Robots need a better GF system!

[Stefan_O]

There is a quite common type of situations,

where the robot bidding-system is very un-helpful.

 

General situation is:

You are in a game-forcing sequence and there is still bidding-space available

to explore a possible slam or finding the best game-contract.

 

One trivial example is this:

 

1♦ - 2♣

2♦ - 3♦

?

 

We are in an obvious situation, where either player might have extra values,

and want to explore a possible (diamond) slam.

The other possibility is, that we want to play 3NT only if the pair

has both majors stopped, otherwise 5♦ is often a better bet.

 

So above, what does 3♥/3♠ mean here?

 

The Gib description simply says, like: 4+cards in the suit bid, 11-21 HCP.

which I hope ZERO descent players would agree with.

 

Even more useless is the continuation...

if opener bids 3♥ above --- after that, what does 3♠ from responder mean?

 

Again, Gib description just says: 4+spades, 12+HCP.

Que?? Why?????

If responder has a spade-suit, he would of course have bid 1♠ or 2♠ in the previous rounds.

 

In the above sequence, I would say 3♥/3♠ from opener primarily shows a stopper,

and asks pd to bid 3NT only with stopper in the unbid suit.

This is so we can avoid a hopeless 3NT when both players are missing stopper in one of the majors.

 

Alternatively, 3♥/3♠ can also be a premature que-bid with slam-interest,

in which case opener in the next round will

proceed with a new que-bid, blackwood or other methods.

 

Isn't this the method that most players use, if nothing agreed?

It's like the Gib definitions for these sequences have been defined by a complete novice.

 

The sequence above is just a typical example, there are many similar situations where

a new suit on the 3 level should not be natural as Gib "believes"

(because either or both players have already denied 4 cards in the suit)

but should be showing/asking for stopper, or possibly be a slamtry,

which will then be clarified in the next round(s).

 

Would be nice, if something could be done about this! :)

 

As it is now, you often have to gamble 3NT not knowing if you have an unbid suit stopped.

And exploring slam often gets restricted to Blackwood as the only option,

rather than cooperative que-bidding, etc, between the players.

[iandayre]

Your point is quite valid, thank you. We all learned some facts about GIB a few months ago. The original programmer's focus was on card play. He was NOT an expert bidder, and the weaknesses of GIB's bidding approach reflect that. BBO management assures us that they are working as fast as they reasonably can to improve GIB's bidding. Mind you, BBO has owned the GIB program for many years now, and many of us think they have not implemented improvements quickly enough. But that is where things stand for now.

 

Be careful though. I am not at all certain that the initial auction you gave - 1D-2C-2D-3D - is forcing with GIB. 1D-2C is its exception to 2/1 GF. I am SURE that GIB would consider 2D passable.

[Stephen Tu]

It would be really weird for 1d-2c-2d to be passable. 2/1 should promise a rebid even if it's not GF. It's responder's 2nd bid that can be passed on some sequences, usually 2nt/3c if 1d-2c is not GF. If GIB thinks 2d is passable then it's another bug.

[Stefan_O]

I am not at all certain that the initial auction you gave - 1D-2C-2D-3D - is forcing with GIB. 1D-2C is its exception to 2/1 GF.

 

In Gib's system, 1♦-2♣ is forcing to 3NT -- check the description next time you see it...

 

But regardless, there are many similar sequences...

Like, 1♥-2♣-3♣, just to give an example...

[Stefan_O]

We all learned some facts about GIB a few months ago.

 

OK... what was it particularly that happened a few months ago?

An announcement? Or what?

[Bbradley62]

In Gib's system, 1♦-2♣ is forcing to 3NT -- check the description next time you see it...

The System Notes, posted here ( http://www.bridgebase.com/doc/gib_system_notes.php ) in 2009, and minimally (if at all) updated since say that 1♦-2♣ is forcing one round. I would expect that, as Stephen Tu says above, it promises another bid by responder, but I would expect that over 1♦-2♣-2♦ any of 2NT, 3♣, or 3♦ is passable. But maybe I'm confused...

[Stephen Tu]

There might have been discussion at one point in time that it was going to be altered to 1d-2c game forcing. But 1d-3c is still a strong jump shift, rather than invitational with clubs, which probably indicates a hole introduced into the methods if the description of "forcing to 3nt" is now accurate. The whole 1d-2c rebid structure likely needs work.

[1eyedjack]

It is a bit of a tightrope to walk, whether to devote resources to fixing bugs in the existing system or to devote resources to changing the system.

 

Probably a bit of both is required.

