts this I'll
look it up.
>Thanks for any input,
>USRobots
>usro...@aol.com
David Montgomery
monty on FIBS
Ron Karr
> 24 23 22 21 20 19 18 17 16 15 14 13
> -----------------------------------------------------
> | O X X | | X X X O |
> | O X X | | X O |
> | X X | | X |
> | X | | X |
> | | | |
> | | | |
> | | | |
> | O | | |
> | X O O O | | O |
> | X O O O | | O O O |
> |_______________________|___|_______________________|
> 1 2 3 4 5 6 7 8 9 10 11 12
>
>score -3:-3, cube=1, pip count: O-145, X-139
>
>O to play 2-1
>
>Maintains builders while challenging X's prime building.
>
Actually, I don't see this play as "challenging" X's prime building. X has all the
numbers to make points that he would have otherwise had. In addition, he now has
numbers to point on you or attack you, many of which (like the 5-5) would otherwise
have been awkward. Splitting like this is much more effective when the 8 point is
stripped (i.e. no spares). Then he can't use it to build a new point without leaving a
direct shot. In this position, for example, if X splits, your builder on the 8 point
would be frozen.
>If I reached this position again, I would probably play one of the safer
>outfield moves, mostly because I have a timing advantage that I had not
>considered over the board.
>
>I would like to pose the following questions:
>1.) Was my actual move the best? If not, was it at least close, or was it
>clearly inferior?
There are a couple of issues here:
1. Should you split?
2. If so, which play should you make?
If you do decide to split, I think the play you made is "too big." Not only are you
subjecting yourself to attacks inside X's board (which negate the flexibility of your
builders, since you can't make new points if you're on the bar), but you're breaking
your midpoint, thereby turning 66 into a killer for X (for the rest of the game), and
leaving 4 exposed blots. If I were going to split, I'd probably play 24-23, 10-8.
However, as you pointed out, you do have a timing advantage. I think it's unnecessary
to split here, since you have no great urgency to escape. You have lots of ways to
improve your board, so I think you should just concentrate on that. You also have safe
plays if you can't make new points. You may also make an advanced anchor or hit a fly
shot WITHOUT splitting.
What play would I make? 11-9 10-9 is not that silly. It's fairly flexible, and leaves
NO real jokers for X (in a game that figures to last for a while, leaving a few killers
every roll can add up). However, it may be worth the risk to play 11-9 6-5. This
leaves 4 builders for the bar and 4 for the 4 point, only 6 numbers to get hit (all of
which otherwise make X's 3 point) and many returns if I'm hit.
>
>2.) What features of a priming game are most important when one is
>deciding whether to split one's runners? Magriel's "Backgammon" merely
>states that such decisions are very difficult, while Robertie's "Advanced
>Backgammon" doesn't seem to cover the issue at all.
I think the main consideration is: whose back checkers are more likely to get trapped?
This depends on 1. the relative strength of the two primes; 2. "timing" or
"playability": the ability to move without having one's position deteriorate. A
player who is about to run out of moves is more likely to want to split. When you have
lots of constructive moves on your side of the board (compared to your opponent),
you're less likely to split.
In addition, there are tactical considerations like: how likely are you to get attacked
if you split? Are you putting pressure on your opponent's builders by splitting? The
advantages of splitting, compared to staying anchored on the 24 point, are that you
give yourself more chances to do good things such as:
1. Make a more advanced anchor (e.g. 23, 22, 21 or 18 points)
2. Hit an indirect shot in the outfield (e.g. 5-2 & 5-4 in addition to 6-2 & 6-4,
assuming X leaves his blots there)
3. Run a checker to safety on the midpoint (5-5 in addition to 6-5)
4. Make it much more risky for X to slot a key point.
The main disadvantage is that you open yourself to being attacked.
X can hit you with otherwise awkward numbers. You experienced the 55s. Other numbers
let him hit you loose & split his back men.
By the way, there's a decent discussion of splitting in priming games with some sample
games in Papazian and Kennedy's book "Backgammon Master Games".
>
>3.) (Bonus question) What are the backgammon chances with four checkers
>closed out?
