Computer bridge

In comparison to [[computer chess], Computer bridge is in its infancy. After many years of limited progress, at the end of the twentieth century computer bridge made big strides forward. In 1996 the American Contract Bridge League (ACBL) initiated official World Championships Computer Bridge, to be held annually along with a major bridge event. The first Computer Bridge Championship took place in 1997 at the North American Bridge Championships in Albuquerque.

World Champions - Computer Bridge

• 1997 Bridge Baron
• 1998 GIB
• 1999 GIB
• 2000 Meadowlark Bridge
• 2001 Jack
• 2002 Jack
• 2003 Jack
• 2004 Jack
• 2005 Wbridge5
• 2006 Jack

Computers versus humans

Strong bridge playing programs such as Jack probably rank among the top few thousand human pairs worldwide. A series of articles published in 2005 and 2006 in the Dutch bridge magazine IMP describes matches between Jack and seven top Dutch pairs. A total of 196 boards were played. Overall, the program Jack lost, but with a small margin (359 versus 385 imps).

Computer play algorithms

Bridge poses different challenges to its players than board games such as chess and go. Most notably, bridge is a probabilistic game: a bridge player has to decide which bid to make and which card to play based on incomplete information. At the start of a game, the information available to each player are just his/her own cards. During the bidding stage and the subsequent play stage, more information becomes available via the bidding of the other three players at the table, the cards of the partner of the declarer (the dummy) being put open on the table, and the cards played at each trick. However, only at the end of the game full information is obtained.

Today's top=level bridge programs deal with this probabilistic nature of bridge by generating many samples representing the unknown hands. Each sample is generated at random, but constrained to be compatible with all information available from bidding and play. Next, the result of different lines of play are tested against optimal defense for each sample. This testing is done utilising a so-called double dummy solver that via extensive seach algorithms determines the optimum line of play for both parties. The line of play that generates the best score averaged over all samples is selected as optimal play.

Obviously, the amount of computation increases with sample size and techniques such as importance sampling are used to increase the computational efficiency.

Websites

• GIB
• Jack (in Dutch)
• WBridge5 (in French)