 

Problems with changing the system are:

 

1) You will never get consensus on what the system should be, which itself will have to be a compromise between what is theoretically optimal and what is popular.

 

2) Changing the system will inevitably cause new bugs to creep in.

 

Problems with concentrating on bugs are ... If the system is going to get changed somewhere down the line anyway, the benefits of a bug-fix may be short-lived.

 

"They" certainly could devote all of their resources to fixing bugs without changing the system and without running out of work to do.

[iandayre]

A number of interesting points here.

 

Unless it has been changed VERY recently, 1D-2C has always been a one-round force. I have seen GIB, as responder, pass a 2D or 2NT rebid. I agree with Stephen that this is not optimal.

 

I have never advocated wholesale changes to the system. It needs improvement to the details, as Stefan accurately gave examples, but as far as the basic structure goes, I am OK with at least 95% of it. If I had to pick one basic change, that would be it - make 1D-2C a GF with 1D-3C invitational with long Clubs.

 

I will Stefan the short version of the "announcement" of a few months ago. I believe I was the instigator. Along with others, I had been critical of GIB's bidding over an extended period of time. Finally, I stated my concern that the ACBL could well discontinue awarding Masterpoints from GIB tournaments unless its performance improved. Fred Gitelman himself responded - the only time he has personally posted to this forum since I have been here. You can probably find that response if interested. Since that time, I have tried to give Fred and his staff the benefit of the doubt, and to give them time to do what needs to be done. We have had one upgrade since - it was a bit disappointing but it did cover a couple of important problem areas.

 

In the meantime I'm still here playing.

[Stefan_O]

One idea, if BBO does not have the motivation, skills or whatever needed to maintain and improve Gib...

 

Couldn't BBO cut a deal with some of the stronger competitors in the computer-championships (Wbridge, Jack, ...)

where (I think?) the software is still under active development...

-- maybe it's time to retire and replace the good ol' Gib :)

 

Has this possibility already been suggested or discussed?

[johnu]

One idea, if BBO does not have the motivation, skills or whatever needed to maintain and improve Gib...

 

Couldn't BBO cut a deal with some of the stronger competitors in the computer-championships (Wbridge, Jack, ...)

where (I think?) the software is still under active development...

-- maybe it's time to retire and replace the good ol' Gib :)

 

Has this possibility already been suggested or discussed?

 

I have asked whether Jack could replace GIB, and IIRC, several others have posted similar questions. Instead of paying independent programmers to make fixes to GIB, that money could be used to pay royalties to the Jack owner. But, cost could be a factor and maybe the Jack owner wants too much money since GIB has already been paid for. Jack may also use a lot more computer resources which would cost BBO more money in having to add additional servers. Only BBO knows :)

[iandayre]

One idea, if BBO does not have the motivation, skills or whatever needed to maintain and improve Gib...

 

Couldn't BBO cut a deal with some of the stronger competitors in the computer-championships (Wbridge, Jack, ...)

where (I think?) the software is still under active development...

-- maybe it's time to retire and replace the good ol' Gib :)

 

Has this possibility already been suggested or discussed?

 

If you were to read Fred's posting Stefan, one thing he said in no uncertain terms was that BBO is indeed motivated to improve GIB. Whether they have the skills, and are truly willing and able to devote the necessary resources, I auppose each of us will have to judge that for ourselves. I have no idea if they would ever consider licensing a more modern and advanced program for use on BBO.

[lycier]

Good topic.

See Gib CC :

 

1♦ - 2♣ : 2♣ only promises 11+hcp, forcing to 3nt

1♦ - 3♣: Soloway jump shift to show rebiddable ♣ with 17+TPs.

 

Now few people use Soloway jump shift, but Gibs use it, I'm not familar with its complete bidding constructure : I have no idea hw to show invitational sequence, sign off, slam and forcing sequences including splinter sequences etc, it seems only BBO knows it.

[johnu]

Now few people use Soloway jump shift, but Gibs use it

 

Many real life world class players play Soloway or other strong jump shifts

[Stefan_O]

Another example of these useless "4+suit" definitions I encountered, is this:

 

1♣ - 1♥

1♠- 1NT

2♦

 

2♦ was described as "4+♦..." but is rather a 4SF/NMF bid, showing at least invitational values and asking responder to describe his hand and go to game with maximum.

 

To make matters worse, there is also another bug related to this situtation:

http://i.imgur.com/gJQmFD5.png

 

As seen, the jump to 3♥ does not show extra-values (i.e. has same meaning as 2♥).

