BBO Discussion Forums: Sometimes you feel GiB is like a selfish child - BBO Discussion Forums

Jump to content

  • 2 Pages +
  • 1
  • 2
  • You cannot start a new topic
  • You cannot reply to this topic

Sometimes you feel GiB is like a selfish child

#1 User is offline   thepossum 

  • PipPipPipPipPip
  • Group: Full Members
  • Posts: 946
  • Joined: 2018-July-04
  • Gender:Male
  • Location:Australia

Posted 2020-October-14, 05:07

This is a general post about GiB style, and I know writing a computer bridge player is not easy, but i certainly has many immature childish and selfish tendencies in bidding and play. You know competing with your partner over suits, not co-operating over defence etc

I'm thinking of checking recent hands where there has been competition over suits during bidding, and those cases where a defensive lead has been made with the intention of coperative defence, and maybe a rare case where the bot helped your line of defence rather than taking it over for its own line

It even destroys the (as far as I know) time honored approaches of leading partner's suit, leading back what partner led, with the overuse of boring passive leads - you forget that rarely it does make a lead that should be led back. Its sad really. recnet example against 3NT when it led what I assumed was another boring passive lead and tried to establish a beautiful long solid heart suit. I found I was competing against my robot partner that managed to make a coupld of lowly long tricks from its originally led suit. My mistake. Saldy the contract was made instead of being defeated
0

#2 User is offline   jeigh 

  • PipPip
  • Group: Members
  • Posts: 16
  • Joined: 2016-May-21

Posted 2020-October-17, 06:41

I have had the robot take me out of a perfectly viable 3D partscore and argue me up to *SEVEN HEARTS* on a hand containing five to the Q9 and 7-8 HCP.

Yes, of course I kept bidding diamonds, just to see if it would ever stop. Just solitaire, not money :rolleyes: . For some reason, while coding in a complete lack of understanding of giving up captaincy, the developers failed to allow GIB to bid to the 8 level.

Feels like you're playing with Yosemite Sam's horse. "When I say 'Whoa' I mean 'WHOOOOOOOOOAAAA'."
0

#3 User is offline   smerriman 

  • PipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 1,850
  • Joined: 2014-March-15
  • Gender:Male

Posted 2020-October-17, 13:09

View Postjeigh, on 2020-October-17, 06:41, said:

I have had the robot take me out of a perfectly viable 3D partscore and argue me up to *SEVEN HEARTS* on a hand containing five to the Q9 and 7-8 HCP.

While GIB is pretty bad, I would be willing to bet you were at fault here. Normally happens when the human misunderstands the system / doesn't read the definitions and continually makes cuebids thinking they're natural, which GIB is forced to remove.
0

#4 User is offline   jeigh 

  • PipPip
  • Group: Members
  • Posts: 16
  • Joined: 2016-May-21

Posted 2020-October-17, 20:55

View Postsmerriman, on 2020-October-17, 13:09, said:

While GIB is pretty bad, I would be willing to bet you were at fault here. Normally happens when the human misunderstands the system / doesn't read the definitions and continually makes cuebids thinking they're natural, which GIB is forced to remove.


There's an easy logic here. Once the human hand has limited its value, GIB's job is to STFU. If its system is one that enables it to go what it "knows" is 3-4 levels too high, it's not the player's fault no matter what the player does. If its system allows it to bid up from a "drop dead" part score to the 7 level against a limited overcall or opener on a five card suit to one lesser honour, then it's a system not worth understanding.

(Human here, who is not a particularly great player, *always* reads the definitions, when there are some, and is never amazed to discover that GIB is usually anywhere from 3-4 to *10-12* points above or below its own definition.)

Apart from all its play shortcomings, such as its propensity to lead Q from Qx in a random suit, to have no concept of carding or signaling, to usually play the worst possible card from its hand on defense ... its bidding seems even worse.

Sorry if there's attitude here. In the past few years since I started playing solitaire here, the game has degraded from fun to painfully useless. I don't know who's trying to "fix" or "improve" things, but they've created something that is a bit heartbreaking out of something that was once a lot of fun.
0

#5 User is offline   smerriman 

  • PipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 1,850
  • Joined: 2014-March-15
  • Gender:Male

Posted 2020-October-17, 21:20

View Postjeigh, on 2020-October-17, 20:55, said:

There's an easy logic here. Once the human hand has limited its value, GIB's job is to STFU. If its system is one that enables it to go what it "knows" is 3-4 levels too high, it's not the player's fault no matter what the player does. If its system allows it to bid up from a "drop dead" part score to the 7 level against a limited overcall or opener on a five card suit to one lesser honour, then it's a system not worth understanding.

