[tabansz]

I frequently encounter a position when I am trying to go for a save or competition and Robot thinks I have 23-24 HCP. Instead of passing it starts asking for Aces. So I realize that 11-22 HCP is simply too wide. I think a simple version of Polish Club, Meckwell Precision or Lauria Versace system may work better. Unfortunately they are all finely tuned and very complicated as the bidding change by position and color (red-white etc.) So maybe it is possible to work around them to find a good solid precision system for robots. It will also give us a chance to practice and learn a world class system.

[1eyedjack]

Resources

[Bbradley62]

I don't even think resources are the issue. It appears to me that the primary purpose of BBO having robots is to allow a majority of members of all skill levels to play bridge without having to arrange foursomes. This means having GIB play a system that the majority of members are generally familiar with.

As you can see from the threads in this forum, GIB has some issues. One theme, however, is that if you bid game opposite a silent robot partner, your partner is going to play you for (very nearly) having game in your hand.

---

You made the same 4♠ bid on this hand; I've posted it from another table, because GIB didn't post explanations for your bids as are provided here. Note that the opening bid limits South to 21 HCP and 22 total points. With the 4♠ bid, North presumes South is at the top of that range. (Best I can tell, GIB will never play you for having decided to sacrifice; he always thinks you're bidding to make in a competitive auction.)

[johnu]

tabansz, on 2014-December-08, 14:10, said:

I frequently encounter a position when I am trying to go for a save or competition and Robot thinks I have 23-24 HCP. Instead of passing it starts asking for Aces.

Check the meaning of your bid before you bid if you don't know what is shows. If your bid shows 23-24 and you only have 12, you are just asking for a disaster to happen. Now most of the time GIB has no business bidding even opposite a 23-24 HCP hand, but GIB is GIB, and now that you know, don't make those bids.

[1eyedjack]

Bbradley62, on 2014-December-08, 16:40, said:

I don't even think resources are the issue. It appears to me that the primary purpose of BBO having robots is to allow a majority of members of all skill levels to play bridge without having to arrange foursomes. This means having GIB play a system that the majority of members are generally familiar with.

No one would suggest disbanding the current system that GIB plays. What was suggested was additional systems, not replacements. The only bar to that is resources. There are insufficient resources even to get it to play its current single system properly.

[helene_t]

There are many auctions in which any bid shows 27+ HCPs, or 12-23 HCPs. Often it is so that even if a bid shows 12-23 GIB will apparently take it as 12-13 and pass it most of the time, maybe because 12 points comes up more often than 23 in the sims. Sometimes you just have to pass partners 4m bid even if you think game is likely to be on because you know it will raise 5m to 6.

I am not convinced that adding more systems would help with that. Maybe playing some kind of walrus precision where both partners show their hcps in first round will eliviate some of these issues but then they will create others. In the meantime, the problem will persist in defensive bidding, as well as constructive bidding for those users who prefer to play 2/1.

I am sure that the developpers are aware of these issues and that they agree that it is a priority. I am also sure that it is easier said than done to do something about it. Patching the holes one by one is probably the only short-term solution.

[Cthulhu D]

1eyedjack, on 2014-December-09, 00:41, said:

No one would suggest disbanding the current system that GIB plays. What was suggested was additional systems, not replacements. The only bar to that is resources. There are insufficient resources even to get it to play its current single system properly.

It has a comprehensive MOSCITO system already implemented - having an option to enable MOSCITO-BYTE with full relays when you have purchased a robot would probably be useful, and potentially low cost as the bidding database already exists. Though Hrothgar has been dismissive of it in the past, so it might suck.

It is possible based on Ginsberg's remarks and knowledge of how GIB's simulations work that it is easier for the robot to play a relay system - GIB's monte carlo simulations are going to be very well positioned after an auction like

1S-1N
2D-2H
3D-?? Now GIB knows that opener has exactly 3=3=6=1 and thus can deal hands that fit the bidding (and thus caculate the optimal place to play), very quickly.

helene_t, on 2014-December-09, 03:59, said:

There are many auctions in which any bid shows 27+ HCPs, or 12-23 HCPs. Often it is so that even if a bid shows 12-23 GIB will apparently take it as 12-13 and pass it most of the time, maybe because 12 points comes up more often than 23 in the sims. Sometimes you just have to pass partners 4m bid even if you think game is likely to be on because you know it will raise 5m to 6.

I am not convinced that adding more systems would help with that. Maybe playing some kind of walrus precision where both partners show their hcps in first round will eliviate some of these issues but then they will create others. In the meantime, the problem will persist in defensive bidding, as well as constructive bidding for those users who prefer to play 2/1.

I am sure that the developpers are aware of these issues and that they agree that it is a priority. I am also sure that it is easier said than done to do something about it. Patching the holes one by one is probably the only short-term solution.

Obviously limited openings (e.g. MOSCITO style 9-14 1DHSN openers) would help this massively. Having it as a toggable option would be very useful.

[cloa513]

They are not going to re-jig the whole system. We could ask for an opening 3NT bid(its a silly natural at the moment) to be an ope 11-16HCP 16-18 TP. 7 card majors- forces 4C to allow the natural major rebid- only continue above game with 8 TP.

[steve2005]

Gib already has MOSCITO theoretically. But how much would be needed to bring it up to date?

How many people know enough MOSCITO for it to be worthwhile? I suspect not enough for cost

[cloa513]

steve2005, on 2014-December-09, 18:31, said:

Gib already has MOSCITO theoretically. But how much would be needed to bring it up to date?

How many people know enough MOSCITO for it to be worthwhile? I suspect not enough for cost

How does GIB have MOSCITO theoretically?

[Cthulhu D]

cloa513, on 2014-December-09, 20:39, said:

How does GIB have MOSCITO theoretically?

It played a homebrewed MOSCITO at the Computer Bridge Worldchampionships - the bid tables have already been developed.

steve2005, on 2014-December-09, 18:31, said:

Gib already has MOSCITO theoretically. But how much would be needed to bring it up to date?

How many people know enough MOSCITO for it to be worthwhile? I suspect not enough for cost

The solution is not to update it, given that it is a unique MOSCITO variant to meet the conditions of contest for US midchart events. Just leave it as is.

Then you can play MOSCITO if you are complaining about the bots not being able to handle wide ranging opening bids, or you can suffer with the implementation of 2/1