[Bermy]

25 years ago I purchased a bridge robot made by Saitek that could not only bid and play better than Gib I also had option of what system and convention I would like to play.

The popularity of robot bridge and the funds arriving from huge robot tournaments must force the proprietors to invest in an all round better robot, if we apply enough pressure.

I would love to see a bridge robot that:

1) Allowed me to choose my own system.

2) Allowed me to choose which conventions I wanted or didn't want.

3) Bid according to bridge scoring (knowing vul from not vul, push the level up and risk a sac in part scores, double poorly bid contracts and sacs for penalties etc)

4) Play according to the type of tournament, in other words change style for imps, mps and Bingo.

5) Partner to give me more assistance in the bidding.

6) Avoid hysterical overbids and passes.

7) Prefer majors over minors and bid for 4-4 major fits.

8) Lead more aggressively especially to No Trumps.

9) More penalty doubles especially at low levels.

10) Bidding explanations that actually mean something.

Perhaps even if Gib could play at one table South in every tournament so players can evaluate just how well or poorly Gib really plays and judge themselves against the Gib score too.

Personally I would love to play Precision with the robot. I can imagine that Precision would be far easier to program as every bid has limited parameters and shape, with a clearly defined controlling bidder. However I would not like to comment on the program complexities, but rather have the programmers read what we the players would like to see in the future.

I invite other members of BBO and robot lovers to comment here on what kind of Bridge robot they would like to play with and what improvements can and should be done, and I ask the programmers to please stay out of the conversation.

[The_Badger]

Maybe it's time for BBO to consign GIB to the garbage bin and have a robot program built either around Jack or Wbridge5 that keep on winning the Robot Bridge World Championships.

Or, perhaps more to the point, for BBO to employ the programmers who update Jack or Wbridge5 to update GIB.

I like the idea of a Precision system-playing robot, but in fairness it is far better, I feel, to concentrate on one system 2/1, the most popular modern system being in use today and get that right, than to spread a robot's tasks too thinly covering a multitude of systems and conventions.

[manudude03]

I think for the sake of tourneys, the system must be constant. It would only add to the luck element if every player was potentially playing a different system. There is a lot that can be improved, but sometimes the cure is worse than the sickness. It would be cool to be able to adjust the system yourself, but it would be with the understanding that you will likely be playing a different system in the tourneys.

[steve2005]

Having different systems would bee nice. It would be a maintenance nightmare. BBO cant handle maintaining I system imagine 2 or evenmore.
I don't thing the system is the problem with Gib true it isn't ideal.

Gib makes ridiculous bids often in competition or on later rounds of bidding. This would still happen if you were playing Precision or something else.

My biggest peeves with GIB is it counts shortness in partners suit when playing in partners suit and when NT is a possibility. Also, with very weak hands it uses KC rather than cuebiding what little it has, also won't cuebid Kings.

[pigpenz]

the original gib would allow different systems
I cant even get real bridge players (club players) to be consistent
it takes a lot of time to get used to what GIB does and doesn't do well
it has been a work in progress but still it works

[Bermy]

I really do not think we need make excuses for the programmers, they seem to make enough for themselves.

My point is that these robots are becoming very popular indeed, and with the interest of the ACBL and with their US$ I am sure enough investment will come very soon.

So let them build the fantasy partner. One that bids and plays like a pro. One that bids your way, One that uses conventions correctly and judges contracts and scores well.

I'm not saying make bridge a perfect science, that would be impossible. Let us play with a robot partner we would love to have, and with E/W opponents that really test our skills. Robots are there to simulate partners and opponents at a bridge match. Surely having different systems and conventions will do exactly that. Why have tournaments where everyone is bidding and playing the same way? Have signal systems, why not? Have options on leads, why not? Then tweak you your system and conventions until you find the formula that gets you the best results. Yes we do need partner Gib options. Opponent Gib can play any way it likes, however I would expect strong opposition.

Gib is not the only bridge robot around, how about introducing different robots to BBO? Bridge robots have been around for 25 years now, and the development ideas already exist, why should we wait another 10 years for Gib to provide us with an alternative bidding system?

[Bermy]

Wow, I have recently downloaded the new ibridgeplus app and ............omg , Gib better start improving soon, as this is a robot for the 21 st century. This new app is what robot bridge should be all about. All the options, all the bidding systems, all the lead and signals even options on opps bidding systems. Even a design it yourself feature. And none of this “best hand “ nonsense, just real bridge. Highly recommended.

[TylerE]

If they could merely get it to play what it CLAIMS to play that would be a huge step up.

[zhasbeen]

TylerE, on 2018-May-01, 20:39, said:

If they could merely get it to play what it CLAIMS to play that would be a huge step up.

BBO MUST get a new robot or prepare to find itself going the way of OkBridge. Talking about all the fixes, bug reports, is a waste of time. I have concluded a long time ago that GIB can not be fixed, or at least not to the point that it can to play like you'd expect from a 21st century product.