If you have bid in a way that 100% definitively set a suit as trumps, and the human then continually bids a new suit despite being warned that it is a cuebid, that is most certainly the player's fault. It is simply a waste of time trying to program a robot to handle cases which can never exist, especially given the complexity of the bidding database.

Not saying that's what happened here - I suspect it, but would need to see the hand to know for sure - but while GIB has many faults, there's no way that I'd accept that particular scenario was not the fault of the player.

In fact, one of the biggest complaints about GIB is that if you make a bid which it doesn't understand, it often passes, and that this should be fixed by defaulting to bidding on. Based on your logic, you seem to want the opposite.
0

#6 User is offline   jeigh 

  • PipPip
  • Group: Members
  • Posts: 16
  • Joined: 2016-May-21

Posted 2020-October-19, 08:50

View Postsmerriman, on 2020-October-17, 21:20, said:

If you have bid in a way that 100% definitively set a suit as trumps, and the human then continually bids a new suit despite being warned that it is a cuebid, that is most certainly the player's fault. It is simply a waste of time trying to program a robot to handle cases which can never exist, especially given the complexity of the bidding database.

Not saying that's what happened here - I suspect it, but would need to see the hand to know for sure - but while GIB has many faults, there's no way that I'd accept that particular scenario was not the fault of the player.


I guess part of my issue is that there is often no hint - neither in the bidding nor the explanation - that GIB has entrenched a suit as trumps. Seems lately its most common explanation is "looking for an explanation."

Playing solitaire anonymously - for some reason you can do all sorts of re-dos. I do this a lot because I'm not competing against anyone or anything, just would like to play a certain hand a certain way. Good practice in case I ever decide to play with/against humans again.

I just had a sequence in which GIB opens a heart. I have an opening bid - stiff HK, 5 good spades (to AJT), 5 good diamonds (to AJ), 2 mid-spot clubs. I bid a spade. GIB bids 1NT and I bid 2D. GIB bids 2S. So far pretty normal. I bid 3D (extra values and enough diamonds to bid twice, seemed pretty natural to me.) GIB passes. (Yes, I'm aware it could have passed 2D too.) So I "undid" 3D and bid 4D instead - just grasping at what it could possibly be "thinking". It raised to 5D.

GIB's hand was a minimum semi-balanced 3-5-2-3 with 3 perfectly reasonable spades and 2 middle diamonds. An obvious 4S game (no idea if it makes, or more to the point, if I as a non-great player would make it - I didn't bother.)

To your point, either my second diamond bid was a cue bid (it'd *better* not pass) or a values bid (it'd also *better* not pass.) It wasn't a bid it could treat as a psyche and pass out of confusion, really.

That is to say, I understand and agree with your point about how it's programmed, but it broke its own programming. It seems to me to do this, emphasis "seems", at least 80% of the time. We'd "agreed" on spades and it wouldn't for love or money let me evaluate its spades to determine if we had game, part score or even 3NT. I either raise spades and hope it hasn't "preferred" on 2 small spots, or I try to get it to show me something about its hand.

I can't recall if in the hand I described in my initial post was about an "agreed" suit; pretty sure there wasn't one. I am dead certain there is no system in which a player would cue bid diamonds 4 times in a row, though. Says here, if it doesn't understand my bid, after the first time, its job is to STFU.

Just one of a dreadful number of things that completely break any joy I used to have in this game. And with apologies to all, that's why I'm upset.
0

#7 User is offline   smerriman 

  • PipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 1,850
  • Joined: 2014-March-15
  • Gender:Male

Posted 2020-October-19, 13:22

View Postjeigh, on 2020-October-19, 08:50, said:

To your point, either my second diamond bid was a cue bid (it'd *better* not pass) or a values bid (it'd also *better* not pass.) It wasn't a bid it could treat as a psyche and pass out of confusion, really.

Unfortunately this in some sense demonstrates my point - that you haven't understood the system. Your correct bid over 1NT is 2 (GIB plays new minor forcing; in this sequence some play both minors as NMF, but you can confirm from the definitions that it just uses 2 here).

