[Stephen Tu]

---

- IMO 2 spades should show 4S 3H. Makes continuation easier if hearts are denied. I think should also deny fifth spade? (5+ spades overcall to begin with or can jump now?) I suppose exception could be 5 spade 4H 18+ too strong for initial overcall, but initial assumption should be 4-3.
- The second cue bid should have higher lower limit since it creates a GF. Particularly after balancing double.
- Should it just bid 3nt itself after 2s, assuming 2s denies hearts?
- 3nt after the second cue bid should just show stopper IMO and not show extra values here like GIB apparently thinks it does.

- maybe it should consider overcall 1nt to begin with?

[lycier]

Stephen Tu, on 2017-January-15, 17:13, said:

- IMO 2 spades should show 4S 3H. Makes continuation easier if hearts are denied. I think should also deny fifth spade? (5+ spades overcall to begin with or can jump now?)
- The second cue bid should have higher lower limit since it creates a GF. Particularly after balancing double.
- Should it just bid 3nt itself after 2s, assuming 2s denies hearts?
- 3nt after the second cue bid should just show stopper IMO and not show extra values here like GIB apparently thinks it does.

Stephen Tu, what you said are really very very good, yes, yes, I agree. This is a normal human's routine in the most of situations.
Now please you would allow me to make my comments which I am afraid you usually would be very angry with me.
I think such sequence is just a Gib TPish bidding style.

1- Whenever Gib really holds 11+TPs, it always will cuebid opp's opening suit.
2- Then often Gib will try to show extra values on the second round - or bidding new suit (for example,only 4-card another major), or cuebid.

Why would Gib do so? I think Gibs need to make some mechanical TP calculation, orelse Gibs can't understand human's routine, just like what you said just now.
Now let's understand Gib well, I would take several examples.

Hand-1 My bais Gib's routine.

---

Hand-2 My experimental sequence.

---

Hand-3 My another experimental hand

---

Stephen Tu, on 2017-January-15, 17:13, said:

- maybe it should consider overcall 1nt to begin with?

Only for this hand, maybe it should do so, however I had ever encountered a hand, I overcalled 1N with 14hcp at four seat, my partner Gib did pass ,note it had 11hcp, so I lost game.

---

[Stephen Tu]

lycier, on 2017-January-15, 19:59, said:

Stephen Tu, what you said are really very very good, yes, yes, I agree. This is a normal human's routine in the most of situations.
Now please you would allow me to make my comments which I am afraid you usually would be very angry with me.
I think such sequence is just a Gib TPish bidding style.

1- Whenever Gib really holds 11+TPs, it always will cuebid opp's opening suit.

You waste a lot of time doing these things without gaining any understanding or making any useful observation for the group.

As has been explained multiple times, it has nothing to do with "TP bidding style". It is not a hand evaluation issue. It is a prioritizing of bidding rules issue.

Do you not agree that more than 1 bid can contain hands of the same TP range? Yes or no? For example if GIB open 1c it require the same 6+ to respond 1H or 1S, no? So how does it decide to bid 1h or 1s? It is not based on TP, it is based on shape. It follow the normal human ordering of bid longer suit, or H if 4-4 or spade if 5+ 5+. How does it know these rules? The programmers put in a lot of them and it just goes down the list matching the hand to first rule that fits. So it is easy for it to be changed to say jump to 2S instead of cue bid with inv values and spades but not hearts. You just put the rule for jump to 2s (4+ spades, 3- hearts, 10-12 TP or whatever) on the list higher than the cue bid. So it will prefer to bid 2s to cue bid since the single suited jump is higher priority like it should be. When it holds both majors it will cue since the jump to 2s denies heart length and it won't match the rule for bidding 2s.

So it is not the case that all hands of particular TP range must be lumped into one bid. When responding to takeout double, there are multiple bids that cater to 10/11+ TP: jumps in a new suit, NT bids, and the cue bid. It is not required that it always cue bid first! It might be currently programmed that way, but the point of my posts is that this is a mistake, and must be changed!! I don't know why you go on and on claiming that this is the way GIB bids and that GIB can't be changed here to bid in a more normal human fashion. It bids this way NOW, doesn't mean it has to bid this way forever.

If it holds invitational values (~10-12), basically it just has to use an ordered rule set:
- with only one major, jump in that major
- with neither major, if holding stoppers, bid 1nt with ~8-10/11 hcp, 2nt with 11/12-13. (11 is on the borderline should be upgraded/downgraded based on long suits/tens etc.)
- without stopper and without major, jump in the long other minor
- with both majors, cue bid.

with GF values, like this hand, it can reasonably cue, but only the first round.

