I am able to calculate things effectively and quickly. Like other areas of the game, it can be improved with practice.
Bob Koca (Ph.D. mathematics professor, Baltimore)
Staring at the screen… You can learn a great deal from playing against a computer but beware of playing against ‘bots’ and not real players for long periods. Playing against humans helps your tolerance threshold stay at normal levels. My friend, Jamie Lee, became so fixated on his quest to overcome his portable backgammon computer that he would play through the night. I would get a desperate phone call from him in the morning. ‘Simon,’ he would splutter. ‘He’s up 143 points to 109. How am I ever going to catch him?’ Another problem is that you don’t get to practice your pip counting because it’s always shown on the screen for you. When you come to play a live match you’ll struggle to work out your pips in your head and know whether you are up or down in the race and by how much. This information is often essential to making the best checker and cube decisions. If you haven’t played against enough human beings, when you do, you might find yourself impatient when your opponent doesn’t move instantly, incredulous when they make an unusual double or move, and intolerant when they count a move wrongly. I do love playing backgammon with people but if you want to continue improving then a good computer program needs to be part of your practice as well. What’s the best computer program to use and how does it work? There are some very sophisticated computer programs on the market now. I’ve tested most of them and my favourite to date is Extreme Gammon (XG). Many, if not all, professionals feel the same way — currently the XG program is streets ahead of its competitors. The developer, Xavier Dufaure de Citres, is constantly updating and improving the product with direct help from several experts. XG is also the most reasonably priced, at around $60 (at the time of writing) compared to others. Computer programs have changed our understanding of the game. The program can roll out the rest of the game from any given position or cube decision in a few seconds. It can then advise you what is the best move if you were playing from the same position. It can also show you if you make the wrong move and how that can affect the rest of the match. Unfortunately, these programs don’t come right out and tell you why one move is better than another, but since they show you how many wins, gammons, and backgammons each move produces, with some practice and coaching you can usually derive enough insight from the numbers to come up with the right conclusions. Doing a rollout is more fun than it sounds because you can get the computer to do all of this analysis while you’re playing a match against it. It can then point out to you if you’ve made a small mistake, which it calls an error, or a big one, which it calls a blunder. (yes, you can turn this function off!) Many people play their matches and games against other humans on XG to save the trouble of having to input the game and positions later, because you can set the computer for two players to play each other. Also, if you play an online game against another player, your play can be automatically imported into your XG program for a full analysis of both your play and your opponent’s. The downside to this is that you cannot kid yourself about how well you’ve played; there really is nowhere to hide! Here’s how to improve your game using a ‘bot’.Always play at ‘world champion level’. You’ll spot where you make mistakes and how your opponent reacts to it. After the match take a look at the mistakes you’ve made and think about what you might have done differently.
When you play a live match and you find an interesting position or cube decision, take a note of it or even take a picture of it. Back at home, you can put this position into your program and it will tell you what it thinks you should have done.
Not only can the program tell you the best move to take, it can also list all the other moves you could make and the result of that move. If you’ve offered a double but it was the wrong decision, the computer will let you know.
Don’t forget to download the phone or tablet version of your program. The iPhone version of XG is fantastic. You can start sneaking in games when you are on the bus, in the dentist’s waiting room, or in the airport departure lounge.
XG also carries many features that will help you to understand the game better. There are advanced tools for match play, including match equity tables, take points and gammon values, racing formulas, and dice distribution analysis that show the effect of a roll from each given position. This program also maintains detailed running statistics so you can get an overview of your game and even see trends and tendencies to guide you towards specific areas of your game that could be improved.
Playing against the bots regularly gives you a wonderful dose of harsh reality. You are playing the equivalent of Mochy or Paul Magriel every time so you’re likely to lose a lot. This difficulty can be reduced if you want to win now and again however!
At the end of each match, the program will tell you how you played. A perfect rating is zero: i.e. your opponent plays at zero level if you set your computer to world champion level. Anybody who plays below a rating of 5 consistently is playing at a professional standard, and below 10 is extremely good. Your long-term average will obviously improve the more good games you play.
Kit Woolsey, one of the great players of the modern era, regularly plays a full match to 11 points against XG. He then takes time to analyse all of his moves and mistakes. John O’Hagan, another backgammon great, plays a match to 13 points every day.
Mochy, the No. 1 Giant in recent years, started a club to register players who can play live, recorded matches and play under a player rating of 4 (a 4.0 PR, the player rating that is shown on the computer). After one year, only six people had qualified. There are probably less than 50 people in the world who could qualify for the ‘Under 4 PR Club’ and maybe 500 who could consistently play under 6 PR. However, if you go to an online server and record matches, there are many playing under 2 PR. You might ask yourself how that could be possible without the aid of a bot? The answer is that many of them are playing with the help of XG or another bot!
Although it’s not as romantic nor as much fun as playing human beings, you will improve markedly if you play regularly, especially if you play in ‘tutor mode’, displaying all your mistakes as you go along.
