When will a robot beat the best human player?


Here is a comment from Hans Leber, one of the authors of Q-Plus Bridge, about the question if a robot will ever beat the best human player.

"I do not see principle obstacles. What has been regarded as difficult, for example, in defense drawing conclusions about declarer's hand from the way he proceeds, or as declarer playing in a deceptive way, is not beyond the scope of a program.

It is perhaps true that there might be a level of human imagination which is difficult to match for a machine, but then not all imaginative actions are successful, and it might be compensated by the strengths of a machine (no problems with concentration and emotions).

Also 'judgement' is not a problem because this is a result of the simulation (which means dealing many possible hands and then calculating the outcome of various actions).The biggest problem is that top players have a large vocabulatory in the bidding (and similarly, but to a lesser extent, in defense play), i.e. they attach sensible meanings to bids in many - thousands - bidding sequences. If a computer program wants to get to the top level, it is necessary to match that, and this means a lot of work to do.

As a quantitative estimate: While about 20.000 hours of work (in the case of Q-plus Bridge) were sufficient to beat the first 96%-98% of the (regular) bridge players, about 30.000 more hours would be necessary to reach the top players "

(This is about mainstream thinking among bridge program authors. Only Tomio Uchida [Micro Bridge] doubts that bridge programs will ever match the best human players. Ian Trackman and Mike Whittaker [Blue Chip Bridge] think that the estimate is about right; Hans Kuijf [Jack] and Yves Costel [WBridge5] believe that not so much effort is needed, but agree that much is still to be done.)

While the development will continue in principle it is unclear if and when the needed amount will be reached.It might sound surprising that so much extra effort is needed, but this also reflects the fact that the difference between human top players and average players is huge - it is no accident that on championships 'always' (i.e. during periods of ten years) the same names appear on the top, which - as a side note - is also a strong indication that bridge is not a game of luck!