For lack of better, with 4315/18HCP, I bid 3♥ anyways, hoping only the description was wrong, but the robot passed with 9HCP.

[cloa513]

Many real life world class players play Soloway or other strong jump shifts

Such as who?

[manudude03]

Another example of these useless "4+suit" definitions I encountered, is this:

 

1♣ - 1♥

1♠- 1NT

2♦

 

2♦ was described as "4+♦..." but is rather a 4SF/NMF bid, showing at least invitational values and asking responder to describe his hand and go to game with maximum.

 

That seems normal enough, though I wouldn't say it promises 4+ diamonds. I'd bid 2D with a 4135 17 count.

[Stefan_O]

That seems normal enough, though I wouldn't say it promises 4+ diamonds. I'd bid 2D with a 4135 17 count.

 

But why should 2♦ promise anything at all about the diamond suit?

I believe (hope :rolleyes: ) most decent pairs play it as 4SF.

 

With your example-hand, 2NT seems like an obvious way to invite.

And with 4144, you open 1♦, not 1♣.

[Stephen Tu]

4sf is usually a responder tool, not by opener. On this auction I don't see any reason why opener can't bid strictly naturally.

[Stefan_O]

On this auction I don't see any reason why opener can't bid strictly naturally.

 

But again -- Why?

 

If you open 4144 with 1♦, the only time you can even have a ♦-suit is the rare 4045.

 

What you normally are interested to know here is:

- Does responder have 4 or 5 hearts?

- Does responder have extra strength or not?

 

Wanna play in diamonds? I think (almost) never... Much more likely to only help defenders by bidding 2♦ naturally...

[manudude03]

But again -- Why?

 

If you open 4144 with 1♦, the only time you can even have a ♦-suit is the rare 4045.

 

What you normally are interested to know here is:

- Does responder have 4 or 5 hearts?

- Does responder have extra strength or not?

 

Wanna play in diamonds? I think (almost) never... Much more likely to only help defenders by bidding 2♦ naturally...

 

It's to allow you to play 5C (or even 4S) when it is right if partner's hearts were particularly weak. If you want to check for a heart fit, then bid hearts. Meanwhile any bid (except maybe 2C) shows extras.

[Stephen Tu]

What you normally are interested to know here is:

- Does responder have 4 or 5 hearts?

- Does responder have extra strength or not?

- Responder denied extra strength by rebidding 1nt. If opener merely wants to bid 3nt if max for 1nt he can just bid 2nt.

- Opener is only interested in the fifth heart if he has extra values with 3 hearts. With that hand opener can just bid 2H.

 

Playing 2♦ as artificial in my mind only makes some sense if you want to play a style where opener always bids 2H with 3 without extras, on 4333/4324 type shapes, catering to responder having 5cd hearts, like Steve Robinson in "Washington Standard" advocates, which I don't think is mainstream.

 

Wanna play in diamonds? I think (almost) never... Much more likely to only help defenders by bidding 2♦ naturally...

Partner can be 5-5 or 4-6 in the reds pretty easily, might well want to play in diamonds.

[Stefan_O]

- Responder denied extra strength by rebidding 1nt.

 

I meant extras relative to his minimum limit :)

1NT is pretty wide range -- like 6-10 or so...

 

Playing 2♦ as artificial in my mind only makes some sense if you want to play a style where opener always bids 2H with 3 without extras, on 4333/4324 type shapes

 

What abt a minimum 4315 (or even 4306 at MP) with ♥HHx or so?

Opener has not shown a balanced hand here...

 

Partner can be 5-5 or 4-6 in the reds pretty easily, might well want to play in diamonds.

 

Well, once in a blue moon it might be better than NT perhaps... :)

But if accepting game-invite, even if 2♦ is artificial, responder can still show 5+suit with 3♦, if he feels unsure where to play.

[iandayre]

Good topic.

See Gib CC :

 

1♦ - 2♣ : 2♣ only promises 11+hcp, forcing to 3nt

1♦ - 3♣: Soloway jump shift to show rebiddable ♣ with 17+TPs.

 

Now few people use Soloway jump shift, but Gibs use it, I'm not familar with its complete bidding constructure : I have no idea hw to show invitational sequence, sign off, slam and forcing sequences including splinter sequences etc, it seems only BBO knows it.

 

I am a huge proponent of Soloway SJS, but only at the two level. I believe I can answer any questions about how to proceed after making one, or if your GIB partner makes one. Of course in "human best hand" tournaments, an SJS by GIB is extremely rare! I have seen only one.