The robots make surprising errors

George Eliot wrote, “Prophecy is the most gratuitous form of error.”

My wife and I play a fair amount of bridge against the robots at Bridge Base Online. Trying to prophesy what the robots will do is a low-percentage exercise.

Take today’s deal as an example. What did West lead against four spades? What did East discard on that trick?

The bidding was not easy. I would have been inclined to rebid two notrump with the North hand. (Many of the human Norths rebid only two diamonds. Then South jumped to three spades, and North raised to game.) Over three diamonds, I was torn. Did we have a slam in diamonds, or was game the limit? And if we could not make a slam, was four spades better than five diamonds?

Eventually, I rebid three spades, partner raised, and I passed with misgivings.

The “normal” lead by West is surely either the diamond 10 or club ace. He expects to win two trump tricks and the club ace. The fourth trick might be a diamond ruff; and perhaps after leading the club ace, seeing the dummy will help. Of the 10 robots with this problem, four chose the club ace, three the diamond 10, one the heart eight, and one, my opponent, the strange spade six.

To make the contract, in theory I had to win with dummy’s jack and abandon trumps. But never guessing that West had led away from the spade queen, I played low from the board. Now East rode to my rescue, discarding a diamond! Why are computers not taught to keep equal length with the board?

So I lost only two spades and one club.