[JLilly]

Hi al, I was playing with the robots and ran into a highly unusual pattern of bids in response to a 1NT opening. I wonder whether the robots were behaving aberrantly or the following is all according to 2/1 Hoyle. Also interested in analysis and criticism.



I was playing uncompetitively so I decided to midly lie to the robots about my holding (4d-5c rather than 5d-5c) to get some experience in playing Cappelletti. A 2c overcall would have been less descriptive, too.
Opener (W) doubles "for penalty", but the explanation in the window doesn't indicate anything other than the usual criteria for a 1NT opening bid. And opener (W) has poor defense in the minors (worthless doubleton and Axx) relative to their point-count.

Partner bids its better minor, which is doubled by the -1-points-holding passed hand:



I do not understand this bid by the robot at all. I promised 5 diamonds; partner selected diamonds (presumably 4+). Robot making the double has negative-one points and three diamonds.

Being fine with playing in 3D doubled, more than another contract, I pass. Opener (W, who bid the initial 1NT) bids 3H. This is the bid I would say I understand the least if I didn't understand the previous bid at all. They (W) have a four-card suit that will produce three (two if they're unlucky) tricks and would need a long suit in partner to get extra tricks. Partner doubles for penalty:



Contract for 3Hx made 7 tricks, so nice score for us.

Commentary and prescriptions for the robots are welcomed. Thanks!

[1eyedjack]

Yup, Cappelletti is broken. GIB should never have been programmed to play it. This forum is littered with far more mundane Cappelletti auctions when the robot commits seppuku.

There are other examples than Cappelletti where "popular" methods are shoehorned into the system without regard to the complex follow-up possibilities. The most common is the extended definition of a take-out double of one-suit opener that contains a multitude of hand types as a knock-on consequence of playing weak jump overcalls and UNT, and yet all follow-up continuations in a contested auction cater only for the possibility that doubler had the classic 3-suiter short in opener.

If you plough through this forum section you will find multiple examples of all of the above, and others.

[JLilly]

Thanks for the info! I guess I'll just have to play humans for more competition-over-1NT experience Would upvote but BBO won't let me.

1eyedjack, on 2017-February-19, 05:06, said:

Yup, Cappelletti is broken. GIB should never have been programmed to play it. This forum is littered with far more mundane Cappelletti auctions when the robot commits seppuku.

There are other examples than Cappelletti where "popular" methods are shoehorned into the system without regard to the complex follow-up possibilities. The most common is the extended definition of a take-out double of one-suit opener that contains a multitude of hand types as a knock-on consequence of playing weak jump overcalls and UNT, and yet all follow-up continuations in a contested auction cater only for the possibility that doubler had the classic 3-suiter short in opener.

If you plough through this forum section you will find multiple examples of all of the above, and others.