BGonline.org Forums

54S-42P-32 — Summary & Analysis

Posted By: Taper_Mike
Date: Thursday, 14 June 2012, at 4:24 a.m.

In Response To: 3rd Roll Position from Robertie (Taper_Mike)

 White is Player 2 score: 0 pip: 161 Unlimited GameJacoby Beaver pip: 158 score: 0 Blue is Player 1
XGID=-b----E-D---eD---b-dAb--A-:0:0:1:32:0:0:3:0:10
54S-42P-32

 Advanced Backgammon, Volume 1, Problem 11

An early 32 is often awkward. One could write a book about the many conundrums it causes in the first three rolls (if only someone would buy it!).

• Having read Matt’s post, I am inclined to agree that, back in 1984, Robertie was ranking plays thusly: S D s T R \$ %. My rollout gives R s T S \$ % D. (A Nactation key is given at the end of this message.)
• Robertie has the best play, R = Run (20/15), ranked 5th. He recommends, in the early game, that one not leave any blots in the opponent’s outer board (on his 9pt, 10pt or 11pt). When you are hit there, says Robertie, your opponent gets to build at the same time. So much for theory.
• Robertie’s 2nd-best play, D = Down (13/11 13/10), came out 7th best in the rollout. After an opponent opens with a point-maker like 31 or 42, it is fundamental not to play two checkers down with rolls like 32 and 43. (The exception is 42P-43, where the alternative split is tied, or slightly ahead of playing down.)

Playing down is an attempt make a prime, at a time when you are already behind in the priming game. A better strategy is to split, bidding for the advanced anchor that will defeat your opponent’s prime. In addition, when you put two targets in the outer board, there is a significant chance that your opponent will hit one of them, and escape one of his rear checkers. That eventuality would make a bad situation much worse.
• In order to give Robertie his due, I gave his recommended play, S = Split (20/18 13/10), 10k trials. It rolls out 0.045 behind the leader. If it had been 0.05, Jeremy Bagai would have deemed it an error, and included it in his book.
• For those who know the replies, Neil Kazaross gives a clever way to ferret out the correct move. Says Neil, “The 63 response to a 42 opening is a close choice between splitting and running, but splitting wins by a smidge. Here we have the same decision except that we already have new spare on the 8pt, and are 5 pips farther ahead in the race. I think that may be enough to change the play in favor of 20/15.”
• In the post cited above, Matt asks, “[Why is s = split (24/22 13/10)] so close behind? Why is it so much better than 20/18 13/10?”

Five years downstream, Matt is better qualified to answer than I am, but nevertheless, here is my take: Of the three best plays, s (split) is the only one that attempts to improve the offense. That is worth something.

On the defensive side, moving to the 22pt threatens White’s stripped 8pt, at a time when White has only 8 checkers in the zone. If White hits, there will often be one or more direct shots in return.

Also, by slotting both anchors, Blue gives himself a good chance to make one of them. Whichever blot does not get hit will often become an anchor. By comparison, moving a blot out to the 18pt walks right into a direct shot from the stack sitting on White’s midpoint. White needs to unload that stack, and giving him a chance to do so with tempo is not the right idea.
• Another question Matt asks is, “Why on Earth is 13/11 13/10 so bad? I understand why it would be behind plays like 13/8 and 20/15, but it’s also behind 13/11 8/5, which leaves a direct, slots while split, and duplicates nothing.” Good question. Any takers out there?
• Keene posts a rollout of 54S-53P-32, an interesting comparison position. Tied at 5-away, his quasi-money rollout features the same top three plays, but this time, advancing the rear checker, and bringing a builder down, is best. There is no need to sacrifice the offense in his variant. My unlimited games rollout of the same position can be found here.
Key (ordered from best to worst): R = Run (20/15), s = split (24/22 13/10), T = Tower (13/8), S = Split (20/18 13/10), \$ = Slot (13/10 6/4), % = Alternate slot (13/11 8/5), D = Down (13/11 13/10).

Mike

Post Response

Subject:
Message: