[USViking]

This occured on Deal 2 of #6141 Express- Free Automated Fun
at about 10:00am Sunday 9/25.

Robot doubles 7NT, on OL, possessing an Ace, partner silent.

OL = low from T643 2nd strongest suit!!!

Is an easy fix possible?

[Bbradley62]

---

Of course, we know that this problem has been reported several times:

6/25/11: http://www.bridgebas...-your-ace-pard/

3/1/11: http://www.bridgebas...ash-ace-vs-7nt/

10/25/10: http://www.bridgebas...-why-double-7n/

This post has been edited by Bbradley62: 2011-September-26, 11:58

[helene_t]

The robot assumes that EW play the gib system so that 4NT shows a balanced hand. Not sure what the shape constraints of 2NT are but maybe they are specifically 3334. The chance that EW have 13 tricks without opening the spade suit is slim and may not occur during the simulations.

So this would be very difficult to fix in a good way. Patching the robots with all kinds of very specific rules such as "always lead an ace against 7NT" is hardly satisfactory.

[xxhong]

The robot shouldn't take human opps' bidding too seriously. In a better set up, I think the robot should only take partner's and robot opp's bidding very seriously and shouldn't simulate based purely on human opps' bidding, which is often not reliable at all. Also many human players tend to distort the minor suit openings, 1NT range and other bids to mislead gib and take an advantage. Gib is especially vulnerable to human opps psyche or semi-psyche, because gib never implements any penalty doubles even at 7 level. I once opened 1D with 4 HCP, one Q and two jacks. My gib pushed me to 6NT, and my opps' gib holds SA and DAK and failed to double this 6NT. This makes psyche when white very cheap. If my gib hangs me by pushing me to slams, I still don't suffer a huge loss and only went down quietly 4 tricks. -200 and -800 is a huge difference and makes psyche very very profitable because in those times your psyche works, it works very very well.

helene_t, on 2011-September-25, 15:28, said:

The robot assumes that EW play the gib system so that 4NT shows a balanced hand. Not sure what the shape constraints of 2NT are but maybe they are specifically 3334. The chance that EW have 13 tricks without opening the spade suit is slim and may not occur during the simulations.

So this would be very difficult to fix in a good way. Patching the robots with all kinds of very specific rules such as "always lead an ace against 7NT" is hardly satisfactory.

[Bbradley62]

helene_t, on 2011-September-25, 15:28, said:

Patching the robots with all kinds of very specific rules such as "always lead an ace against 7NT" is hardly satisfactory.

There wouldn't be "all kinds of" such rules; but this one doesn't sound like a bad idea.

[USViking]

helene_t, on 2011-September-25, 15:28, said:

The robot assumes that EW play the gib system so that 4NT shows a balanced hand. Not sure what the shape constraints of 2NT are but maybe they are specifically 3334.

3334 sounds a bit too restrictive and inflexible.

helene_t, on 2011-September-25, 15:28, said:

The chance that EW have 13 tricks without opening the spade suit is slim and may not occur during the simulations.

I would like to see simulation results. I cannot believe they would be anywhere near zero.

helene_t, on 2011-September-25, 15:28, said:

So this would be very difficult to fix in a good way. Patching the robots with all kinds of very specific rules such as "always lead an ace against 7NT" is hardly satisfactory.

I will go along with Mr. Bradley's earlier comment here, and add that if specific rule patches
can be made in the form of what I called an "easy fix" then why not make them?

[barmar]

Here's a similar case where it doubled 6NT with two cashing aces, and went to bed with one of them. This was in a Robot Rebate.

---

[steve2005]

helene_t, on 2011-September-25, 15:28, said:

So this would be very difficult to fix in a good way. Patching the robots with all kinds of very specific rules such as "always lead an ace against 7NT" is hardly satisfactory.

As a human I follow that rule don't see why GIB cant should be easy.

[uday]

I agree, patching to force the lead of an A vs 7N seems manageable but there will be other similar cases.

The next gib update will be coming out within the next couple weeks, I hope ( Georgi & I are trying to shorten the release cycle ) and I'll do what I can