Learning from the Machine



In these articles we are going to look at how you can use modern technology to improve your game and also get more enjoyment from it on a day-to-day basis. To do that we have to understand the history of backgammon and also how we, as human beings, develop our understanding of games and therefore our skills. Let’s start by looking at the history.


Backgammon is approximately 5,000 years old and for the first 4, 900 years the pace of development and learning was extremely slow. Initially instruction would have been by word of mouth only and I suspect the levels of skill remained the same for centuries.

Backgammon appears in art and literature throughout time and is mentioned both by Chaucer and by Shakespeare. The invention of the printing press could have given the game a leap forward but despite its popularity instruction manuals just didn’t appear (in contrast to chess which developed a bibliography early in its history).

Backgammon does feature in Hoyle but one book was not going to change matters. And then in the mid 1920’s the concept of doubling was introduced and the game was changed forever. Thousands of people ‘discovered’ the game and the clubs of New York and Boston were full of backgammon players. Suddenly everyone rushed to publish a book on the game in an effort to cash in on the new craze.
Most of these books were basic primers but they at least got backgammon on the map as far as published material was concerned. Then came the Wall Street crash, overnight the money disappeared and the game went into hibernation for thirty odd years.

And there it might have remained but for the efforts of Prince Alexis Obolensky who created the international backgammon circuit in 1964. Suddenly the game was back on the map and importantly the money, created by the boom in oil, was back in town. The second backgammon boom had begun.

As before a plethora of new books – again virtually all primers – appeared on the market. Some of them, like Bruce Becker’s “Backgammon for Blood” were truly awful, but at least it was a start.
In this boom a number of very bright young men, some ex chess masters, were attracted to the game and began to develop the theory of the game beyond the basics that had been around for centuries. For the first time doubling cube theory got serious attention.

The culmination of this effort was the production of “Backgammon” by Paul Magriel – the bible of the game and still the finest exposition of backgammon basics today –thirty years after its original publication.

Kent Goulding took the game another step forward when he began the practice of recording games and subjecting them to post-match analysis – a hitherto unexplored avenue of learning.
By the early 1980’s books began to appear that took our levels of learning to new heights and seminal works such as Bill Robertie’s “Advanced Backgammon” and Jeff Ward’s “Doubling Cube” were published.


The books became more sophisticated and our knowledge levels increased but there was a problem. In any given backgammon position how did we know which was the right play? For many years we took the word of the acknowledged experts of the day – after all they were the experts.
Then the bright young men of the 1970’s developed the concept of rollouts. Surely if we took a position and played the game to conclusion enough times (rotating the first roll to allow for the 36 possible rolls of the dice) we would get an idea of the ‘correct’ play?

Yes we would and this was certainly better than anything that had gone before. Rollouts significantly improved our understanding of the game. The problem was the accuracy. To get a reasonably accurate result (from a statistical perspective) required hundreds, if not thousands, of rollouts and we simply didn’t have the time.

Just when everyone was getting really frustrated computers came to the rescue. The PC had arrived just in time to assist in the development of the game. Early attempts to create a backgammon-playing computer had met with a modicum of success – in 1979 Hans Berliner created the ‘Gammonoid’ program that beat (luckily) the then world champion Luigi Villa. However that program ran on a mainframe computer and it would be another 13 years before the first PC program arrived.

This was Tom Weaver and Tom Johnson’s “Expert Backgammon” which ran on both Macintosh and PC. Whilst not great this program was a huge step forward and critically, not only could you play against the program, but it included a rollout facility. You could feed it a position and ask it to rollout (hundreds of times) your chosen moves. Gone were the long tedious hours spent rolling out positions by hand. Whilst we slept computers could do the work. This was a huge step forward for backgammon.

“Expert Backgammon” was based on the same type of theory as most chess programs. Take a position, look at all possible future moves, evaluate those resulting positions and select the best one using a predefined algorithm.

This produced reasonable results but the problem with backgammon is that after a couple of moves there are so many possible variations that even the most powerful computers can’t look far enough ahead (in a reasonable time) to give the ‘best move’ in any given position. This was frustrating but just when all seem lost a new technology arrived.

Neural Networks

In the IBM labs in White Plains, New York, a scientist named Gerry Tesauro was playing around with neural networks. Without going into the theory too deeply the idea was to produce a computer that could more accurately model the human brain in its working. The idea would be to give the computer the basics of a problem and then let it work out how to solve the problem rather than giving it clear directions.

For Tesauro backgammon was the perfect testing vehicle. He taught his program “TD-Gammon” the rules of backgammon and then told it to work out for itself the best strategy and tactics. It did this by playing half a million games against itself.

The end result was fascinating as some of the plays that humans had thought for centuries were correct were shown to be errors. That TD-Gammon was no mug was quickly proven when the best players of the day could only just show a slight edge when playing long sessions against it.
The game of backgammon had changed forever.


In this article we have covered the necessary historical context. In subsequent articles we will learn how to use modern day technology to enhance our understanding and enjoyment of the backgammon game.защитное сстеклокупитьBinaryнациональный антикоррупционный центр харьковhouseплита керамичекаяwheels markeрезина летняятатьяна шашникоаglobal translationпланшеты на андроид 4вышиванкаgirlsукраинына политологиenglishmanufacturing product life cyclethe towers key midtown-miamiпоследние новости из харьковкупить резину 225 65 16 летняя резинаноутбуки нысколько денег брать в грузиюпоследние новости об ани лоракблендерсайтskachat-point-blank.ruplayworldoftanks.ruскачать взломанный вормикс на андроидultrasurf rus.Netнакрутка ок в одноклассникахvk-editor.ruvzlom-minecraft.ruнакруткаХакерские программыбесплатновзломпрограмма шпионvkspyскачать бесплатно взлом одноклассниковбесплатночитатьчитыNeo OKs скачать бесплатноlink2sdвзломцентрTM binary options robotпинг тествозрождение

Leave a Reply

Your email address will not be published. Required fields are marked *