[jdonn]

steve2005, on 2015-June-14, 07:25, said:

GIB used to miscount losers in response to blackwood, causing overbidding such as grand slams off the queen of trumps or even off a keycard. That will no longer happen.

Bug not fixed see link to forum post below

http://bridgebase.co...wn-cashing-ace/

That is a different bug. The bug that was fixed was something like 4NT 5H (no queen of trumps) 7S (grand slam anyway). The one you linked to has to do with overruling partner after he picked a strain other than the suit that blackwood was bid for. It's something on our radars that we are working on.

[jdonn]

baraka, on 2015-June-15, 11:13, said:

And here is an other one that's never been fixed...

---

Can anyone push GIB any harder to bid 3NT with Diamond stoppers ? No ! Yet he bids 4D instead of 3NT. What is the use of that 4D bid when 3NT is bypassed ? Am I to bid 4NT to play? GEEZZZ !?!?!?!?

We got your bug report on that and it's in Georgi's queue. Thanks for reporting it. Agreed that it's a major problem.

[jdonn]

baraka, on 2015-June-15, 11:28, said:

And again...

---

My 2nd bid is pass signaling minimum. So, when I bid 2D GIB should realise I'm just competing. What's the point of bidding 2NT ? Why not pass or bid 3D ? He's got 4 of them and knows I'm gonna be short clubs and he has no C stoppers ?!?!?! Exactly the opposite of the previous hand. You want him to bid NT, he does'nt ! You dont want him to bid NT, he does !

I agree this is a bug as well. The real problem is that I see for the 2D bid, GIB thinks that shows fewer than 5 diamonds, max of 4. It decided that with an 8 card fit, notrump would play better, which seems to me like a very reasonable judgment looking at his hand (based on what it understood the bids to mean). I think this particular bid, pass then 2D in this situation, should show exactly 5 since with fewer you wouldn't bid it and with more you wouldn't pass 1NT. I'll sent a note to Georgi. Thanks

[iandayre]

jdonn, on 2015-June-16, 12:41, said:

I agree this is a bug as well. The real problem is that I see for the 2D bid, GIB thinks that shows fewer than 5 diamonds, max of 4. It decided that with an 8 card fit, notrump would play better, which seems to me like a very reasonable judgment looking at his hand (based on what it understood the bids to mean). I think this particular bid, pass then 2D in this situation, should show exactly 5 since with fewer you wouldn't bid it and with more you wouldn't pass 1NT. I'll sent a note to Georgi. Thanks

It's a very, very common GIB issue that the descriptions of bids indicate shorter length than normal bridge practice would indicate. If this were fixed, I suspect that GIB would be introducing new 3 and 4 card suits at high levels much less frequently than it currently does.

[baraka]

Another one bites the dust !

---

To those who want to gamble 3NT with ♠Q4 against a known opp 8 card fit and knowing that partner will be short... I say good for you, but that is not really my point here. Against an opp known 5-5 in the majors and a known 7+ card fit in one major, partner is most likely not to have 4-4 in the majors. Therefore, partner has 4+ ♦'s. If I had wanted to play 5♦ in front of a minimum hand, I would have bid 5♦ myself. Therefore, my 4♦ bid is at best invitational if not for pure competition. Why on earth is GIB bidding 5♦ on his bare minimum hand is beyond me ! Please correct !

[1eyedjack]

Personally I think that 4D is an overbid. I agree that 3N is not an attractive alternative. But Pass is. Maybe X at MP.

[Bbradley62]

GIB does not understand the concept of a purely competitive bid. (Well, maybe he's willing to make such a bid, if simulations indicate that he expects to go down less than his opponents would score declaring, but he doesn't ever think his partner has just done exactly that.) This is an often-reported and system-wide problem. Note that the description of your 4♦ bid does not have an upper limit.

[jdonn]

1eyedjack, on 2015-June-20, 12:47, said:

Personally I think that 4D is an overbid. I agree that 3N is not an attractive alternative. But Pass is. Maybe X at MP.

From a pure bridge perspective (meaning not adjusting due to playing with GIB) I think double is completely automatic. We have game forcing strength, no stopper to bid 3NT, and have already shown the diamond support. But of course, that has nothing to do with the meaning of 4♦, which yes should definitely be just competitive.

I think the source of this particular bug is that it was widespread when GIB was first written that to play on the 4 level we need like 25+ combined total points, and the 5 level needs 29+. That's why for so long, bidding 5 of a minor to play was taken as a slam try. Unfortunately, there is no switch to fix all of these at once. They are just being fixed piece by piece, which is not a fast process.

[1eyedjack]

jdonn, on 2015-June-20, 13:45, said:

From a pure bridge perspective (meaning not adjusting due to playing with GIB) I think double is completely automatic. We have game forcing strength, no stopper to bid 3NT, and have already shown the diamond support. But of course, that has nothing to do with the meaning of 4♦, which yes should definitely be just competitive.

