[1eyedjack]

MP, robot tourney, best hand South.

---

Trick 1: ♣2, ♣3, ♣K, ♣9
Trick 2: ♠3, ♠A, ♠6, ♠4
Claim 12 (effectively).

Before anyone piles in with the comment that how can GIB possibly be expected to get this right against a psychic opener, may I just point out that swap the ♠2 with a red J and South then has a systemic 2N opener (although I grant not a 3♦ rebid).

OK, perhaps denying a Major in response to Stayman compounded the confusion. Actually one other player outscored me: Idendical auction except that in response to Stayman opener admitted to having a 4 card Spade suit. (Why??). Anyway, the effect of that was to elicit ♦3 lead and an overtrick. Not fair, of course - East was never given an opportunity to shine by cashing the setting trick.

GIB places 100% trust in every bid at the table being within defined parameters (at least until that cannot be possible). Rightly so in the case of partner's bids, and I acknowledge that it will be hard to program GIB to think otherwise of opponents' bids.

But I don't think that any human player sitting East would find the same defence, even if he decided to trust South.
South can in fact just about have a hand consistent with the bidding:
♠AKQ
♥AJx
♦QJxx
♣QJ9

While it is true that a ♠ switch at trick 2 then costs nothing neither does it gain anything. It just prolongs the agony.
Even if there was a prospect for a ♠ switch to set up a second undertrick, in that event you should assume that the rest of the room will not be in 6N and simply taking it one off will net nearly all the matchpoints.
What is more, if South has 3-3-5-2 shape, you could be taking the first 5 tricks in defence ... *IF* you continue with ♣AT at tricks 2 and 3.

Is it unreasonable to expect GIB to be programmed to cash the setting trick at trick 2? I would sincerely hope that it would do so at IMP, but I cannot test that.