[shyams]

Link to hand: http://tinyurl.com/hm8w4ma
Note: Despite the diagram showing me as North, this is a "human declares" tournament. I was South, switched to North for the play of the hand.

I assume GIB deals sample hands at each stage that match the bidding to date. If so, how many "sample hands" can GIB deal where it fails to count 13 cold tricks as North?

There are 13 cold tricks in almost all configurations. Even a 4-0 split with West holding all trumps is not fatal. In essence, we are talking about fewer than 1 in 50 "sample deals" where the grand will fail whereas in all others, 7♠ is cold.

The reason for my post is to get a sense (if possible) as to why GIB would not consider bidding a grand slam here.

[1eyedjack]

shyams, on 2016-July-23, 12:22, said:

I assume GIB deals sample hands at each stage that match the bidding to date.

I am not certain, but I think that this assumption may unwarranted, although I do not believe that it is important to your point. I think that there are some auctions where, particularly in the early stages, it works on a pure rule-based approach.

That aside, there is another relevant principle to add into the mix. 5N following a Blackwood 4N is a consultative bid, promising all key cards. Responder to Blackwood may be able to count 13 tricks in that knowledge, and if he gives a King-count in response there is an inference that he has concerns. It is perhaps arguable that with such solidity in Hearts you *might* have bid 7 yourself. Personally I would have chickened out. But then accurate GS bidding seems to me a low priority for resources. Maybe I am too pessimistic.

[shyams]

I disagree that South has enough information to bid a grand slam. I can only count 11 or 12 tricks from the South side. Even if hearts are pulled in for 5, I might still have a diamond loser.

As far as North is concerned, South has shown: ♦A, ♥A, ♥K, ♠K, and a 4-5-x-x (OR 4-6-x-x) distribution.

My point was simple: How often (out of 50 or 100 sample deals) will GIB find a hand that yield only 12 tricks instead of all 13 tricks? If GIB uses deal sampling to make decisions, this hand is a test for its ability to rely on the outcome of such sampling.

[johnu]

North has a clearcut 7♠ bid as the club loser can be pitched on the ♥K and you just have to set up diamonds for no losers (or ruff 2 small diamonds) and avoid ♠J9xx offside.

South should just jump to 7♠ after the 5NT bid as ace third of clubs and minimal honor cards will probably yield 13 tricks. GIB has showed a very good hand on this sequence.

[lycier]

shyams, on 2016-July-23, 12:22, said:

Link to hand: http://tinyurl.com/hm8w4ma
Note: Despite the diagram showing me as North, this is a "human declares" tournament. I was South, switched to North for the play of the hand.

I assume GIB deals sample hands at each stage that match the bidding to date. If so, how many "sample hands" can GIB deal where it fails to count 13 cold tricks as North?

There are 13 cold tricks in almost all configurations. Even a 4-0 split with West holding all trumps is not fatal. In essence, we are talking about fewer than 1 in 50 "sample deals" where the grand will fail whereas in all others, 7♠ is cold.

The reason for my post is to get a sense (if possible) as to why GIB would not consider bidding a grand slam here.

shyams, on 2016-July-23, 14:13, said:

I disagree that South has enough information to bid a grand slam. I can only count 11 or 12 tricks from the South side. Even if hearts are pulled in for 5, I might still have a diamond loser.

As far as North is concerned, South has shown: ♦A, ♥A, ♥K, ♠K, and a 4-5-x-x (OR 4-6-x-x) distribution.

My point was simple: How often (out of 50 or 100 sample deals) will GIB find a hand that yield only 12 tricks instead of all 13 tricks? If GIB uses deal sampling to make decisions, this hand is a test for its ability to rely on the outcome of such sampling.

Hi

Good hand example and good topic.

1- Even so, however I never agree with your opinion partly.
The main reason is I think you should rebid 6♣ to show specific ♣K after 5N instead of 6♥ showing ♥K.
Any ideas?

2- If there is a correct Gib CC to show specific King, it's very easy to bid up to grand slam for sure. I think there is a serious issue on how to show specific King correctly after 5N on Gib CC. Now please see my example hands.

---

Result : 6♠N+1
Note it's me to make rebid-6♦ because North Gib want to return 6♠ directly after 6♣.
Even thru 6♣ - 6♦ - 6♥ sequences, North Gib still returns to 6♠, and at the meanwhile, 6♥ says wrongly " Cuebid ---5+♥,4+♠,♣K,♦A,♥A,16-18TPs,forcing", actually rebid-6♥ should show ♥K !
So the programmers need to improve showing specfic King after 5N on Gib CC.

3- Now let's compare with two hands.

Your hand :

---

Basic Gib hand:

---

My conclusion : Even there is a serious issue on your hand and there is no obvious issue on basic Gib hand, your hand played by advanced Gib is better than basic Gib hand,

[lycier]

johnu, on 2016-July-23, 17:16, said:

North has a clearcut 7♠ bid as the club loser can be pitched on the ♥K and you just have to set up diamonds for no losers (or ruff 2 small diamonds) and avoid ♠J9xx offside.

South should just jump to 7♠ after the 5NT bid as ace third of clubs and minimal honor cards will probably yield 13 tricks. GIB has showed a very good hand on this sequence.

Your opinions are a worst method since 5N says " Don't promise extras " at first, so how would you think " Gib has showed a very good hand on this sequence."?
Illogic thinking.
On this hand, the exact issue mainly comes from worse Gib CC instead of wrong judgement.