I think the source of this particular bug is that it was widespread when GIB was first written that to play on the 4 level we need like 25+ combined total points, and the 5 level needs 29+. That's why for so long, bidding 5 of a minor to play was taken as a slam try. Unfortunately, there is no switch to fix all of these at once. They are just being fixed piece by piece, which is not a fast process.

I am not sure that Qx of Spades is worth taking into consideration, and without it I would not regard the hand as GF strength. Maybe initially, but in light of the subsequent auction to date I would rather have that holding in defence. On balance I still think that the initial 2H cue says it all.

[baraka]

Looks like GIB still doesn't get it...

---

You need Aces to bid 6 in this situation ! Not some stupid Quacks ! Even less a Q in opp's suit ! 4Hearts makes easy. I have to save. Why does GIB kills me every time ?

Why is it that GIB has such a hard time evaluating an honour in opp's suit like a singleton K of clubs after the opp's go 1 club - 3 clubs ? That was one I got once.

[Bbradley62]

baraka, on 2015-July-28, 08:05, said:

Why does GIB kills me every time ?

GIB does not ever consider the possibility that partner is sacrificing. Note the explanation of 5♣. GIB thinks you are announcing 29 total points, enough to make 5♣ opposite a yarb with small doubleton support. Of course, you're very nearly there as it is...

[baraka]

Bbradley62, on 2015-July-28, 18:21, said:

GIB does not ever consider the possibility that partner is sacrificing. Note the explanation of 5♣. GIB thinks you are announcing 29 total points, enough to make 5♣ opposite a yarb with small doubleton support. Of course, you're very nearly there as it is...

It should not be that hard to fix, at least in part. Both opps bid hearts. GIB has 3. Therefore I have at most 1. What will Qxx in opp's suit bring me ? Nada ! That Q should be set at a big fat zero in value and proceed from there. It might be simplistic but not so far off. It's just like that stiff K of clubs when opps both bid and raised clubs. A big fat zero ! It's called duplication of value. Bridge is a game of odds. Every time GIB detects a duplication of value he should down grade because the odds got lowered.

[steve2005]

baraka, on 2015-July-28, 19:20, said:

It should not be that hard to fix, at least in part. Both opps bid hearts. GIB has 3. Therefore I have at most 1. What will Qxx in opp's suit bring me ? Nada ! That Q should be set at a big fat zero in value and proceed from there. It might be simplistic but not so far off. It's just like that stiff K of clubs when opps both bid and raised clubs. A big fat zero ! It's called duplication of value. Bridge is a game of odds. Every time GIB detects a duplication of value he should down grade because the odds got lowered.

have to agree completely, hand evaluation is not a strong point of Gib. What really kills me is Gib gives points for shortness in partner's suit, I think even in NT.

[Bbradley62]

steve2005, on 2015-July-28, 20:20, said:

have to agree completely, hand evaluation is not a strong point of Gib. What really kills me is Gib gives points for shortness in partner's suit, I think even in NT.

We have been told that GIB now uses total points when deciding on suit contracts and HCP when deciding on NT contracts. This was not always the case, and GIB's methodology for counting total points is very rudimentary.

[diana_eva]

GIB Version 35 is here.

Changes:

• If GIB preempts and partner raises to game and the opponents then bid, GIB will stay "out of the way", and let partner decide between passing, doubling, or bidding.
• The ranges for rebidding a suit on the 2nd or 3rd level after fourth seat balances with a double have been suitably increased. GIB used to bid 1♠ - P - P - X - 3♠ with not nearly enough strength.
• The range has been corrected when a player makes a takeout double, partner makes a bid on the 2nd or 3rd level, and the doubler rebids 3NT. Previously an auction like 1♠ - X - P - 3♦ - P - 3NT showed 22-25 points and GIB always bid slam over it, but no more.
• GIB will no longer respond 2♦ to a third seat opening bid of 1♥ or 1♠ on a 3 or 4 card suit (a side effect of when GIB started to play drury, since 2♣ was no longer an option). 2♦ will always be a 5+ card suit.
• After an overcall, GIB will give more thought to introducing a long suit instead of making a cuebid raise of partner. So when it goes 1♠ - 2♣ - P, GIB won't bid 2♠ with 3 card club support and a 6 or 7 card heart suit, it will show the hearts instead.
• Similarly, when GIB overcalls and partner makes a cuebid raise, the overcaller will give more consideration to showing a 4 card major on the side. After 1♥ - 2♣ - P - 2♥ - P, GIB will show his 4 card spade suit if he has it.
• The range has been corrected so when the auction goes 1♥ - 2♦, GIB will cuebid raise with 3♦ if appropriate instead of bidding 4♥ on too good of a hand.
• A bug has been corrected where GIB "forgot" it opened the bidding after a 1NT overcall and transfer. Previously when the auction went 1♠ - 1NT - P - 2♦, opener-GIB would bid as though the auction was 1NT - P - 2♦ to him, and bid 3♠ with inadequate strength.
• Other various bug fixes and improvements.