I say this as someone who has probably played more robot tournaments than since my first in Aug 2017 than 90% of BBO customers. I have also done very well from a competitive standpoint, but I am getting closer and closer of breaking my BBO habit. The only thing that has kept me going the last couple few months has been getting involved in the ACBL yearly masterpoint race, but I'm now in process of letting it go. The frustration of playing with GIB has worn me down to the point where it's not worth it.
I don't even want to tell you how much I've spent. Hear that BBO? Support ignored my last complaint.

I'm tired of the frequency of incorrect definitions.
I'm tired of being left in 4-2 or 5-1 fit when GIB doesn't know what to do.
I'm tired of GIB blasting to slam off 2 aces.
I'm tired of having GB bidding one 5-card suit 2 or 3 times while ignoring a 2nd 5-card suit.
I'm tired of having to make a choice of selling out to opponents and having GIB leap to slam or game if I don't.
I'm tired of BBO doing nothing about it.

The list goes on and on.
Despite all our complaints (with 100's, if not thousands of examples) I don't see GIB as being one bit better than it was 8 months ago when I started. The forums are just a place for customers to vent their frustrations while the moderators keep trying to make excuses for GIB or have them believe that it can somehow be fixed.

I probably have not played my last game, but it is a virtual certainty that it will only be a small fraction I what I have been.

For the most part BBO has a lot to like, but that big elephant in the living room cancels out a large percentage of it.

[hrothgar]

Here's my vision of a perfect world

BBO partners with one of more commericial developers such that there program can run on the BBO server.

I envision a system in which you sign up for a tournament and indicate what Bot you want to play with.

• If you have purchased a copy of program "foo", there is no upcharge
  
• If you have not purchased a copy of program "foo" you can license it for this tournament for an additional fee

Tournament sponsors decide which bidding systems / conventions you can choose to play with your "Bot"

This all sounds good, but there is a lot of work that would need to be done.

At the very least, you now require:

• An interface by which my bot can disclose methods tou you an your bot
  
• An interface for displaying hands
  
• A payment system
  
• Willing partners

Of these, I expect that item 1 would be the most difficult. I suppose that you can sidestep this if you partner with ONLY Jack or ONLY Winbridge

Personally, I am kinda surpirsed that this hasn't happened already.
Maybe, someday...

[johnu]

The_Badger, on 2017-August-10, 04:01, said:

Maybe it's time for BBO to consign GIB to the garbage bin and have a robot program built either around Jack or Wbridge5 that keep on winning the Robot Bridge World Championships.

Or, perhaps more to the point, for BBO to employ the programmers who update Jack or Wbridge5 to update GIB.

IMO, replacing GIB with Jack or Wbridge would take a whole lot less time than trying to "fix" GIB. I also assume that Jack and WBridge people almost certainly don't have any experience in GIB's programming.

[johnu]

manudude03, on 2017-August-10, 05:48, said:

I think for the sake of tourneys, the system must be constant. It would only add to the luck element if every player was potentially playing a different system.

I strongly disagree. In the real world, players can play anything that's legal (except maybe for some very rare "one" convention card tournament). I don't see why robot tournaments need to use the same systems. If you choose to play a "better" system than the rest of the field, don't you deserve to win?

As a practical matter, the program's implementation of different systems/conventions may have wide levels of competence. Experience players may learn from experience that certain "book" bids are very likely to lead to a disaster (e.g. Making off shape takeout doubles with strong hands). BBO should be able to run extended simulations on specific systems/conventions that are made available and rate how well they work compared to the "standard" approach.

[Bermy]

TylerE, on 2018-May-01, 20:39, said:

If they could merely get it to play what it CLAIMS to play that would be a huge step up.

Yes, I agree, they do need a little work on the user interface, however the hard part has been done. This is a real bridge playing robot that really simulates a bridge match. It bids,plays and defends rather well too. The hard work has been done, without excuses as to how difficult AI is. They prove that what Gib programmers claim is not really impossible. Programming my partner to bid, lead and signal my way makes all the difference. I don’t have to endure the rather ridiculous methods of the programmers. I don’t have to endure their bidding opinions either. Sorry Gib and BBO but I have a new alternative now. Hooray 😃

[Bermy]

zhasbeen, on 2018-May-02, 08:44, said:

BBO MUST get a new robot or prepare to find itself going the way of OkBridge. Talking about all the fixes, bug reports, is a waste of time. I have concluded a long time ago that GIB can not be fixed, or at least not to the point that it can to play like you'd expect from a 21st century product.