Quote

2- Then often Gib will try to show extra values on the second round - or bidding new suit (for example,only 4-card another major), or cuebid.

Yes, and it should have properly ordered rules here too. Like not bid 4 cd major in suit partner bypassed. And bid some number of NT if appropriate as higher priority to cue bid a second time.

Quote

Why would Gib do so? I think Gibs need to make some mechanical TP calculation, orelse Gibs can't understand human's routine

It makes some TP calculation to determine how strong it thinks its hand is. But then it uses rules and the priority of those rules, which are based on BOTH TP and distribution to decide which bid to choose. There is nothing that forces it to cue bid as first priority with TP > 10, other than the broken order of the current rule set. If the programmers reordered the rules, it could be made to jump with one suiters in preference to cue bid, reserving cue bid for only both majors and 13+ TP hands. The problem has nothing to do with using TP to guage hand strength. No one is disputing on these particular deals how strong the hand is. It is the priority of choosing one bid over another. Single suited jumps should be higher in priority than cue bid. Cue bid is fine this hand since it has enough to GF. It just needs more and better defined rules for responding to the cue bid, and for the next round of bidding, and also for another round if it chooses to cue bid a 2nd time.

[m1cha]

Stephen Tu, on 2017-January-15, 17:13, said:

- IMO 2 spades should show 4S 3H. Makes continuation easier if hearts are denied. I think should also deny fifth spade? (5+ spades overcall to begin with or can jump now?) I suppose exception could be 5 spade 4H 18+ too strong for initial overcall, but initial assumption should be 4-3.
- The second cue bid should have higher lower limit since it creates a GF. Particularly after balancing double.

I agree with most of your post but I think you go wrong in one point. I seem to read from both of your posts that you think the second cue bid shows strength but that's not what it does. It shows length. I'm not saying it is right or wrong to do this; I'm just saying that what it does show, according to the label, is a genuing diamond suit ("4+ ♦") with no additional values (11+ TP) and it is not strictly GF but invitational because it is forcing to 3S, not 3N. I agree with you that 3S and 3N here is practically the same to you and me because South's double should deny 5 spades and it should be known to North that there is no spades fit; but since - as you pointed out - the double fails to deny 5 spades, the 3♦ bid is treated as invitational, GIB seems to consider 3♠ a possible final contract.

Stephen Tu, on 2017-January-15, 17:13, said:

- 3nt after the second cue bid should just show stopper IMO and not show extra values here like GIB apparently thinks it does.

If 3♦ means what the labes says it means, 3NT from South should show additional strength (though 16+ is ridiculous) and not necessarily a stopper. If 3♦ means what you say it should mean (and I'm not arguing against it), 3NT should just show a ♦ stopper.

Stephen Tu, on 2017-January-15, 17:13, said:

- Should it just bid 3nt itself after 2s, assuming 2s denies hearts?
- maybe it should consider overcall 1nt to begin with?

Yes, I would consider bidding 1NT in the first round and I would consider 3NT even more so in the second round. It's part of the problem that the requirements for 1NT in the first round and for 3NT in the second round are almost identical. GIB may have considered the hand a little bit weak but I rather believe that the real point behind the failure to bid NT here is - although it seems strange to us - that GIB does not consider JT982 a stop because it does not contain either of {AKQ}. This creates the paradoxical situation that GIB offers to play in this suit (by bidding 3♦) but is not willing to play NT against opponents bidding the suit (thus failing to bid 2 or 3NT).

[Stephen Tu]

m1cha, on 2017-January-16, 23:25, said:

I agree with most of your post but I think you go wrong in one point. I seem to read from both of your posts that you think the second cue bid shows strength but that's not what it does. It shows length. I'm not saying it is right or wrong to do this; I'm just saying that what it does show, according to the label, is a genuing diamond suit ("4+ ♦") with no additional values (11+ TP) and it is not strictly GF but invitational because it is forcing to 3S, not 3N.

It's absurd to use 3d as a natural inv only forcing bid. South has shortness, can't play in diamonds and it's forcing anyway, you have no spade or heart fit, so it's essentially forcing your side to play 3S in a 4-2/4-3 or 3nt understrength, both of which are terrible. If the rules were set up according to my list, any inv only hands without a spade fit would have bid something else than cue bid the first time. It would have chosen to bid one of 2h/1nt/2nt/3c. So 3d should only contain GF hands at this juncture, and those with length, most can bid NT unless something like xxxx in diamonds. 3d has to be a directionless GF, the current meaning simply must be changed. Then 3nt need not have extras.

Quote

Yes, I would consider bidding 1NT in the first round and I would consider 3NT even more so in the second round. It's part of the problem that the requirements for 1NT in the first round and for 3NT in the second round are almost identical.