[1eyedjack]

diana_eva, on 2016-March-15, 03:37, said:

GIB Version 35 is here.

Changes:

• If GIB preempts and partner raises to game and the opponents then bid, GIB will stay "out of the way", and let partner decide between passing, doubling, or bidding.
• The ranges for rebidding a suit on the 2nd or 3rd level after fourth seat balances with a double have been suitably increased. GIB used to bid 1♠ - P - P - X - 3♠ with not nearly enough strength.
• The range has been corrected when a player makes a takeout double, partner makes a bid on the 2nd or 3rd level, and the doubler rebids 3NT. Previously an auction like 1♠ - X - P - 3♦ - P - 3NT showed 22-25 points and GIB always bid slam over it, but no more.
• GIB will no longer respond 2♦ to a third seat opening bid of 1♥ or 1♠ on a 3 or 4 card suit (a side effect of when GIB started to play drury, since 2♣ was no longer an option). 2♦ will always be a 5+ card suit.
• After an overcall, GIB will give more thought to introducing a long suit instead of making a cuebid raise of partner. So when it goes 1♠ - 2♣ - P, GIB won't bid 2♠ with 3 card club support and a 6 or 7 card heart suit, it will show the hearts instead.
• Similarly, when GIB overcalls and partner makes a cuebid raise, the overcaller will give more consideration to showing a 4 card major on the side. After 1♥ - 2♣ - P - 2♥ - P, GIB will show his 4 card spade suit if he has it.
• The range has been corrected so when the auction goes 1♥ - 2♦, GIB will cuebid raise with 3♦ if appropriate instead of bidding 4♥ on too good of a hand.
• A bug has been corrected where GIB "forgot" it opened the bidding after a 1NT overcall and transfer. Previously when the auction went 1♠ - 1NT - P - 2♦, opener-GIB would bid as though the auction was 1NT - P - 2♦ to him, and bid 3♠ with inadequate strength.
• Other various bug fixes and improvements.

It is early days with version 35, and I don't like to rain on anyone's parade, but I confess to being somewhat underwhelmed by this announcement. The first 8 bullet points don't strike a chord with my recollection of the major faults with version 34 or with the issues repeatedly raised in the forum to which no official responses have been given for some time.

As just one of I suspect many examples, what about the problem with a t/o double of 1-suit opener being any one of a few strong and dramatically diverse hand types, only one (and the weakest of which) being a 3-suited hand of modest opening strength short in opener's suit, but the follow-up continuations allowing only for that possibility.

Frankly, given the length of time that version 34 has been out without upgrade I had expected something more. I may yet be pleasantly surprised that the 9th bullet point "other various bug fixes and improvements" conceals some gems, in which case as a marketing style the announcement might do with some improvement.

[lycier]

Gib is gradually improving day in day out, it is easy to find many wonderful Gibs hands, so I would suggest that it should pin a special thread to display wonderful plays by Gibs from version 35 on, its goal :


1- Show wonderful playing skill of Gibs.

2- Show Gibs' technical achievements and encourage the programmers.

3- More importantly, let our fans learn Gib CC more better and get pleasure from Gibs.

[smerriman]

Quote

A new robot version has been released! This is GIB version 36, a major update to the BBO robot.

The interpretation of the point range of 1NT in the following sequence has been changed from 18-19 HCP, to 12-14 HCP:
1C - (P) - 1♥ - (Double)
1NT <--- changed from 18-19 to 12-14

ROPI and DOPI sequence has been fixed.

GIB will now take any 6NT as a signoff no matter who used 4NT initially.

After 1M - 1NT, depending on whether the 1NT is forcing or not, GIB now understands the minor suit rebids can show either 3 or 4 cards so can pass or correct more accurately. Example:
1M - 1N (forcing)
2♣/2♦ ---> 3+ cards.
vs
P - (P) - 1S - (P)
1N (non forcing) - (P) - 2♣/2♦ ---> natural suits, 4+ cards.

Improved cuebidding sequences. Some examples:
(2♦) - P - (3♦) - 4♦ <--- shows majors.
and
(2♥) - P - (3♥) - 4♥ <--- shows spades + a minor. And in this continuation, 4NT now will be asking for the minor.
and
(2♠) - P - (3♠) - 4NT <--- shows both minors.

Various bug fixes and improvements.

I wonder what the various bug fixes and improvements are, since I don't recall ever have issues with the above situations, but repeatedly with many of the ones I've seen reported in the forums

[Stefan_O]

smerriman, on 2017-February-24, 00:04, said:

I wonder what the various bug fixes and improvements are, since I don't recall ever have issues with the above situations, but repeatedly with many of the ones I've seen reported in the forums

Where did you find this release notes Quote, you are citing?

afaics, it's not in this thread.
And I cannot find it by google either...

[smerriman]

It was in the news feed. News feed items seem almost impossible to find once they've disappeared from the sidebar within BBO, but here:

http://webutil.bridg...tch.php?id=3877