|
Backgammon Articles
|
Backgammon Computers History
1st World Backgammon Champion Loss to a Computer
The first landmark in backgammon computers history was in 1979, when BKG 9.8 beat newly crowned World Backgammon Champion Luigi Villa 7-1. BKG 9.8 was developed by computer scientist and former World Correspondence Chess Champion Hans Berliner, together with backgammon master and author Paul Magriel (X-22). Berliner, who took home $5,000 on the behalf of its creation, had later tried to consulate the beaten Villa by ascribing the backgammon computer's victory to fortunate dice rolls.
1st Champion Level Skill Backgammon Program
In backgammon, unlike pure skill games such as chess, a less skilled player, whether a computer program or a human being, can coincidentally outplay a superior player even a world champion. Thus, it took more than a decade until TD-Gammon, the first championship level backgammon program was developed. Created by Gerald Tesauro, a computer scientist of the IBM's Watson Research Center in New York, whose previous attempt in the backgammon computers field sent Neurogammon to win the 1989 International Computer Olympiad, TD-Gammon, was self taught using a learning technique called Temporal Difference, used in the fields of artificial intelligence and machine learning.
1st Internet Backgammon
Around the same time, early 1990s, the first internet backgammon server (FIBS) was launched, enabling, for the first time, real time backgammon games played by human players from all over the world, around the clock. FIBS has also established a rating system that ranked the players according to their wins and losses, based on the ELO rating formula adapted by world chess association (FIDE), allowed players to watch other player's games, record their games for study purposes, and basically broke the ground for all the backgammon rooms yet to come.
1st Commercial Backgammon Bots
Yet before 128 bit encryptions have made online backgammon for money conveniently secure, neural-net backgammon programs were released for commercial use. The first one was Jellyfish, which was capable of choosing the best possible play in every given position using equity calculations and performing reliable rollouts, as good as the best backgammon players around. Then, Snowie (known as SnowWhite at the beginning), free GNUbg and other well trained backgammon bots came along, all making backgammon matches, games and positions analysis accurate and systematic.
|
|
|
|