From looking at my PC GIB, I think it wants 16-18 for a direct 1nt overcall, so 15 isn't enough. I guess it is OK with 5422 shapes for overcalling 1nt though it would never open 1nt holding that shape. Instead it goes for a gross 1d-1s-2d rebid, ugh.

Second round, simply 3nt has to be prioritized over a 2nd cue bid if it is a reasonable option. Or even 2nt, 2nt over 2s after cue bidding should arguably be GF if all inv hands could have done something else, vs. direct seat doubles. Maybe over a balancing double 2nt can be NF with 13-14 if we want to cater to staying low opposite a light 9 point reopening double.

Quote

GIB may have considered the hand a little bit weak but I rather believe that the real point behind the failure to bid NT here is - although it seems strange to us - that GIB does not consider JT982 a stop because it does not contain either of {AKQ}. This creates the paradoxical situation that GIB offers to play in this suit (by bidding 3♦) but is not willing to play NT against opponents bidding the suit (thus failing to bid 2 or 3NT).

No, if I strengthen the hand a point on my PC it will overcall 1nt treating the jt982 as a stop.

[slar]

Since a balancing NT is 11-14 HCP, 4th seat GiB should bid 2NT over 2♦. From there things are easy.

[Stephen Tu]

Rebidding 2nt over cue bid would be a really non-std way of bidding. The primary purpose of cue-bid is to locate 4-4 major fit. 2nt normally denies a 4 cd major on this sort of sequence. Most often advancer has some 4-4 in the majors and doesn't want to pick one, and then raises doubler's major to the appropriate level. That doesn't work so well if doubler is going to conceal 4 cd majors.

[helene_t]

Stephen Tu, on 2017-January-24, 09:25, said:

2nt normally denies a 4 cd major on this sort of sequence. Most often advancer has some 4-4 in the majors and doesn't want to pick one, and then raises doubler's major to the appropriate level. That doesn't work so well if doubler is going to conceal 4 cd majors.

I think standard is that
x-2♦
2NT*
shows a balanced hand too strong to bid 1NT initially. It doesn't deny a 4-card major.

Other than that I agree with your analysis.

[helene_t]

Stephen Tu, on 2017-January-15, 22:57, said:

You waste a lot of time doing these things without gaining any understanding or making any useful observation for the group.

No need to be so unfriendly Lycier's analysis may not address the OP issue but it is IMO a very useful clarification on how this auction actually works according to GIB.

[Stephen Tu]

helene_t, on 2017-January-24, 09:42, said:

I think standard is that
x-2♦
2NT*
shows a balanced hand too strong to bid 1NT initially. It doesn't deny a 4-card major.

Maybe over balancing seat double. Mike Lawrence says it doesn't apply after direct seat doubles + cue bid, he says just play 2nt as FG since doubler will be reasonably sound if doubling with no 4cd major (3325 I guess). Which makes sense since 18+ would be incredibly rare if advancer has a cue bid. Over balancing NT ranges, I guess it makes more sense to play it as the 15-17 or whatever range, but one could simply bid the same way as over direct seat, bid 4 cd M (which is F1 normally, cue bidder promises a rebid), bid game/3nt later depending on continuations.

helene_t, on 2017-January-24, 09:57, said:

No need to be so unfriendly Lycier's analysis may not address the OP issue but it is IMO a very useful clarification on how this auction actually works according to GIB.

Maybe you've missed the back and forth over the past year. I'm only unfriendly because of Lycier's persistent annoying habits, I'm not biting his head off just based on this one post. He continually clogs up threads posting lots of diagrams of different sequences using the same hand, which are usually completely irrelevant and should be opened in separate threads. (Choosing different bidding sequence, even if it is based on same deal, exposes entirely different bugs in different rules in the DB, and should be discussed separately IMO, given the forum's one bug per thread sticky request in "when reporting GIB's bugs", and criticizing human actions is irrelevant to whether a robot's bid is buggy if the human actually had their bid and the deal was different.) Also he inaccurately links many bugs (this and others), to GIB's use of TP evaluation, when the TP is totally irrelevant to GIB's poor bid choice. It is inaccurate to say that GIB's use of TP to gauge the strength of hands forces GIB to cue bid with all 10+ or 11+TP. More than one bid can share a TP range. Shape and stopper considerations and a prioritized rule set can make GIB choose other bids instead of cue bid. It cues with too many hands now, that doesn't mean auto-cue with 10+ is fundamental to GIB's bidding style and can't be changed. It is simply a bug in the current rule ordering. It's not "GIB uses TP, therefore fundamentally it must always cue with 11+ regardless of shape, humans must accept this" as Lycier has claimed.