I'll leave this question to someone with rollout data.
Ron Karr
ka...@best.com
FIBS: ronkarr
Kevin Cline
David Montgomery <mo...@cs.umd.edu> wrote:
|>In article <3sd49i$k...@newsbf02.news.aol.com|>
|>|> 24 23 22 21 20 19 18 17 16 15 14 13
|>|> -----------------------------------------------------
|>|> | O X X | | X X X O |
|>|> | O X X | | X O |
|>|> | X X | | X |
|>|> | X | | X |
|>|> | | | |
|>|> | | | |
|>|> | O | | |
|>|> | X O O O | | O |
|>|> | X O O O | | O O O |
|>|> |_______________________|___|_______________________|
|>|> 1 2 3 4 5 6 7 8 9 10 11 12
|>|>
|>|>score -3:-3, cube=1, pip count: O-145, X-139
|>|>
|>|>O to play 2-1
|>|>[ USRobots played 13/11 24/23 ]
|>|>
|>|>1.) Was my actual move the best?
|>
I have no idea, but I would not split here. After keeping track of
this for 17 years, I have found that 97.2% of the time
my opponents roll doubles immediately after I split.
|>
|>My first reaction is to play 24/23 10/8. I think this is
|>better than 24/23 13/11 because...
|>
|>1) The 11 point is not such a valuable point, being 6 away
|> from the 5 point. The midpoint isn't vital either, but
|> I think since it provides better coverage of the outfield
|> and blocks 6-6, it is a slightly better point to own.
|>2) 10/8 leaves only 1 outfield blot, subject to 2 hits (6-4),
|> while 13/11 leaves 2 outfield blots, subject to 4 hits
|> (6-3, 6-6, 3-3) of which two are tremendous rolls for X.
3-3 and 6-6 are tremendous rolls no matter what, and if X rolls either
number you will be doubled out unless you can roll your own 3-3 or 6-6
immediately. So if you leave blots that are hit by those numbers,
you have actually reduced the number of winning rolls for X. So leaving
a blot on the 13 point is safer than leaving one on the 11 point.
|>3) The fact that 10/8 removes a builder 6 away from the 4
|> point is not that significant because O has a lot of uses
|> for sixes already, either running or making the bar point
|> or even making the two point.
|>
Yes, but 10/8 also removes a crucial builder from the bar point.
Making the bar point is more important than making the
four point here because it will be relatively easy to make the four
point after making the bar point. If O makes the bar point and
X rolls a couple of 6's then I think O can double X out.
I think I would play 11-9 6-5, leaving four builders for the bar point and
creating a new builder for the four point. If X makes his bar point first,
then you will be forced to split.
Kevin Cline (TheSkeptic)
--
Kevin Cline
Kit Woolsey
: Hi everyone,
: Here's a position for your consideration:
: 24 23 22 21 20 19 18 17 16 15 14 13
: -----------------------------------------------------
: | O X X | | X X X O |
: | O X X | | X O |
: | X X | | X |
: | X | | X |
: | | | |
: | |BAR| |
: | | | |
: | | | |
: | O | | |
: | X O O O | | O |
: | X O O O | | O O O |
: |_______________________|___|_______________________|
: 1 2 3 4 5 6 7 8 9 10 11 12
: score -3:-3, cube=1, pip count: O-145, X-139
: O to play 2-1
My choice would be 24/22, 11/10. With the offensive four point open,
there is great value in making the ten point. One of X's paths to
freedom is via the four point -- by making the ten point you do a lot to
block that patch. In addition my play leaves no outfield blots, which
could prove to be very important if things start to go bad.
On the defensive side of the board, the split to the 22 point is the
right idea. This split puts O in position to escape, but it is not the
point X wants most. Thus, even if X points on O X's four point will
remain open. The danger of being blitzed is not too great -- after all,
O does have the stronger inner board. By splitting up to the 22 point O
takes the sting out of a lot of rolls which would otherwise be good for X
-- O is attacking the blot on X's nine point, so if X rolls something
like 2-1 or 3-2 he can't quitely make the four point (the point he wants
to make) without giving O some retaliation. If O just stays put then X
will quietly build his prime, and O is likely to have to worst of it in a
priming battle since X owns the offensive bar point but O does not.
