Re: Alex to move

I know this aint for me and I have no idea what Gnubg's breakdown is. For
Snowie however, X wins 63% after 20/14, 20/15 and 56.5% after 21/16, 10/4.

From a different angle, O has 3 ways to win

1. Scramble and muddle his checkers to safety (if there is such a thing)
about 12% of the time.

2. slots after X bolts, gets lucky and converts the slot to a point and
scrambles home about 8% of the time

3. Re-circulates after a hit and loosens the squeeze on his inflexible
position and then scrambles home about 17% of the time

In variations 2 and 3, O has about a 25% of making a new point and for X to
present him with the 21-pt is a gift well received for O will surely (if
played correctly and gets lucky) fight for the steaming-hot 4-pt. For this
pt is the most important pt in the position.

N.Merrigan
<pauldepstein@xxxxxxx> wrote in message
news:4b87e02d-3a61-4efe-a738-ac326b965427@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
On May 16, 3:29 pm, David C. Ullrich <dullr...@xxxxxxxxxxx> wrote:
On Fri, 15 May 2009 17:49:53 +0100, "N Merrigan"



<acum...@xxxxxxxxxxxxxxxx> wrote:
I see what you are getting at.

Question: should X hit if O leaves an immediate shot? Yes!!!!!! Because X
can not afford for O to make a new pt. This factor takes precedent over
any
other.

From this variation, It seems that X will be doing the waltz and now and
then break dancing between hitting and trying to make his board. Now I
understand the expression - 'I dont know whether I'm coming or going'.

An interesting question is: should O slot before X can do anything about
his
inner-board on rolls 61, 51 After 20/14, 20/15?

After stepping-off the 21-pt, O is forced to blot on 18 rolls and X will
hit
20/36 of the time. However, he'll be hitting and breaking the extremely
valuable 22 or 23-pt to do so. If played out correctly, this battle has a
long way to go and for X to give up his defence this early could be
curtains.

If O is forced to blot (8/36) after X steps-off the 20-pt, X has a 11/36
(30%) chance of hitting from the 21-pt and in the process maintaining his
defence.

I have just realised that a key issue is after 20/14, 20/15, most of O's
remaining 28 rolls are somewhat forced and his position will deteriorate.

After stepping-off the 21-pt, 18 of O's rolls are forced to slot; but
this
is a good thing (a) he could make the 21-pt if missed and (b) if not, X
will
be forced to hit and break his defence.

In a nutshell, its good for O after X steps-off the 21-pt because his
position has a 50% chance of improving as opposed to a 22% after X
steps-off
the 20-pt.

You haven't demonstrated this. You're still assuming that it's bad
for X to hit on the 21 point and "break his defense". The whole
question is whether this is bad or not! I've had very good results
hitting on the 21 point and trying to contain that checker right
now - the idea being to move X's back men into the outfield making
it impossible for O to bring that man home. At least one point of
view is that X is not breaking his defense here, he's simply
shifting to the forward part of the game. He has to "break his
defense" at some point because he can't close O out with fewer
than 12 men - the question is _when_ it's a good idea to start
this phase, and your calculations are still _assuming_ that
it's too early (or so it seems when you say that hitting right
now weakens his defense...)

O has to get very lucky to make any new points at all
without slotting, and any time O slots X still has many
shots at the slotted checker.

To put it another way: Typically we don't want to try
to go forward until after O crashes. Seems to me that
O's position right now is extremely crashed - if we
don't take steps he may un-crash. The play that
maximizes the chance of hitting O right now is
to leave a blot on the 21 point...

_Try_ a manual rollout and let us know what happens.
The relevant rollout can't be done with gnubg bcause
we're not just comparing one play to another play,
we're comparing one overall longterm stratefy to
another, and there's no way to persuade to gbubg
to try to play the game out "my way".

Play it out N times the way you're saying it should
be played.

Then play it out N times like so: Break the 21 and
make the 4 point, leaving a blot on the 21 point.
Note you now have a very bad 3-point board,
but O has just an incredibly awful 2-point board.
So you should have an advantage in a hitting
contest. Play out the rest of the game like so -
in roughly decreasing order of priority:

(i) Hit blots in O's home board whenever possible,
not worrying about whether you're "weaking
your defense".

(ii) Distribute men in the outfield so that it will
be very hard for O to bring men home after being
hit.

Keep O in the air no matter what, so he can never
make another inner point. Or try to - make it a
hitting contest.

(iii) Try to make your 5 and 6 points.

(iv) Try to maintain the 22 and 23 points so you still
have a reasonable back game in case O does escape
after an "early" hit.

When I tried it I got _much_ better results with the
second strategy. A gnubg rollout proves nothing,
for reasons noted above. I'm not convinced by
any of your arguments about what the best
current play is, because it seems to me you're
assuming that the first _strategy_ is better (I
agree that breaking the 20 point is better _if_
we're following the first strategy). The only way
to see which _strategy_ is better is to try it.
Let us know what happens when you try it.

(For example, _my_ opinion of the best strategy
_did_ change somewhat when I actually played
it out: At first my plan was not to worry about
making the 5 and 6 points until very late, more
or less walking home and outside partial prime
instead. But I found that it happened quite often
that O bounced enough, and/or exposed a second
blot often enough, that I was able to make my
board much sooner than I expected.)

When I _tried_ it I almost always got an "early"
hit, and I was usually able to contain one or two
of O's men. In the games where O brought his
man back home I often still had a reasonable
back game - O's position was still very awkward
in those games, because I'd kept him on the bar
so much he hadn't been able to build much of
a board.