I say this as someone who has probably played more robot tournaments than since my first in Aug 2017 than 90% of BBO customers. I have also done very well from a competitive standpoint, but I am getting closer and closer of breaking my BBO habit. The only thing that has kept me going the last couple few months has been getting involved in the ACBL yearly masterpoint race, but I'm now in process of letting it go. The frustration of playing with GIB has worn me down to the point where it's not worth it.
I don't even want to tell you how much I've spent. Hear that BBO? Support ignored my last complaint.

I'm tired of the frequency of incorrect definitions.
I'm tired of being left in 4-2 or 5-1 fit when GIB doesn't know what to do.
I'm tired of GIB blasting to slam off 2 aces.
I'm tired of having GB bidding one 5-card suit 2 or 3 times while ignoring a 2nd 5-card suit.
I'm tired of having to make a choice of selling out to opponents and having GIB leap to slam or game if I don't.
I'm tired of BBO doing nothing about it.

The list goes on and on.
Despite all our complaints (with 100's, if not thousands of examples) I don't see GIB as being one bit better than it was 8 months ago when I started. The forums are just a place for customers to vent their frustrations while the moderators keep trying to make excuses for GIB or have them believe that it can somehow be fixed.

I probably have not played my last game, but it is a virtual certainty that it will only be a small fraction I what I have been.

For the most part BBO has a lot to like, but that big elephant in the living room cancels out a large percentage of it.

I could not have said it better myself. However I have been playing with Gib every day for near 20 years now, if you want to work out what I have spent, so I have witnessed all their so called. Upgrades and improvements, bringing us to the rubbish we have today. Oh,and you forgot to mention rebidding a 4 card suit 3 times. How many more upgrades do we need? The limitations of the poor 2/1 bidding system that Gib uses (Not saying all 2/1is bad, just Gib modified version ) makes it impossible to ever get the structure to work.

[pescetom]

Bermy, on 2018-May-03, 00:35, said:

The limitations of the poor 2/1 bidding system that Gib uses (Not saying all 2/1is bad, just Gib modified version ) makes it impossible to ever get the structure to work.

This is an important point: it's much harder for a human to explain bidding to a robot than to explain card play. With today's technology it should be possible for a pair of AI based robots to beat even a pair of World Masters if it wasn't for the auction. The robot can refine card play by playing against itself, just look at how easily AlphaGo rose to beating world champions at Go.

[steve2005]

zhasbeen, on 2018-May-02, 08:44, said:

BBO MUST get a new robot or prepare to find itself going the way of OkBridge. Talking about all the fixes, bug reports, is a waste of time. I have concluded a long time ago that GIB can not be fixed, or at least not to the point that it can to play like you'd expect from a 21st century product.

I say this as someone who has probably played more robot tournaments than since my first in Aug 2017 than 90% of BBO customers. I have also done very well from a competitive standpoint, but I am getting closer and closer of breaking my BBO habit. The only thing that has kept me going the last couple few months has been getting involved in the ACBL yearly masterpoint race, but I'm now in process of letting it go. The frustration of playing with GIB has worn me down to the point where it's not worth it.
I don't even want to tell you how much I've spent. Hear that BBO? Support ignored my last complaint.

I'm tired of the frequency of incorrect definitions.
I'm tired of being left in 4-2 or 5-1 fit when GIB doesn't know what to do.
I'm tired of GIB blasting to slam off 2 aces.
I'm tired of having GB bidding one 5-card suit 2 or 3 times while ignoring a 2nd 5-card suit.
I'm tired of having to make a choice of selling out to opponents and having GIB leap to slam or game if I don't.
I'm tired of BBO doing nothing about it.

The list goes on and on.
Despite all our complaints (with 100's, if not thousands of examples) I don't see GIB as being one bit better than it was 8 months ago when I started. The forums are just a place for customers to vent their frustrations while the moderators keep trying to make excuses for GIB or have them believe that it can somehow be fixed.

I probably have not played my last game, but it is a virtual certainty that it will only be a small fraction I what I have been.

For the most part BBO has a lot to like, but that big elephant in the living room cancels out a large percentage of it.

like he said
dont forget the ridiculous defenses where Gib throws away high cards that could be useful
Gib doesn't make leads the way the commercial program does. Something has been screwed up in the current program possibly even messing up Gibs calculation of best move from sampling. Certainly it can't even cash out for -1 sometimes, there has to be a bug in the sampling program!

[steve2005]

Bermy, on 2018-May-03, 00:35, said:

The limitations of the poor 2/1 bidding system that Gib uses (Not saying all 2/1is bad, just Gib modified version ) makes it impossible to ever get the structure to work.

Nonsense. Gib is a non standard 2/1 version which is a problem not because it is bad but because people don't know it. Also, Gib program has a problem as it doesn't follow it's system and make the standard bids.

I submit that you could program any system including a system with almost no conventions and it would be trivial to get the program to bid well with no competition.

It is in competition that programing becomes incredibly difficult.

One of Gib solution to competition is to make balancing which humans might make on 13 cards show a slam try and is just ludicrous!