You got a second chance when it showed you three spades and could have raised them - but didn't. So in both cases, it's going to take inferences from the fact you haven't done so.

In particular, 2 then 3 is described as an invitational hand with 6, so it will always pass with a minimum. 2 then 4 probably shouldn't exist, but it's described as game forcing with 6 diamonds; since you chose not to show a 5th spade, it's always going to raise diamonds.

You have most certainly not agreed spades at any time in this sequence, since you may simply have 4; you know you have an 8 card fit, but GIB only knows you have 7.

It's similar to a sequence like 1 - 1nt - 2 - 2 - this doesn't agree spades, since it may be a preference bid with 2 spades. If opener continues with diamonds, responder is under no obligation to correct back to spades.

In both cases, the person who knows about the fit must insist upon it before it's too late - so the error here was on your side.
0

#8 User is offline   jeigh 

  • PipPip
  • Group: Members
  • Posts: 16
  • Joined: 2016-May-21

Posted 2020-October-19, 14:37

View Postsmerriman, on 2020-October-19, 13:22, said:


You have most certainly not agreed spades at any time in this sequence, since you may simply have 4; you know you have an 8 card fit, but GIB only knows you have 7.

It's similar to a sequence like 1 - 1nt - 2 - 2 - this doesn't agree spades, since it may be a preference bid with 2 spades. If opener continues with diamonds, responder is under no obligation to correct back to spades.

In both cases, the person who knows about the fit must insist upon it before it's too late - so the error here was on your side.


This suggests that GIB has 3 spades. Turned out it did, but it would have bid exactly the same, I think, with two tiny spades and three mid-range diamonds. At least, I'm pretty sure that's what I see it do.

I suppose I could/should bid clubs in that situation. I'm pretty sure when I've tried that on other occasions, GIB has gotten real excited about clubs, which I would in this hand have been bidding with a doubleton.

I also suppose that I really have a strong dislike for this system even on the rare occasions when GIB bids and explains it accurately. Seems about half the time, it's not possible to find a bid to describe my hand that GIB will understand and react to. Good ol' SAYC and almost any plain wrapper 2/1 works way better at that, for me anyway.

I marvel a bit at a system that leaves you with no choice but to defend hands you could have owned and made, and then is entirely random at defending. No carding, no logical choice of suit, seems more often than not no standards for which card in a suit to play. It's of no satisfaction at all to constantly make 2-3 tricks more than I deserve due to its horrific defense - I'm not trying to "win", I was hoping to sharpen and maybe even learn a bit.
0

#9 User is offline   jeigh 

  • PipPip
  • Group: Members
  • Posts: 16
  • Joined: 2016-May-21

Posted 2020-October-19, 14:39

P.S. I really do appreciate you taking the time to discuss this.
0

#10 User is offline   johnu 

  • PipPipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 3,855
  • Joined: 2008-September-10
  • Gender:Male

Posted 2020-October-19, 14:50

View Postsmerriman, on 2020-October-19, 13:22, said:

In particular, 2 then 3 is described as an invitational hand with 6, so it will always pass with a minimum. 2 then 4 probably shouldn't exist, but it's described as game forcing with 6 diamonds; since you chose not to show a 5th spade, it's always going to raise diamonds.

Amazing that GIB has that definition for a wild jump to 4 when it doesn't have definitions for much more common and lower bids.

If the definition is as you say, that's nuts. For most human bridge players (don't know much about other robot systems), you would respond 2 and then bid spades to show longer diamonds than spades (already in game force after 2/1). This will almost always keep the bidding below 3NT after responder's 2nd bid instead of being at 4 with no fit known. There would be no reason to have the 4 jump mean the same thing in a sensibly designed system.
0

#11 User is online   sfi 

  • PipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 1,864
  • Joined: 2009-May-18
  • Location:Oz

Posted 2020-October-19, 14:56

View Postjeigh, on 2020-October-19, 14:37, said:

I also suppose that I really have a strong dislike for this system even on the rare occasions when GIB bids and explains it accurately. Seems about half the time, it's not possible to find a bid to describe my hand that GIB will understand and react to. Good ol' SAYC and almost any plain wrapper 2/1 works way better at that, for me anyway.

