Using All Your Tools

Using All Your Tools

(Or, How to Find out Why)

By Phil Simborg

(with lots of help from Perry Gartner)

 

 

How do you decide which checker play is best?  If you are like most “good” backgammon players, you look at the position of all the checkers on the board, you look at your opportunities, your threats, the race.  You even consider the key “big picture” factors such as score, position of the cube, doubling windows and take points, price of gammons…and you might even factor in the skill of your opponent to decide which play is best.

 

 If you are like most good backgammon players, you might move some checkers to test the candidate plays and consider which play appears to be strongest. The play that leads to the highest potential for advancing the prime might be what convinces you.    You might search your memory for similar positions or think about some general rule of thumb that will lead you to the right play.  I know that often, just remembering that if I am winning a race I should attempt to minimize or break contact.  Remembering that I am better off not getting into a hitting battle if my opponent has a stronger board is a rule of thumb I think of often…and there are many more that lead me in the right direction. 

 

 If you are a good backgammon player, the above approach will probably suffice for you to make good choices much of the time.

 

I consider myself to be a “good” backgammon player, and I do go through the above process most of the time, and I usually make good plays.  But there are still far too many times where I am baffled, and far too many times when I am relatively confident only to  discover later that I am completely wrong.

 

It was comforting to me recently to see that I was not alone.  In fact, I was recently involved in several discussions about plays where some of the best players in the world were completely wrong.  I could give you many examples, but for the sake of this discussion, take a look at the 1-1 play below where not only I got it wrong, but so did a noted world-class player who was in the game.  In fact, he and I were so sure of the wrong play that we not only made the play, we also bet on it (and lost).

 

Please take a look at this position and decide how you would play 1-1, and while you are deciding, think about your decision-making process, because I think I can offer a few surprising suggestions for you in that area later on in this article.

 

 

 

 

 

 

Position 1


 

 

 

 

There are four plays that make the most sense: 

 

  1. make your 4-point
  2. make your bar point, your 9-point, and 6/5
  3. make your bar point, and clean up your blot moving 15/13
  4. make the 9 point, 15/13, and 6/5

 

 

The play that we made and bet on was play No. 4.  The best play is to make your 4 point, and according to rollouts on Extremegammon, it’s better by nearly 10%.  That’s a pretty significant error, and usually an error like that suggests that I’ve really missed an important feature of the position.

 

When that happens and I don’t really understand what I’m missing or why the right play is so much better than another, I generally make a copy of the position and send it to some of the best players in the world to help me.

 

I got a very quick, excellent response from Stick Rice who told me that in almost all of these kinds of situations, I should just make the 4 point and I will usually be right. In other words, rather than making an improvement to a broken prime, (creating an outer priming point while moving out of the way of a direct shot), it’s better to make another key, inner-board point. 

 

Stick’s advice was extremely helpful…that gave me another rule of thumb, or reference position if you will, that will make it a lot easier for me to avoid making this kind of error in this kind of position.  But Stick also warned me that of course, it is not always right, and he said something else I consider extremely valuable and I have turned into a “rule of thumb” for future reference:  “Every position is different.”  So rules of thumb are helpful, but they don’t apply to every situation.  Move a couple of checkers, change the score, change the position or level of the cube, and plays can change completely.

 

A few days later I was at the Novi tournament with my good friend and mentor, Perry Gartner, and we had a conversation about this position.  As Perry usually does, he is not interested in talking about “what” the right play is, but “why” it is the right play and “how” to find that out. 

 

While I treasure the lessons I have had from some of the best minds and teachers in the game, they have all told me the what, and most have also told me why, but only Perry has gone to the “how” level with me.  How to know why, or the methodology to determine why one play is better than another is probably the most important lesson I’ve had in my 52 years of playing this incredible game.  By the way, as a teacher for over 20 years, I now make that my highest priority to my own students.  It is more important to know why something is right than what is right.  “What” only tells you what to do in that particular situation, but knowing “why” will help you make the right play in many situations, and you really “internalize” the answer.

 

 The reason it’s best to make my 4 point is that it is an effective counter to fend off my opponent’s most likely Game Plan while simultaneously advancing my own goals..  

 

I highlighted a part of the above sentence, as I believe it could be one of the most important concepts you will ever learn. 

 

You see, in order to decide what the best play is for me, Perry emphasized that in addition to considering all the usual variables that go into the decision-making process, I must also carefully consider what strategy my opponent will deploy to counter my GP (Game Plan) and I must determine what his GP is likely to be.