_Try_ it.

N.Merrigan

"David C. Ullrich" <dullr...@xxxxxxxxxxx> wrote in message
news:pv2r05pknaemfmleh5v4rl5oudhehuvlk4@xxxxxxxxxx
On Fri, 15 May 2009 16:02:24 +0100, "N Merrigan"
<acum...@xxxxxxxxxxxxxxxx> wrote:

Questions about note 1: just a general rule of thumb.

Right. I was curious whether that was the point to that, fine.

But the things you say below really don't answer my
real question(s). I understood how all those factors would
seem to indicate that breaking the 20 point is better.
But it seems to me that the difference between breaking
the 21 point and the 20 point with regard to all those
factors is not that large, while the difference in the
number of _immediate_ rolls that force O to leave
a blot is _very_ large. How do we know that these
factors are more important than that one?

And probably what it really comes down to is
how do we know that the standard "wait until
you have a good board before hitting" is
correct here, given the fact that O's position
is very bad right now and is only going to be
_improving_? See (i) and (ii) below. Certainly
_if_ (i) is correct then breaking the 20 point
is better. But if in fact (ii) is correct, and _if_
gnubg is playing as though (i) is correct,
which seems possible, then the rollout
proves nothing.

???

Note 2. X does better with as many of O's checkers infront rather than
behind. If X steps-off the 21-pt and O manages to make it, then the
checkers on the 20-pt have lost some of their contact strength.

Also, the 21-pt is an excellent safe-landing zone given O's candlestick
formation.

If O manages to make the 20-pt, then all of O's checkers are still
infront
of X

If O fails to make it, then the vacant 20-pt is still exerting full
contact
through X's remaining defence.

N.Merrigan

"David C. Ullrich" <dullr...@xxxxxxxxxxx> wrote in message
news:1smq051fe336iotu7laq1p7jeo6n90lvao@xxxxxxxxxx
On Thu, 14 May 2009 14:05:17 +0100, "N Merrigan"
<acum...@xxxxxxxxxxxxxxxx> wrote:

**A 324 cubeful rollout yielded the following**

Well thanks for that.

I don't understand this (I must be playing it all wrong, since I
do much better breaking the 21 point. Of course the manual
rollouts were a much smaller sample.)

Questions about two notes below;

(note) In race bear-offs its generally a good idea to be 4-pt heavy.
In
turn, gaps in the inner-field are relatively harmful.

Typo for "less harmful"?

A 4-pt gap hurts you
the most.

Was this supposed to be directly relevant here or just a general
comment? I mean O's not worried about losing the race, right?

(note) When behind in a race opt for maximun contact. Stepping off
the
21-pt is a cardinal sin.

This is the one I really don't get. Yes, of course X needs to
maximize
contact. Seems to me that breaking the 21 point gives _mote_ contact.
It certainly gives more immediate blotting numbers...

???

Specific to this situation, the checkers on the 21-pt have further to
go
in
making the all important 6 and 5-pts.

The negligible difference between M1 & M2 is X fairs a little better
by
moving additional wood into the outfield and closer to home.

After 20/14 20/15, X has a 55% chance of making the 6 or 5-pt in 2
rolls.

X has a 66.66% chance on the first roll and 44.43% on the second when
O
fails to roll a 2 or 11. Therefore:

(24/36*24/36)+66.66/2=55.55%

I didn't quite follow that calculation, so I can't do the analogous
calculation myself. What do you get for X's chances of making
the 5 or 6 point in 2 rolls if he makes the same play except
breaking the 21 point instead?

Never mind that - since O blots with more numbers there's
really no way to compare the two.

I just had a thought. It _could_ be that gnung is actually
wrong about this. It's been suggested that

(i) It's too early for X to hit - he should wait until he has a
better
board.

Of course in typical backgames hitting too early is a huge
blunder (don't ask me how I know). X is still behind some
sort of broken prime, O still has some sort of board, X breaks
a back point to hit, O comes in and eventually makes the
back point that X broke and X loses big.

But it seems to me that that's backwards here, because
O's board is as bad as it's ever going to be. Seems to
me that _here_

...

read more »

David,

Out of interest, approx what percentage of games do you win when
playing X in your strategy?
It seems to me that X's position is much better than gnubg thinks it
is.

Paul Epstein


.

Relevant Pages

  • Re: Alex to move
    ... Re-circulates after a hit and loosens the squeeze on his inflexible ... then break dancing between hitting and trying to make his board. ... he'll be hitting and breaking the extremely ... we're not just comparing one play to another play, ...
    (rec.games.backgammon)
  • Re: Alex to move
    ... then break dancing between hitting and trying to make his board. ... view is that X is not breaking his defense here, ... shifting to the forward part of the game. ... we're not just comparing one play to another play, ...
    (rec.games.backgammon)
  • Re: golf question.
    ... Hitting off an Astroturf mat is more forgiving than hitting off a ... a round though isn't about learning to hit for me - it's about getting ... ready to play. ...
    (rec.sport.golf)
  • Re: Abreu needs to drop
    ... pitches he gets to hit (with his eye and hitting in front of ARod ... Probably be a good idea to check the historical records of ... and Matsui and Damon have suddenly lost the ability to play the ...
    (alt.sports.baseball.ny-yankees)
  • Re: Rules Question
    ... he was hitting a provisional. ... If you want to to proceed in that manner, his second ball is ... immediately in play no matter if you find the first one or not. ...
    (rec.sport.golf)