[pescetom]

Good news, thanks.

[diana_eva]

Hi all,

GIB has been upgraded to Version 42.

The only change in this release is a bugfix for the recent regression observed in basic robots behaviour, where the basic GIB sometimes plays a low card as 3rd hand.

The V3 interface has also been upgraded to offer a feature that had been part of Version 41, but not visible until now which allows the robots to claim, when all their tricks are high. This feature is available in daylongs and challenges for now. We plan to offer it in the rest of the games in the future updates.

Players in Instant tournaments may notice the robots at their table making bids and plays different from the original hands robots, due to the discrepancy between GIB versions. These differences will iron out and disappear in a few weeks, as the deals used for comparisons will pick more recent deals using the latest GIB version.

Hope you enjoy, and have fun playing

[smerriman]

diana_eva, on 2025-January-13, 13:59, said:

Fixed an issue where GIB attempted a finesse when it could have claimed all remaining tricks.

Seems it's still finessing.

[benellis58]

I play at least one robot game each day. Consequently, and not at all surprisingly, I am reminded each day of how astoundingly incompetent the GIB robots are. Miraculously, I enjoy the games in spite of this robotic ineptitude, but I would enjoy them infinitely more if the GIB robots did not do so many inexcusably stupid things.

[diana_eva]

We’ve upgraded our robots to GIB Version 43

What’s new:

• Robots will now claim in all games when the rest of the play is obvious (expanded from claims only in daylongs and challenges)
• Declarer play by GIB advanced is now a bit better

Enjoy, and - as usual - feedback is welcome.

[Derrygirl2]

It would be great if BBO would provide robots who play ACOL as an option for all robot games.

[smerriman]

Any chance of expanding on "a bit better"?

[Ditherer]

diana_eva, on 2025-May-07, 09:10, said:

We’ve upgraded our robots to GIB Version 43

What’s new:

• Robots will now claim in all games when the rest of the play is obvious (expanded from claims only in daylongs and challenges)
• Declarer play by GIB advanced is now a bit better

Enjoy, and - as usual - feedback is welcome.

When humans claim, the system will reject the claim if there is a possible distribution that would invalidate the claim. On the other hand, when robots claim I suspect they do so on the basis of the actual lie of the cards. At least, that's what seemed to happen the other day when a robot opponent claimed: the claim appeared to be based on the assumption (rather than certainty) that my robot partner could not over ruff -- sorry but I didn't record the board. I hope my suspicion is unfounded and the robots aren't peeking at our cards before making a claim!

[diana_eva]

Ditherer, on 2025-May-08, 07:24, said:

When humans claim, the system will reject the claim if there is a possible distribution that would invalidate the claim. On the other hand, when robots claim I suspect they do so on the basis of the actual lie of the cards. At least, that's what seemed to happen the other day when a robot opponent claimed: the claim appeared to be based on the assumption (rather than certainty) that my robot partner could not over ruff -- sorry but I didn't record the board. I hope my suspicion is unfounded and the robots aren't peeking at our cards before making a claim!

The robots aren't peeking. They make claims based only on the cards that have already been played - not on hidden knowledge like "this suit splits 3–3." They claim when all their remaining cards are high and the outcome is certain based on what's known. You can always reject the claim if you don't "see it", by the way.

They tend to claim like an expert - having already counted everything - so while we might still be thinking it through, or we missed a spot, the robot always knows, for example, that only two cards remain from a specific suit, and it holds the A-K, so it doesn't matter who has those two cards and in which way the suit breaks.

[lorserker]

 smerriman, on 2025-May-07, 13:30, said:

Any chance of expanding on "a bit better"?

0.05 imps per board

[smerriman]

I meant in terms of whether any specific bugs had been fixed / old ones would need rereporting if they keep occurring, anything in particular to look for, etc.

[Maestra108]

How do you accept the robot claim and go on to the next board

[Khrystyna]

2 questions:

Is the robot able to change what it plays/bids in competitive games depending on what the ops play /bid? Today my robot made a bid that made no sense given the bidding, a double which was described in the notes as a “responsive” double. Did not make any sense at all- it certainly left me in a quandary and our side bottom score. BTW no other robot made this bid. given the same bidding up to that point that my robot doubled.

Also, the robot posts a 2/1 cc showing Standard Carding. Is it capable of this? I personally doubt it given what I've seen playing.

Your responses will be good guides. I learn a great deal playing with them!

[smerriman]

Khrystyna, on 2025-May-08, 18:32, said:

Is the robot able to change what it plays/bids in competitive games depending on what the ops play /bid?

No, it assumes everyone at the table plays its own system.

Khrystyna, on 2025-May-08, 18:32, said:

Today my robot made a bid that made no sense given the bidding, a double which was described in the notes as a “responsive” double. Did not make any sense at all- it certainly left me in a quandary and our side bottom score.

If you're referring to this hand, then the bid seems normal - it shows diamonds and hearts, like the description says. (Here's a page about what responsive doubles are and how they work.)

Khrystyna, on 2025-May-08, 18:32, said:

BTW no other robot made this bid. given the same bidding up to that point that my robot doubled.

There were only 5 other tables with robots in the North seat, but in none of them was the bidding by the humans the same up to that point.

[Ditherer]

diana_eva, on 2025-May-08, 08:43, said:

The robots aren't peeking. They make claims based only on the cards that have already been played - not on hidden knowledge like "this suit splits 3–3." They claim when all their remaining cards are high and the outcome is certain based on what's known. You can always reject the claim if you don't "see it", by the way.

They tend to claim like an expert - having already counted everything - so while we might still be thinking it through, or we missed a spot, the robot always knows, for example, that only two cards remain from a specific suit, and it holds the A-K, so it doesn't matter who has those two cards and in which way the suit breaks.

Thank you for confirming that my suspicions were unfounded. I managed to find the board and the robot declarer did indeed have sufficient information to claim at the point that it did, i.e., it knew that my partner would not be able to over ruff. I simply hadn't been paying attention when my robot partner discarded when a trump was led a few tricks earlier as I could tell that there were not going to be any decisions for me to make in the remaining tricks.