 

Wait a minute. You might say in this particular position my MCV’s (Most Common Variations) OTB (Over the Board) thoughts should have included the possibilities that my opponent would try to establish a forward anchor, possibly escape one checker, or prime, or hit my lone checker in the outfield if he gets an escaping roll. I did think of all of these scenarios. However, there was one strategy I did not give very serious consideration to, and that was the potential for my opponent to attack the last remaining checker in his inner board. 

 

Here is why!  Hitting me loose on my ace point seemed really weak since 4 points were open and there was a wide gap between the 5 and the 1. If he rolled 5-4, that would be one of his few pointing numbers--almost in the joker category, but I still would have plenty of play left.  His 5-5 is one of those few that point and gives him a great game because it also makes the 3, but it is a joker so I needn’t been concerned, I thought. After considering these ‘attacking’ numbers I discounted his ‘attack’ as not being powerful enough to win enough, and therefore thought that my attention needed to be focused on how best to contain his back checkers and went on to contemplate to what extent would I improve my prime by leaving a blot on my 15. Was my improved prime worth getting hit I asked myself?  I didn’t think getting hit on the 15 or even on the 9 was “game over” for me because his inner board was relatively weak so I indeed should consider risking a hit for a better priming formation.

 

But, Perry said that most of his wins will come from ‘attacking’, starting now or later, and went on to say:

 

“There are 2 battlefields where the potential for action is: one is where your fight will be to contain one or both of the checkers presently on your 1 point and the other potential conflict is where you want to escape from his 1 point and he is the one that needs to contain you.”

 

At that point I remembered what I had read in both Robertie’s and Trice’s books that if the player has two checkers back you should tend to prime, or block them, and with one checker back, you are forced to attack it.

 

It’s often easiest to discern the opponent’s most potent GP by considering the MCV’s, using his very next rolls as clues.   Categorize rolls to save time and look for patterns.

 

After you make your 4, there are 12 numbers that make inner points for him (3-3 is not counted here as it hits in the outfield and makes an inner point so it is discussed separately).  There are 4 additional numbers that hit loose on your ace point. I lump these 16  together  because by  I recognize they are part of the same attacking GP.  

 

 

It’s not often you need to go through all the numbers and in this case OTB experience with similar attacking positions could be enough to convince you without counting.  An expert however, could count them all quickly OTB and recognize this as a CV (common variation).

 

What is your opponent likely to do to follow up these numbers ?   On your next roll if you don’t escape, or hit back, or dance, you will probably (the MCV) bring the blot in the outfield to safety or onto a point in your outfield out of direct range. He will then go into Attack Mode if he can. The best way for him to stop the lone checker back from escaping and simultaneously stop the forward movement of your prime, is to attack that checker. He is likely to still have checkers in the zone or able to reach inside his inner board from slightly further away.  He need not close all inner board points to succeed. He may be able to put you on the roof long enough for him to free up his back checkers. If he only can bring one if these to safety and have 3 or 4 points in his inner board, his game will have markedly improved at your expense.

 

Now let’s consider the pluses for you. You already have a third inner point so he is playing catch-up in point making, and if you are hit soon you are likely to be stronger or at least equal in board strength. The quality of your return hits are high as he has only the lower points to come in on, he can dance, and some of his come-in rolls are really unpleasant.

 

Now let’s review the other plays that don’t make the 4 point in the context of his GP.   His attacking chances are better as he is likely to have as many or more point than you do when he attacks. He can freeze your forward priming movement. He can split during his attack if you are on the roof even for only half the roll. No need to examine these plays any further in comparison to this variation as the consequences should be clear by now.

 

 Lets look at the other CV which is being hit in the outfield 15 times. You should see at a glance that 13 numbers hit on your 15 point and 2 on your 9 for a total of 15. In these variations his 2 point board with a five checker stack on his 6 makes his forward progress awkward.  Again your 4 point makes a big difference. Getting hit back when you have the 4 is a much bigger loss to him. With another checker back you have the possibility of anchoring during his attack.  Priming both your checkers back is more difficult for him than for you.  You look like a favorite either way when he gives up his midpoint to hit. But for the reasons we just went over it should be clear you are a bigger favorite owning the 4 wherever you are hit.

 

That leaves 6 numbers that remain of 36 that we haven’t touched on.  OTB you might pay these scant attention (just consider MCV’s) as your other numbers are overwhelmingly in the direction of making the 4 point. You can totally avoid getting hit in the outfield with the play we made, but you face being frozen in place in this scenario and it turns out to be number 4 play on the list.

 

Since we want to have all the insight we can into the position, (it’s a great one for future reference), we should explore these 6 numbers as well. . 6-1 makes his bar point. This point inhibits your escape and potentially these checkers can attack or form a prime that holds in the rare variation where he wins prime vs prime. Since it doesn’t make an inner point (all inner points are better for the attacking game plan) it’s not nearly as good as those numbers that do. It looks like it’s a slightly worse than his average number.