[lycier]

And your opinion never help to resolve any issue.

[sfi]

lycier, on 2016-July-23, 18:44, said:

And your opinion never help to resolve any issue.

Johnu was spot on in his assessment of the mistakes in evaluation on both sides, as well as the fact that North has shown a very good hand on this sequence. Your solution was to cue bid a non-existent king, which didn't seem to work anyway.

Your insults are unwarranted.

[johnu]

lycier, on 2016-July-23, 18:42, said:

Your opinions are a worst method since 5N says " Don't promise extras " at first, so how would you think " Gib has showed a very good hand on this sequence."?
Illogic thinking.
On this hand, the exact issue mainly comes from worse Gib CC instead of wrong judgement.

By cue bidding, GIB showed 14+ total points. By bidding Blackwood, GIB showed 16+ total points. In what world is 16+ total points not a very good hand after partner shows 16-18 total points with 4+ card trump support??? 5NT shows all the keycards and the queen of trump. Rethink your faulty analysis.

[johnu]

lycier, on 2016-July-23, 18:44, said:

And your opinion never help to resolve any issue.

And if I want your opinion, I'll ask

[1eyedjack]

lycier, on 2016-July-23, 18:42, said:

Your opinions are a worst method since 5N says " Don't promise extras "

True, 5N doesn't promise extras. 4N does

[shyams]

IMO, we are getting distracted by South and South's actions. This has nothing to do with North's decision.

North is GIB - we are told, GIB does not use human logic for decision-making. Instead it deals a sample list of hands that fit the bidding so far, then decides a course of action.
In that case, my point is quite clear. In order for North to justify its 6♠ bid, its sample should have hands where 13 is not a logical outcome. How many hands can be randomly dealt where GIB finds it cannot make 13 tricks?

[1eyedjack]

shyams, on 2016-July-24, 03:20, said:

IMO, we are getting distracted by South and South's actions. This has nothing to do with North's decision.

You may be right, but I am not convinced, at least in principle. If North is running sims, it would be entirely justified in excluding those hands in which it judges that, in its view, South should have jumped to 7 over 5N

[shyams]

1eyedjack, on 2016-July-24, 03:56, said:

You may be right, but I am not convinced, at least in principle. If North is running sims, it would be entirely justified in excluding those hands in which it judges that, in its view, South should have jumped to 7 over 5N

You are right, of course. My fault entirely for raising this topic.
Thank you for clarifying.... much appreciated.

[iandayre]

I agree that this hand should have been bid to a grand slam. I have posted a few hands in the past where GIB could count 13 certain tricks in NT after the 5NT response, but failed to bid the grand. Now that I think about it, I have never seen GIB bid 5NT, receive a favorable response, and go on to the 7 level. It appears that the programmers have made GIB's 5NT call a pointless exercise of telling the opponents what our side has, without ever using that information for our benefit. Can anyone remember, and preferably post, a hand where GIB bid a grand slam after having bid 5NT?

[Bbradley62]

1eyedjack, on 2016-July-24, 03:56, said:

You may be right, but I am not convinced, at least in principle. If North is running sims, it would be entirely justified in excluding those hands in which it judges that, in its view, South should have jumped to 7 over 5N.

Even if we exclude such hands, I think OP's point is correct that the overwhelming majority of "includable" hands would allow North to count 13 tricks. Worst case should be: ♠Kxxx, ♥AKxxx, ♦A, ♣Jxx. Even that simulation has some play; most would be cold.

[Bbradley62]

shyams, on 2016-July-24, 03:20, said:

IMO, we are getting distracted by South and South's actions. This has nothing to do with North's decision.

I completely agree with you about who can and cannot bid 7. However, you now see why some of us have learned to take the extra couple of minutes to edit the hand record to show only the hand being discussed. Some members of the peanut gallery just can't help themselves.

[Bbradley62]

sfi, on 2016-July-23, 18:52, said:

Your solution was to cue bid a non-existent king, which didn't seem to work anyway.

As you can see, I have upvoted your post. However, it may be (input from staff would be helpful here) that GIB interprets 6♣ as second-round club control (assuming spades as trump) and 6♥ as denying second round club control. Of course, even if this is the case, "lying" and bidding 6♥ still might be better on this hand that also includes ♥q, but I don't think it's clear whether 6♣ shows a non-existent king.

[lycier]

I am very surprised and very pity to see that Bbradley62 opinion looks like a expert, Jhon and Sfi, even as programming staff, just look like beginner.
Actually, how simply issue !
Obviously, there are some issues on showing specific king of Gib CC, only !
I am wondering why it is tough for programming experts to analysis it, even very difficult to get correct conclusion.

[johnu]

Bbradley62, on 2016-July-24, 15:28, said:

However, it may be (input from staff would be helpful here) that GIB interprets 6♣ as second-round club control (assuming spades as trump) and 6♥ as denying second round club control. Of course, even if this is the case, "lying" and bidding 6♥ still might be better on this hand that also includes ♥q, but I don't think it's clear whether 6♣ shows a non-existent king.

Talking about 6♣ isn't even a red herring, at best it is a peripheral question on this hand since GIB N has everything it needs from the 6♥ response.

That being said, showing a singleton club could work on some hands, as long as partner doesn't think it is a king and put you in an unplayable NT slam. Maybe that's why, AFAIK, nobody shows singletons in response to a bulk ace asking bid, eg. spiral scan, RKC, Gerber, etc.