[johnu]

Stephen Tu, on 2017-January-15, 17:13, said:

- IMO 2 spades should show 4S 3H. Makes continuation easier if hearts are denied. I think should also deny fifth spade? (5+ spades overcall to begin with or can jump now?) I suppose exception could be 5 spade 4H 18+ too strong for initial overcall, but initial assumption should be 4-3.

According to GIB descriptions in the hand diagram, 4(+) spades and 3(+) hearts is exactly what 2♠shows.

[johnu]

duplicate

[johnu]

slar, on 2017-January-24, 08:40, said:

Since a balancing NT is 11-14 HCP, 4th seat GiB should bid 2NT over 2♦. From there things are easy.

4th seat is a human, not GIB. Bidding 2NT over 2♦ is an awful bid with a quacky 11 count. As helene_t notes, this shows better than a 1NT balance so you would likely end up in slam opposite the North hand.

[johnu]

Stephen Tu, on 2017-January-17, 01:32, said:

Second round, simply 3nt has to be prioritized over a 2nd cue bid if it is a reasonable option.

3NT = 100
3♦ cue bid showing 4+ diamonds = -100
Other bids = 0

2nd cue bid is totally without purpose. Again, GIB has forgotten the previous rounds of bidding. 2♠ showed 9+ total points. I think it is 100% that 2♠ would not be bid with a 16-23 count, so how does 3NT now show 16-23, especially since there was an opening bid by the opponents and GIB N is looking at a 15 count.

Still, I have to give Stephen 100% of the blame for not checking the description of 3NT before bidding since we all know that GIB will be GIB.

[Stephen Tu]

johnu, on 2017-January-24, 12:04, said:

According to GIB descriptions in the hand diagram, 4(+) spades and 3(+) hearts is exactly what 2♠shows.

My argument is that it should remove the (+) part, assume South exactly 4-3 in the majors for the time being. 4-3 is different from 4(+)-3(+) in the DB. It affects followup actions if it has to cater to (+).

johnu, on 2017-January-24, 12:30, said:

Still, I have to give Stephen 100% of the blame for not checking the description of 3NT before bidding since we all know that GIB will be GIB.

I did check. The problem is I don't really have any other options that rate to work well. If I bid 3h on 3 I rate to get raised. If I bid 3s I may get raised on 3 also, since I didn't deny 5+. 4c bypasses the most likely game. I chose 3nt as least of evils, hoping that GIB wouldn't have extras and leave it in, not invite, or if that if GIB bid 4nt I could make it (which I did after opp GIB misdefended). But all this is entirely besides the point. The purpose of these posts is to bring attention to bugs for BBO staff to fix rules in the Gib database so that GIB chooses better bids (3nt instead of cue bid a 2nd time), and give humans better options (not have 3nt show extra values after the second GF cue bid). It isn't to second guess or blame humans for not finding the best non-std action to work around GIB's bugs. I'm generally pretty annoyed at criticizing human bids in this forum, when GIB has made clear errors, or bids with clearly poor definitions, and the human has bid reasonably normally by human stds, and the point of the post is to correct GIB bugs, not questioning best strategy to manage bidding with GIB. Even if human has made egregious errors, in my mind if GIB's bids are still bad (assuming human actually had a hand fitting his own bid), I think we should generally ignore those and focus on fixing GIB. I only criticize human bids in this forum section if the GIB actions are normal for human adv/expert and the human has very non-std expectations of what bids ought to mean, and there is arguably no bug present in the auction in question.

[slar]

johnu, on 2017-January-24, 12:14, said:

4th seat is a human, not GIB. Bidding 2NT over 2♦ is an awful bid with a quacky 11 count. As helene_t notes, this shows better than a 1NT balance so you would likely end up in slam opposite the North hand.

Okay, I completely read this wrong. According to the card, GIB can make a 1NT overcall with 15-18 HCP and a stopper. If this is done then there are no issues. It is not clear to me why GIB did not make this bid in the first place. After that the continuations are easy.

[johnu]

Stephen Tu, on 2017-January-24, 13:30, said:

I did check. The problem is I don't really have any other options that rate to work well. If I bid 3h on 3 I rate to get raised. If I bid 3s I may get raised on 3 also, since I didn't deny 5+. 4c bypasses the most likely game. I chose 3nt as least of evils, hoping that GIB wouldn't have extras and leave it in, not invite, or if that if GIB bid 4nt I could make it (which I did after opp GIB misdefended). But all this is entirely besides the point.

All true, but then I still have to give you 100% for playing with GIB