The Backgammon Quiz Book - #89 Part III

Here is the final roll for problem #89:


As I've mentioned before, I'm a cautious player... probably too cautious. Here's how Nick, Kansil and I played it:

Sean: bar/1/18
Nick: bar/1, 11/5
Kansil: bar/1, 11/5

Well, I'm the odd man out here, though Kansil gives me partial credit for my timid play. Kansil observes that if you are going to leave a blot, let it be a blot that works for you, so slot the 5 point. My play was based off of some advice I got from a different backgammon book (the title escapes me) that said that hit blots in you back area are less of a liability, because the checker doesn't have as far to go to make up lost ground.

For instance, if we played my move and white were to hit me on the 18 point, my pip count would go from 147 to 154, a loss of only 7 pips. If we played Kansil's pick and 5 point got hit, my pip count goes from 147 to 167, a loss of 20 pips. However, Kansil's move does have the advantage of securing an anchor in white's home area. In the end, I think I like Kansil's play better than mine.

Lets see what GNU BG says:

1. Rollout bar/24 11/5 MWC: 45.55%
0.359 0.081 0.008 - 0.641 0.256 0.025 CL 46.92% CF 45.55%
[0.002 0.001 0.001 - 0.002 0.002 0.001 CL 0.04% CF 0.11%]
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 818966479 and quasi-random dice
Play: 0-ply cubeful prune [expert]
Cube: 0-ply cubeful prune [expert]
2. Rollout bar/24 13/7 MWC: 44.46% ( -1.09%)
0.328 0.073 0.006 - 0.672 0.257 0.026 CL 46.46% CF 44.46%
[0.002 0.001 0.001 - 0.002 0.002 0.002 CL 0.04% CF 0.11%]
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 818966479 and quasi-random dice
Play: 0-ply cubeful prune [expert]
Cube: 0-ply cubeful prune [expert]
3. Rollout bar/18 MWC: 43.68% ( -1.87%)
0.324 0.087 0.004 - 0.676 0.437 0.019 CL 45.37% CF 43.68%
[0.000 0.000 0.000 - 0.000 0.000 0.000 CL 0.00% CF 0.00%]
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 818966479 and quasi-random dice
Play: 0-ply cubeful prune [expert]
Cube: 0-ply cubeful prune [expert]

Looks like Kansil's pick held up to the rollout!

The Backgammon Quiz Book - #89 Part II

Continuing with the second roll for problem 89 in Prince Joli Kansil's The Backgammon Quiz Book.

Hmmm... 3-2. My instinct is play it safe with bar/22, 24/22. A more aggressive play might be bar/22, 11/9, or even splitting your mid-point with 13/11. As I look back at games I've played, I feel that conservatism is my biggest fault... there are times when courage is required at the table if you want to win. However, black is down 24 pips and getting hit again, perhaps even twice, would be a game ending situation, so I'm going to play it safe. "Nick", the previous owner of my copy of Kansil's book, apparently felt the same:

Sean: bar/22, 24/22
Nick: bar/22, 24/22
Kansil: bar/22, 24/22

Again, the three of us are in agreement. But the real question is gnu-bg's roll out:

1. Rollout bar/22 24/22 Eq.: -0.158
0.463 0.108 0.004 - 0.537 0.136 0.004

2. Rollout bar/22 13/11 Eq.: -1.000 ( -0.842)
0.321 0.090 0.004 - 0.679 0.415 0.023

3. Rollout bar/22 11/9 Eq.: -1.000 ( -0.842)
0.322 0.092 0.004 - 0.678 0.430 0.024

Well, you can see what gnu-bg thought of my more aggressive plays... both almost equally bad, with only 0.321 chance of winning. Not that the recommended play is much better, but it is a lot closer to 50% than the others.

The Backgammon Quiz Book - #89

The first set of quizzes I'll look at will be taken from The Backgammon Quiz Book by Prince Joli Kansil, published in 1978 by Playboy Press. Joli Kansil, perhaps best known as the inventor of Bridgette, a popular 2 player version of bridge, was the backgammon editor of GAMES magazine. His quiz book contains 112 quizes for players players at all levels and for every aspect of the game. Each quiz shows a board position and 3 separate rolls to evaluate. His answers include the "right" move as well as one or two alternate plays, along with a little commentary. The used copy I aquired has the previous owner's answers penciled in, so I'll be including his picks as well as my own. We'll call the previous owner "Nick".

The first problem is #89, found in chapter 12, "Various Positional Plays." Here is the board:


The first roll Kansil gives us is snake eyes, or "the dog thow" as the ancient Roman's called it. Here is how Nick and I played it, followed by Kansil's answer:

Me: bar/24, 11/10, 6/5(2)
Nick: bar/24, 11/10, 6/5(2)
Kansil: bar/24, 11/10, 6/5(2)

Three minds with but a single thought. This one seems pretty obvious, though Kansil gives partial credit for 8/7 in lieu of 11/10. 11/10 is preferred because you are only 6 points from your 5 point and have a good chance to make the point next roll. 8/7, to me anyway, seems weak because I'd rather not fool around with my prime. What does GNU BG have to say on the matter?

1. Rollout bar/24 11/10 6/5(2) MWC: 50.96%
2. Rollout bar/24 8/7 6/5(2) MWC: 50.77% ( -0.18%)

So, GNU BG is on our side, giving the nod to 11/10. Score one for Kansil!

Welcome

Welcome to the Bar Point, a blog about backgammon.

Over the last two years, I've become increasingly obsessed with this game and its history, both ancient and modern. During the 1970's the game enjoyed a surge in popularity and a small publishing empire grew up around it, much like bridge. My compulsive book buying has landed many of these books on my shelves and as I've read them, I've wondered how well the conventional backgammon wisdom of the 70's has held up given the advances made in computer analysis in the last 40 years. Programs like gnu-bg have brought strong analytical tools to the desktop computer, enabling the backgammon enthusiast to gain a deeper understanding of the cruelest game.

The inspiration for this blog came from several directions. First, the excellent backgammon blog BlotBlog, which offers expert analysis on a variety of positions and problems. Second, the book Classic Backgammon Revisted by Jeremy Bagai, in which the author examines 120 positions in several seminal backgammon texts where the original solution contains a significant error. Finally, I was struck by this passage in Lewis Deyong's delightful Playboy's Book of Backgammon, published in 1977:

The fluid play of backgammon is something that will always resist mechanical analysis... in fact, it [the computer] might very easily blow up, like any gambler caught in a hopeless losing situation.

While it is true that the "fluid" nature of backgammon is why there are no child prodigies at backgammon as there are in chess and why human players will always stand a chance against even the strongest backgammon AIs, Deyong was clearly no futurist. The affordability of powerful computers for the home user and the development of neural net technologies have vastly changed our understanding of the game.

My plan, at least initially, for this blog is to look at backgammon problems and quizzes presented in the popular backgammon books of the 70's and re-evaluate the solutions using modern tools, mainly gnu-bg. Gnu-bg is an open source backgammon program that offers great analytical tools and plays at a world class level. I'll look at openings, checker play and doubling decisions. I'm not a strong player and my hope is that this exercises will be edifying for both myself and anyone who cares to read this blog.