One major problem is that BBO's system is not at all suitable for newer players, and not terribly suitable for people who don't already know 2/1 fairly well. It approximates some version of "expert standard" 2/1, so there are subtleties that are not always well explained or understood. And then there's the fact that the robots often don't get those subtleties themselves, which can lead to silly results without the player doing anything bad.
0

#12 User is offline   smerriman 

  • PipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 1,850
  • Joined: 2014-March-15
  • Gender:Male

Posted 2020-October-19, 16:45

View Postjeigh, on 2020-October-19, 14:37, said:

This suggests that GIB has 3 spades. Turned out it did, but it would have bid exactly the same, I think, with two tiny spades and three mid-range diamonds. At least, I'm pretty sure that's what I see it do.

That may well be true; after all, by bidding 2 you are asking it to give a preference between diamonds and spades, and a 5-2 fit is usually preferable to a 4-3 one.

Thus the reason for the artificial 2 bid; then you find out definitively whether they have 3 spades or not.

View Postjeigh, on 2020-October-19, 14:37, said:

I suppose I could/should bid clubs in that situation. I'm pretty sure when I've tried that on other occasions, GIB has gotten real excited about clubs, which I would in this hand have been bidding with a doubleton.

The club bid is 100% artificial and the description doesn't say anything about clubs, so the other occasions must have been unrelated bidding sequences.

View Postjeigh, on 2020-October-19, 14:37, said:

Seems about half the time, it's not possible to find a bid to describe my hand that GIB will understand and react to.

There are certainly plenty of sequences that GIB can't handle properly. But I wouldn't say this is anywhere near half - if you keep posting hands where you're stuck, you may find it was just a case of missing what the correct sequence was (like the artificial 2 above).
0

#13 User is offline   smerriman 

  • PipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 1,850
  • Joined: 2014-March-15
  • Gender:Male

Posted 2020-October-19, 16:48

View Postjohnu, on 2020-October-19, 14:50, said:

Amazing that GIB has that definition for a wild jump to 4 when it doesn't have definitions for much more common and lower bids.

.. There would be no reason to have the 4 jump mean the same thing in a sensibly designed system.

I agree - that's why I said 4 probably shouldn't exist. The definition will most likely just be an automatically generated combination of the level and rebid showing a minimum number of points and length.

Given 2 is NF, what do you think 4 should mean in a sensibly designed system?
0

#14 User is offline   johnu 

  • PipPipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 3,855
  • Joined: 2008-September-10
  • Gender:Male

Posted 2020-October-19, 18:18

View Postsmerriman, on 2020-October-19, 16:48, said:

I agree - that's why I said 4 probably shouldn't exist. The definition will most likely just be an automatically generated combination of the level and rebid showing a minimum number of points and length.

Given 2 is NF, what do you think 4 should mean in a sensibly designed system?




I don't know how sensible it is, but if it came up undiscussed with an expert partner, I would assume it was a self splinter, very good 6+ spades, singleton/void in diamonds.
0

#15 User is offline   smerriman 

  • PipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 1,850
  • Joined: 2014-March-15
  • Gender:Male

Posted 2020-October-19, 18:40

Of course, but that's not the auction we're talking about, which is 1H - 1S - 1N - 2D - 2S - 4D.

I think you misread something :)
0

#16 User is offline   jeigh 

  • PipPip
  • Group: Members
  • Posts: 16
  • Joined: 2016-May-21

Posted 2020-October-20, 22:02

View Postsmerriman, on 2020-October-19, 16:45, said:

That may well be true; after all, by bidding 2 you are asking it to give a preference between diamonds and spades, and a 5-2 fit is usually preferable to a 4-3 one.

Thus the reason for the artificial 2 bid; then you find out definitively whether they have 3 spades or not.


The club bid is 100% artificial and the description doesn't say anything about clubs, so the other occasions must have been unrelated bidding sequences.


There are certainly plenty of sequences that GIB can't handle properly. But I wouldn't say this is anywhere near half - if you keep posting hands where you're stuck, you may find it was just a case of missing what the correct sequence was (like the artificial 2 above).


Appreciate all your points. I'll keep an eye out for that club bid; I've played enough solitaire here that I'm pretty sure I'd have recognized it if there was an explanation that went with it, so I'm guessing the system uses the bid but doesn't explain it. I may have just missed it, though.

I really strongly dislike this system, and regret that a system that looks highly contrived and filled with gaps and widgets for workarounds should be the required standard of the many thousands of us from all over the world and all over time to be stuck with.

You or others may disagree, but to me, this is a camel system - camel being the result of trying to design a horse by committee.

