[Joeydinky]

HI All,
I played two games this morning and had 3 impossible bids by GIB resulting in bad scores. One hand was game 9048, hand #1. I opened 2C - partner bid 2D, I bid 2S, pard bid 3H. I jumped to 4S and GIB bids 6NT with no D stoppers(2 little). How can they be programmed to bid that way? Was down 4.

Another bad bid was game 9496, hand #7. I opened 1NT with Ax, AKQ9x, Kxx,9xx Pard bid 2H transfer to S. I bid 2S. Then Gib bids 3D, a game force, with 5 HCPs and only 2Ds. His hand was KJxxx, J10xx, xx, xx. WHY? Makes no sense.

Game 9496 hand #11 I opened 1NT with QJxx, KQxx, KQx, Qx. They overcall 2C (single suiter)- P - 2H (own suit). I passed and it went P by the 2C bidder. Now Gib bids 2Ss with 9xxx, 10xx, 9xx, Axx. We get doubled. Again WHY?

I thought GIB had been imrpoved. I have not found that to be true. HELP to keep this from happening again. Thanks

[Bbradley62]

Welcome to the Forums, Joey.

---

[Bbradley62]

Joeydinky, on 2012-April-09, 09:00, said:

Another bad bid was game 9496, hand #7. I opened 1NT with Ax, AKQ9x, Kxx,9xx Pard bid 2H transfer to S. I bid 2S. Then Gib bids 3D, a game force, with 5 HCPs and only 2Ds. His hand was KJxxx, J10xx, xx, xx. WHY? Makes no sense.

I think you misread this auction; it was not your partner who bid 3♦...

---

[Bbradley62]

Joeydinky, on 2012-April-09, 09:00, said:

Game 9496 hand #11 I opened 1NT with QJxx, KQxx, KQx, Qx. They overcall 2C (single suiter)- P - 2H (own suit). I passed and it went P by the 2C bidder. Now Gib bids 2Ss with 9xxx, 10xx, 9xx, Axx. We get doubled. Again WHY?

---

Edited to add this table with a different auction...

---

This post has been edited by Bbradley62: 2012-April-09, 11:38

[Stephen Tu]

I think you are being a bit tough on GIB. I've played with it a lot over the past couple weeks, and definitely find it to be making far fewer nutty bids than it used to a couple years ago. Bidding is really complex, there are a zillions of possible sequences, it's hard for it not to occasionally go haywire on certain sequences. Just report bugs (preferably one per post, not 3 like here), the bots will be fixed eventually if it's a real bug, and can be addressed without too much difficulty.


First complaint: 6nt with no stopper. GIB holds approximately x AJxxx Qx Kxxxx, you had AKQJx K Jx AQxx After your sequence, 2c-2d-2s-3h-4s, really it's kind of stuck. It has to guess between pass, 6s, 6nt. On most hands, it probably expects you to be a bit stronger (your hand is borderline for a 2c opener, IMO), and you'll hold DK or DA way more often than not. Would you be happy missing the slam if your HK was the DK instead? If you had opened 1s, or maybe not jumped to 4s over 3h, or maybe jumped to 3s over 2d demanding cue bids, perhaps the auction goes differently, though probably still GIB has some trouble interpreting cue-bidding inferences.

Second complaint: rebid 3d with 5422 5 count. This is your error here, not GIBs. If you look at the hand record closely, it is your *opponent* who balanced with 3d, not your partner. Partner passed the 2s, 4th hand bid 3d, you bid 5d (which is bad even if partner was the one to bid 3d).

Third board: Well GIB probably thought it could avoid the double in most instances, and really it probably thinks it makes 2s a reasonable fraction of the time, and drives them higher a lot. It's not a great bid, though it might work out more often than you think it should at matchpoints. Remember at matchpoints, getting out for two undoubled for -100 instead of letting opponents make -110 or -140 is an equal triumph to even out the disaster of -300. If it achieves -50/-100 more often than -300 it's arguably the right bid. I think it should pass here, but this is probably one of the harder sequences to program. Maybe it should require shorter hearts to balance.

[barmar]

Stephen Tu, on 2012-April-09, 11:05, said:

First complaint: 6nt with no stopper. GIB holds approximately x AJxxx Qx Kxxxx, you had AKQJx K Jx AQxx After your sequence, 2c-2d-2s-3h-4s, really it's kind of stuck. It has to guess between pass, 6s, 6nt. On most hands, it probably expects you to be a bit stronger (your hand is borderline for a 2c opener, IMO), and you'll hold DK or DA way more often than not. Would you be happy missing the slam if your HK was the DK instead? If you had opened 1s, or maybe not jumped to 4s over 3h, or maybe jumped to 3s over 2d demanding cue bids, perhaps the auction goes differently, though probably still GIB has some trouble interpreting cue-bidding inferences.

Unfortunately, GIB doesn't know about the treatment where jumping to 3♠ sets trumps and demands cue bidding. I don't think it considers 2♣...2♠...3♠ to be much different from 2♣...3♠.

This is probably something we could add, but I don't consider it high priority.

Quote

Third board: Well GIB probably thought it could avoid the double in most instances, and really it probably thinks it makes 2s a reasonable fraction of the time, and drives them higher a lot. It's not a great bid, though it might work out more often than you think it should at matchpoints. Remember at matchpoints, getting out for two undoubled for -100 instead of letting opponents make -110 or -140 is an equal triumph to even out the disaster of -300. If it achieves -50/-100 more often than -300 it's arguably the right bid. I think it should pass here, but this is probably one of the harder sequences to program. Maybe it should require shorter hearts to balance.

GIB does tend to be pretty aggressive at balancing in these auctions. I haven't looked at the rules involved, but I suspect simulations are to blame.

[Leo LaSota]

Bbradley62, on 2012-April-09, 10:37, said:

Welcome to the Forums, Joey.

---

An interesting point to consider on the 6nt hand for JoeyDinky is that in live bridge, many players play "fast arrival" where the jump to 4S shows a minimum given what has been shown with the hand thusfar. While I probably would open just 1S, I can understand the thinking was that there are only 4 losers (Losing Trick Count), so they decided to open 2C. Also, they probably thought jumping to 4S should be a sign off showing a minimum 2C opener.