Granted splitting loses badly when X rolls something like 5-5, but
overall it figures to be the winning action.
Kit
Kit Woolsey
8 point, not four men on the bar point as I had thought. Therefore O is
not hemmed in, so there is no need to split the back men into the jaws of
death. 11/9, 10/9 is now my clear choice.
If X did have the four men on the bar point, then splitting would be
superior.
Kit
Walter G Trice
> 24 23 22 21 20 19 18 17 16 15 14 13
> -----------------------------------------------------
> | O X X | | X X X O |
> | O X X | | X O |
> | X X | | X |
> | X | | X |
> | | | |
> | |BAR| |
> | | | |
> | | | |
> | O | | |
> | X O O O | | O |
> | X O O O | | O O O |
> |_______________________|___|_______________________|
> 1 2 3 4 5 6 7 8 9 10 11 12
>score -3:-3, cube=1, pip count: O-145, X-139
>O to play 2-1
The big issue in this position is whether to split, and as
O I would not. The reason is that O is ahead in the prime
building contest, and splitting lets X shift plans and
attack. X is well placed to attack -- one important factor
is the sheer number of checkers he has close to his home
board.
If O just keeps working on his prime he rates to keep an
advantage. X can't easily respond to the priming threats
by splitting or running because then O can shift to
attacking himself, since his board is better and escaping
his back men is not yet a problem.
Given no split, my 1st choice is 11/8 keeping 3 builders
each for the bar and 4 points and only 1 blot. 2nd would
be 13/10. The 10 point has some utility being 6 away from
a key point -- it's a builder that can't be hit and it
might block a number that would otherwise run in a turn or
2. Other multi-blot plays seem a bit rich.
-- Walter Trice
USRobots
From my last post:
"Thanks for all of your replies! After looking them over, my personal
preference is 11/9 6/5, which I saw neither over the board nor during my
analysis. This play keeps 4 builders for *both* the bar point and the
4-pt. I don't think splitting was that bad, because of its many
advantages noted by Ron Karr, but was definitely not the best.
"Kit's play, 11/9 10/9, seems a little too conservative, if I may be so
bold as to disagree with him :) It leaves only 3 builders for each key
point. The difference between 3 and 4 builders is often huge! In this
case, after 11/9 6/5, 27 out of the 36 rolls for O (assuming no hits) make
either the 4- or bar point. After 11/9 10/9, 18 rolls do so. The
tradeoff seems worth the extra 6 shots.
"Any other opinions/JF rollouts?"
Oops, forgot a couple of things.
One is the position:
24 23 22 21 20 19 18 17 16 15 14 13
-----------------------------------------------------
| O X X | | X X X O |
| O X X | | X O |
| X X | | X |
| X | | X |
| | | |
| |BAR| |
| | | |
| | | |
| O | | |
| X O O O | | O |
| X O O O | | O O O |
|_______________________|___|_______________________|
1 2 3 4 5 6 7 8 9 10 11 12
score -3:-3, cube=1, pip count: O-145, X-139
O to play 2-1
The other is that 3-3 hits after 11/9 6/5, bringing the total number of
shots up to seven. Of course, it's not clear whether it is proper to hit
after 3-3! Therefore, my original opinion stands.
Thanks,
USRobots
USRobots
Thanks for all of your replies! After looking them over, my personal
preference is 11/9 6/5, which I saw neither over the board nor during my
analysis. This play keeps 4 builders for *both* the bar point and the
4-pt. I don't think splitting was that bad, because of its many
advantages noted by Ron Karr, but was definitely not the best.
Kit's play, 11/9 10/9, seems a little too conservative, if I may be so
bold as to disagree with him :) It leaves only 3 builders for each key
point. The difference between 3 and 4 builders is often huge! In this
case, after 11/9 6/5, 27 out of the 36 rolls for O (assuming no hits) make
either the 4- or bar point. After 11/9 10/9, 18 rolls do so. The
tradeoff seems worth the extra 6 shots.
Any other opinions/JF rollouts?
Thanks,
USRobots