I can't imagine it being a useful thing to many people to spend that much time learning the nooks and crannies of the system unless one would *ever* want to play that system with human partners. I wouldn't. Obviously, it's entirely likely some would.

It seems really short-sighted that any large-scale organization would bank its entire world on a ridiculously complicated and unintuitive system, when there are so many really easy and intuitive ones it could use instead. Especially when there are such huge gaps in its game anyway - why subject everyone to a bidding system that requires a ton of time to learn or a bunch of guessing to use, and then completely ignore defense, carding, etc?
0

#17 User is offline   smerriman 

  • PipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 1,850
  • Joined: 2014-March-15
  • Gender:Male

Posted 2020-October-20, 22:39

If by 'system' you mean the NMF convention, almost every pickup partner you'll find online will also play it. Those that don't will likely play some form of xyNT where both 2 and 2 are *both* completely artificial bids, unrelated to clubs and diamonds. I would agree that's an improvement, though given you were wanting to bid diamonds naturally, probably not what you were looking for.
0

#18 User is offline   jeigh 

  • PipPip
  • Group: Members
  • Posts: 16
  • Joined: 2016-May-21

Posted 2020-October-21, 15:39

View Postsmerriman, on 2020-October-20, 22:39, said:

If by 'system' you mean the NMF convention, almost every pickup partner you'll find online will also play it. Those that don't will likely play some form of xyNT where both 2 and 2 are *both* completely artificial bids, unrelated to clubs and diamonds. I would agree that's an improvement, though given you were wanting to bid diamonds naturally, probably not what you were looking for.


Could be, and it's probably me - I haven't played bridge with other people in over 30 years, though I play a lot of solitaire here and watch a lot of experts and near-experts play. And I look at the explanations when they exist, which they often don't. I don't in any way pretend to be up on this.

But this didn't strike me as a place where NMF applies. I guess it might, but I assumed after GIB started with 1H, rather than a minor, and then preferred my spades over my diamonds, that 3C would confirm spades and be a cue or it would confirm short hearts. It may not apply anymore, but I always learned that in this case, re-bidding my diamonds showed extra strength/length in diamonds, to give partner a better understanding before settling on a suit. The "extra" component opposite an opening bid, I assumed, would force a further bid.

I probably stand corrected, but I don't stand any happier with the system.
0

#19 User is offline   smerriman 

  • PipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 1,850
  • Joined: 2014-March-15
  • Gender:Male

Posted 2020-October-21, 16:34

The NMF part was referring to a 2 bid, not a 3 bid. The key is that 2 is a non-forcing bid that could be passed - in fact, swap its third spade for a diamond and it will - so you're never going to be able to force after that. Rebidding diamonds shows extra length, but not extra strength that you've already denied.

Descriptions below:


0

#20 User is online   sfi 

  • PipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 1,864
  • Joined: 2009-May-18
  • Location:Oz

Posted 2020-October-21, 17:25

The other thing to note is that you've never shown 5 spades. The auction 1H-2D is forcing to game, so hands with 4S/5+D start by bidding spades. The auction 1H-1S; 1NT-2D is entirely consistent with a weak hand and four spades and six diamonds, for example.

Even in the auction smerriman is talking about:
1H - 1S
1NT - 2C* (artificial, inv+ hand)
2S (min, 3-card support)

3D still doesn't show a fifth spade. In fact it denies it and shows an invitational hand with longer diamonds.

So what do you do when you actually have a spade fit? I don't know the details of what the bots play, but a jump into a new suit might be a cue (it might also be a shortage). A 3H bid is likely to be available as an artificial forcing bid (partner denied 4H, if you have four you already have a spade fit, and if you had 5-5 I think you could have jumped to 3H over 1NT). The point is you need to explicitly set spades since partner doesn't yet know you have five of them.

This all arises because if all your bids are natural you can't show everything. Consider the auction:
1H - 1S
1NT

With long spades you would like to be able to show weak, invitational and forcing hands. But only 2S and 3S are available, so something has to give. New Minor Forcing, and the array of more sophisticated complicated conventions give up a natural 2C bid (and frequently a natural 2D as well) to try to sort these ranges out. They do work well once you get used to them.
0

Share this topic:


  • 2 Pages +
  • 1
  • 2
  • You cannot start a new topic
  • You cannot reply to this topic

1 User(s) are reading this topic
0 members, 1 guests, 0 anonymous users