Cash game, center cube.

White to play 6-3.

This position shows one of the most commonly misplayed early game situations. White opened with a 43 and made the building play, 13/9 13/10. It’s the most aggressive choice for an opening 43, designed to make a prime quickly and create a complicated position. Black responded with a 44, making his 9-point and electing to make an advanced anchor rather than hit. (Hitting is also very reasonable.) Now White has a 6-3 to play.

White’s two blots are under attack, and he can deal with the problem in a few obvious ways. Let’s look at their merits and demerits.

Play (A): Make the bar with 13/7 10/7. This play cleans up one blot while making the bar-point. While not a blunder, it’s also not a play to get excited about. Once your opponent has moved up to the 5-point anchor, the bar point doesn’t have much value. It doesn’t give you the strength of an inner board point, and it doesn’t really have much blocking value. In addition, it leaves the midpoint stripped, which is usually a nagging weakness early in the game.

Play (B): Make the 3-point with 9/3 6/3. Compared to the bar-point, this play has a few advantages. It makes an inner-board point, it unstacks the 6-point, and it leaves a spare on the midpoint. However, the 3-point isn’t that strong at this stage, and White still doesn’t have much of a game.

Play (C): Make the 10-point with 13/10 9/3. This is a better plan than the first two since the 10-point has more blocking value. Although we haven’t left a blot, that’s not a big deal. Black would have to break his great anchor to hit anything, so a blot would represent a very small risk.

If these three plays were our only choices, we’d go with Play C. At this point, however, we need to step back and look at the position as a whole. So far we’ve just been focused on cleaning up our two blots in some way. Now let’s look at the position with fresh eyes and see what’s really going on.

Black has both a great defensive anchor and a good building position. If we make one of the plays above, next turn almost all Black’s rolls will make a good point on his side of the board – either the 5-point, 4-point, bar-point, or 3-point. (Only 5-1 and 3-2 miss altogether, while 6-4 at least makes the 2-point.) After Black makes a new point, we’re going to have to get our back men moving before they get blocked in any further. Maybe the dice will let us do that, and maybe they won’t. But even if they do, we’ll be running at a time when Black’s game is even stronger than it is right now.

If we need to move our back men quickly (and we do) then we’re much better off moving them now, while we can and before Black has any home board. The right idea is therefore 24/18 13/10! This prepares to make a great anchor and covers one of our blots, while ensuring that if Black hits our checker on the 18-point, he can’t also make an inner-board point without throwing a perfecta. Although the play looks loose, since Black can hit somewhere with almost every number, it’s hard for Black to make real progress with such a weak inner board.

—–

The first commercial neural net to make an appearance on the backgammon scene was Jellyfish, in 1994. When players started practicing against Jellyfish, one of the first things they noticed was that it was almost impossible to pin Jellyfish behind a prime or back in an ace-point game. Those were standard game plans in master play, but you could play an entire session against Jellyfish without ever seeing those positions arise (unless you were the one stuck in an ace-point game!) The reason were plays like 24/18 13/10. Jellyfish had a better sense than the players of that era of the importance of splitting, even into apparently strong threats.

Here’s a simple idea to take away from this position: When your opponent makes a strong advanced anchor quickly, you need to split more than ever and try to equalize with an anchor of your own, even if the split seems to involve some risk.