The Hidden Pitfalls Of the Routine Play

by Gerry Tansey

I don’t worry too much about the plays I agonize about over the board.  If I care enough to take a picture of the position, I know I’ll get further feedback on my decision, first from other players, and then, most definitively, from a bot.  Sometimes I’m right or wrong by a lot; sometimes the plays I’m considering are so close that I shouldn’t have cared that much.  But no matter what, I’ll know that there was a backgammon problem that I wasn’t sure how to solve, and I’ll get an answer.

No, what I worry about are the positions I just fly through, thinking there is no thinking required.  My live matches are rarely recorded, so if such a situation arises then, I’ll never know about it.  Fortunately, most of my matches are either played against a bot or online, and when I’m done, XG will let me know about every play it thinks I butchered, whether I was aware of it at the time or not.

The following position came up in an online match.  The score is 6-5 to 7, and we have passed the Crawford game, so the trailer was allowed to use the cube.  Therefore, we are at Double Match Point (DMP).  Whoever wins this game wins the match.  Gammons and future cube actions matter not a whit.  Both players should make plays that maximize their chances to win this game.


In the game, White wasted little time with this play.  He might have uttered a silent curse to himself, but he quickly played 13/2.  Black failed to hit, White rolled 65, and soon the players were shaking hands (virtually, of course).  Nothing to see here, right?

XG took a look at this position and did its best impression of Frank Costanza’s Airing of Grievances:  “I got a lot of problems with you people!”

1.Rollout1	13/8 13/7	eq: +0.137 	Player: Opponent:	56.84% (G:8.52% B:0.12%) 43.16% (G:3.11% B:0.12%)	Conf.: ± 0.002 (+0.135...+0.139) - [100.0%] Duration: 1 minute 48 seconds 2.	Rollout1	13/2	eq: +0.093 (-0.044) 	Player: Opponent:	54.65% (G:8.33% B:0.16%) 45.35% (G:2.39% B:0.05%)	Conf.: ± 0.002 (+0.091...+0.095) - [0.0%] Duration: 1 minute 39 seconds 3.	1-ply	13/7 6/1	eq: -0.340 (-0.476) 	Player: Opponent:	33.02% (G:5.28% B:0.05%) 66.98% (G:30.86% B:1.08%)

1. Rollout1 13/8 13/7 eq: +0.137
Opponent: 56.84% (G:8.52% B:0.12%)
43.16% (G:3.11% B:0.12%) Conf.: ± 0.002 (+0.135…+0.139) – [100.0%]
Duration: 1 minute 48 seconds
2. Rollout1 13/2 eq: +0.093 (-0.044)
Opponent: 54.65% (G:8.33% B:0.16%)
45.35% (G:2.39% B:0.05%) Conf.: ± 0.002 (+0.091…+0.095) – [0.0%]
Duration: 1 minute 39 seconds
3. 1-ply 13/7 6/1 eq: -0.340 (-0.476)
Opponent: 33.02% (G:5.28% B:0.05%)
66.98% (G:30.86% B:1.08%)

XG prefers clearing the midpoint, a play that leaves two blots and twenty shots, to 13/2, which leaves just one blot.  It is not a close decision.  The right play wins the game (and thus the match) over two percent more than the second best play.  What is going on here?

Well, first note that 13/2 is not immediately as safe as one might think.  Let’s count the shots:  Any roll containing a 3 (eleven numbers), 21 and 11 (three more numbers), and 62, 44, and 22 (four more numbers).  That’s eighteen hitters, only two fewer than the 2-blot play.  But still, it is two fewer.  What else is happening?

This is the one

If White gets away with 13/8 13/7, the match is basically over.  All of White’s rolls play safely, and most of them leave at most one point to clear in front of Black’s 20-point blot.  White’s worst roll is 31, which leaves two points to clear (played 8/7, 5/1).

This is the one

Now, if White gets away with 13/2, there is more work to do.  Let’s assume that Black has not rolled 55, 66, or 65, and is thus able to keep roughly the same defensive position (potentially with a made 2-point).  White leaves another shot on the rolls 51, 41, 31, 21, 62, 42, 32, 53, 33, 22, and 11, a total of 19 numbers.  White is an underdog to roll a safe number!  Let’s do some rough arithmetic.  After 13/2, Black will hit initially with 18 numbers.  On the 18 misses, White will leave another shot about half the time, which Black will hit about a third of the time.  18 times (1/2) times (1/3) is 3 more hits, bringing the total number of hits to about 21.  All of a sudden, the 1-blot play looks less safe than the 2-blot play.

There are some shortcuts we are taking, such as ignoring Black’s big numbers that compromise his defensive position, and using one-third for the proportion of Black’s hits in all cases when he gets a second shot.  But over the board, this type of analysis is quite manageable, if one will only recognize that it is necessary and then take the time to do it.

In the online match, it was my opponent who made this mistake, but I’m counting it as my own error, as I am 100% certain I would have done the same thing if faced with the same decision.  I would have thought, “One blot good, two blots bad,” and just made the “safe” play.  Normally it is right to worry much more about the immediate shots you leave rather than what might happen on your next turn.  However, when the shot count for two plays is close, the ease of cleaning up after survival can become the deciding factor.

One thought on “The Hidden Pitfalls Of the Routine Play

  1. Additional thought: after White moves 13/8 13/7, Black’s 2 and 3 to hit are duplicated against the 2 or 3 he needs to cover the 2 point. That might be a (distant) secondary reason to make the play versus the future probability analysis already provided in the article (excellently, I might add), but it certainly can’t hurt!
    Of course 13/2 leaves 3 as the direct shot as well, so the difference is probably minuscule. But I thought the observation might be worth a discussion none-the-less.
    I’d be curious, just from a learning perspective, what the rollout difference would be if Black’s 2 builders in his inner board happened to be on the 3 and 6 points, say, rather than the 4 and 5 points. It would probably be a bigger difference if Black’s outfield point was the 16, say, instead of 15 (giving a 4 shot against the 13/2 play rather than a 3 shot).
    All this is to say this seems like an interesting position to explore the relative merits of the duplication concept. If I had XG I would do it myself… But who needs XG when you have Gerry?!


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s