Good, Better, Best: A comparison of bridge bidding systems and conventions by computer simulation

Bridge players constantly argue about the 'best' bidding systems and conventions. Strong Club or Natural? With or without relays? Standard or 2/1? Weak or strong notrump? Was the Blue Club system really better than the natural systems in the 1950s and 1960s? Is two-way Stayman as good as Stayman and transfers? Disciplined or undisciplined weak twos -- which work better? And many, many more...

Well, now we have the answer to those questions, provided by exhaustive analysis and the results of computer simulations. And not only which is better, but also by how much, typically tested in thousands of 24-board IMP matches.

Playing the same deals via the same AI software eliminates the variable of player skill, and bidding methods are judged solely on their own merits. It is as objective a test as modern technology can provide.

And the results may not be what you would expect.

Jan Eric Larsson (Sweden) is a professor in artificial intelligence at Lund University in southern Sweden, and manager of a small software company that delivers intelligent monitoring and diagnosis systems for national power grids. He is also one of Sweden's top ten tournament directors and the chairman of the Swedish Bridge Laws Commission.