These computers really do have the answers as to what is the best move because they are ‘neuro-nets’ which means that they constantly improve their performance by learning from experience tens of thousands of times. In a way, all of the thinking behind our checker and cube play is really a series of stories we tell ourselves to explain what is scientifically proven by computers.
If you had the chance to play against the world champion or a beginner, which would you choose? In boxing, you might prefer to take on the novice, but in backgammon, the answer is always to take on the champ and see how you get on. This is even more appropriate when playing against the computer for two reasons:
Firstly, if you win you really are achieving something and not kidding yourself into thinking you’re good by winning against the program’s beginner level. If you play against the computer on its beginner setting, it makes hilariously unrealistic and poor moves.
Secondly, as well as noting your own moves, you’ll have a masterclass from a brilliantly ruthless machine whose every move you can watch and learn from. At times, it will do things that might surprise you; at this point you can take a step back and try to figure out why it has made a particular move. Every move it makes is an excellent learning opportunity.
There used to be a column in the Sunday Express called Keep a Six Off Your Card. It was a weekly prize for amateur golfers who could play a round of golf without shooting more than a 5 on any hole This is not nearly as easy as it sounds because most amateurs will have a bad couple of holes, even if they do well on most of them. If you manage it, it’s considered a highly acceptable round and you’ll be seen as one to beat.
You can apply the ‘keep a six off your card’ challenge to backgammon. When you’re playing against XG, you can go wrong in three categories: ‘error’, ‘blunder’ and ‘huge blunder’. If you can play a match and avoid making any huge or regular blunders (i.e. keeping all blunders off your proverbial card), you will be playing at a very high standard indeed. I don’t believe it’s possible to play a match of reasonable length either against a human being or a machine without making some errors.
As soon as you start making blunders you will lose matches far more frequently against good players and if you make huge blunders you will be taken to the cleaners. Errors are less devastating though, just think hard about each move and just try to avoid blunders you will be amazed how fast your game improves.
Think of it this way. The perfect backgammon player (i.e. a computer) will play at a PR of 0. The best human beings in the world will play, on average, at a PR of between 2 and 4. If you avoid blunders in a match you will be playing to a PR of around 10. If you make blunders you might be playing to a PR as low as 20 or 25. If you are playing, on average, at a PR of 10 and you face a professional who is playing at a PR of around 7, the difference between your PRs is only 3. If you are playing that same professional and you are playing at a PR of 25, the difference jumps to 18. If you’re only a few PR levels worse than your opponent, you have a chance of winning, even over quite a long match (there is enough of a luck factor to make up for the difference). But if there is a massive gap between your ratings, you will struggle to beat him.
Can we humans play better than the best rollout computers like XG? The answer is yes and no.
No human can calculate the moves and percentages of a backgammon game as accurately and fast as a computer. On the other hand, the computer has one weakness: it always assumes it is playing another computer, or a player as good as a computer. The computer doesn’t account for normal human error.
Knowing the strengths and weaknesses of your opponent can give you an edge that the computers simply don’t have. For example, many beginners fall into patterns of flawed play that are easily spotted. You should familiarise some of these.
The most obvious beginner traits are as follows:
There are also players who will automatically reject a double if they are losing the game, even if mathematical logic says they should accept it. You can gain a big advantage against a player who does this over a period of time.
On the flip side of that coin, some inexperienced players will almost always accept a double just because they want to see how the game turns out. Even some very good players do this, including one former world champion, as I previously mentioned.
Just like in poker, there are players who have ‘tells’. They will tell you with their body language whether they are going to accept or decline before you offer them the double. If you are in a money session, some players get very conservative when they are ahead for the day and drop just about every game when offered the cube. These same players become ‘steamers’ when they are losing, taking wild chances in an attempt to turn the match around. Of course, these approaches will tend to accomplish just the opposite of their what they intended to achieve, especially if the player’s’ opponents notice and learn how to exploit this weakness.
If you’re playing very good tournament players, they’ll be well aware of these pitfalls and will be trying to avoid these same mistakes, but if you’re playing friends or at your club, or with other people online, you will start to see the players who have some of these bad habits. Currently the bots don’t account for these weaknesses as they assume they are playing against the best.
In 1918, Max Planck won the Nobel Prize for physics and went on a tour giving talks from town to town. His chauffeur boasted to him that he knew the talk off by heart and so at one town it was agreed that the chauffeur would give the talk and Max Planck would watch from the audience. This plan went swimmingly well and the talk was very well received, until somebody from the audience asked a technical question. Quick as a flash the chauffeur replied: ‘that’s such an easy question that I’m going to let my chauffeur, who is in the audience, answer it.’ The point here is that it’s not enough to learn by rote dozens of good moves from playing repeatedly against a computer. You need to understand why you are making the moves, so that you can adapt to the millions of other situations that will come up. In other words, a true understanding of the subject is better than a parrot-fashion repeating of moves.