 

5-6 escapes one checker. True you could avoid this escape if you make the bar but cater to his 5-6?  The same explanation holds true for the 6-6 and the 3-3. There are almost always Jokers that hurt you with any play, and when they roll those jokers you will not be happy.  But if you play to stop Jokers and give your opponent many other good rolls instead, you are going to be unhappy more often.  That is the essence of giving MCV your greatest consideration, as what is most common is that Jokers will not happen.  There are many situations where stopping the Jokers is right, but only if the sacrifice to MCV isn’t too great.  This is NOT one of those times, as the numbers prove.

 

Not that I don’t trust Perry’s insights, but I wanted to see for myself what Extremegammon thought and wanted to learn for myself if the GP that Perry predicts is also predicted by the bots. We went back to Extremegammon (and of course you could do this on Snowie or GNU as well) and made my play and then hit the “dice” button which shows us every possible roll our opponent could roll, how he would play each roll, and what his equity in the game is after each play.

 

Sure enough, this exercise showed that White would, indeed, hit loose with just about every roll he couldn’t make a point or hit in the outfield. It showed that because he usually doesn’t get hit back, his equity does go up with that play as opposed to other plays he could make. 

 

We then went back and made the “best” play according to the bot (and Stick and Perry)-- making the 4 point, and we hit the dice button, and we saw a dramatic difference in the equity gain for White if and when he hits.

 

Here is what the information looks like in Extremegammon when we hit the dice button.  It shows us how White should play every roll after we make our play, and what his equity is after each play.

 




 

We took the exercise one step further.  After choosing an “average” roll for white from the dice log, we looked to see what White’s equity is after making each play, and sure enough, that showed that again, making the 4 point was better. 

 

Then we looked to see how many rolls would give us a double after making each play.  If we make the 4 point, there are 6 rolls out of 36 where red would have a double on the next roll, even if White made the best play.  If we made the second or third best play, we found that there is not a single thing White could roll that would give us (Red) a double on the next roll!  This was another strong reason to make the 4 point that I, and the top players who looked at this position, probably didn’t consider when they chose their play.  (One of my Rules of Thumb is to consider how your play affects cube action.)

 

Because of this exercise, I now know WHY it was right to make the 4 point, and I know HOW to determine WHY.  The HOW was to decide what game plan was my opponents best chance of winning.  The HOW was to look at the next roll and see how White would play each roll, and I saw that he would attack my last checker if he could.

 

What this exercise also taught me is not only how to make better decisions over the board, but how to use the tremendous tool of the bot to provide the insight behind the numbers.  Bot evaluations and rollouts only tell us which moves are better—in this case they told us that making the 4 point is 10 percent better than making the 9 point and safetying checkers, but that doesn’t tell us why. Now without Perry holding my hand, I can use the many tools and great information the bots offer to be able to discover the “how” and “why” of a given play. 

 

I learned long agao that one of the best ways to learn why one play is better is to change the position slightly and see how being a little farther ahead or behind in the race affects the numbers, or how one more point or one less shot might change the numbers.  But looking at the dice button to see what the most likely rolls and plays are and how much the affect the numbers (equity) was a tool I hadn’t really used properly before.  (I want to add that my other good friend and coach, David Rockwell, first introduced me to this dice button a few years ago, but until now I never really used the information properly.)

 

The bots have other neat buttons you can push which also provide great insights, including the ability to see match equity, take points, doubling windows, gammon price, and how all these numbers change at different cube values.  Using all of these properly will help you learn why one play or cube decision is better than another, and as it did in this case, it helped me understand what my opponent’s GP would be and how to best foil his plans!

 

Here are the major lessons I have learned from this exercise that I hope you have learned as well:

 

1.            Even incredibly gifted, experienced, players don’t always see things properly and don’t always make the best plays;

2.            No matter how good or bad you are, you will play better, learn better from your mistakes, and remember the lessons better, if you take the time to learn why something is right instead of just what is right;

3.            In order to learn why something is right, you have to know how to find the answer.  There is methodology and there are tools to apply to every checker and cube decision.   Over the board we don’t have all the exact tools, but we have methodology that will help us approximate.  In the above example, what I, did and perhaps the expert in the game did as well, was to dismiss without going into depth, the Attacking GP, and to what extent this ‘direction of play’ (an expression coined by Paul Magriel) dominates this particular game.

 

 

 

"The MOST IMPORTANT THING I try to teach my students is to how to find out why something is right on their own."--Phil Simborg

Comments