[Andy_L]

Apologize in advance for not posting hand here... Thing is I don't know to "Import" a hand to here. Hand is Hand #7 from: "#6408 Robot Duplicate -(MP)".

3♦ is played by S (29) times. In every case when Declarer (S) plays ♦'s the sequence is A♦ followed by a 2♦ to dummy. In (19) cases West plays small ♦ followed by small ♦. In (10) cases West plays small ♦ followed by Q♦... giving away the ♦ suit. What logic can possibly support GIB's play of the Q♦?

In a previous post a reply was given saying (paraphrase): In tournament play the GIB's will act the same.

[mgoetze]

Traveller: http://www.bridgebas...username=andy_l

---

---

The difference as you see is the play of the ♦9 vs. the ♦2 at trick 3. GIB's play is based on double dummy analysis. With the ♦9 it is 100% obvious to GIB that it doesn't make a difference whether it plays the queen or the 3, so it will choose more or less randomly between these cards (given that it doesn't simulate any entry problems either way). With the ♦2 it is just a bit less obvious. In any case the simulations diverge at this point so even if it is still random you might get a different result.

Now, to see whether you understood what I wrote, here is a quiz. What is the best way to play the following suit for 5 tricks vs. GIB?

Dummy: K6542
Declarer: AJT3

[barmar]

That's correct -- small differences in declarer play can affect the hands in the simulations, so the robots will play differently. I had a similar thing come up last night.

---

Most tables had the same auction, and West started with two top ♣ and a ♣ ruff. Then it's up to East to find the right card to get back to partner for another ruff. My opponent found the killing ♦ exit, so I went down; others switched to ♥J and allowed it to make with an overtrick (one allowed two overtricks by later discarding a ♥). It all depended on the specific clubs that declarer played on the first 3 tricks.

When I was trying to figure out what happened, I had to look at some of the hands 3 or 4 times before I noticed this difference.

[Bbradley62]

barmar, on 2012-April-05, 08:16, said:

... West started with two top ♣ and a ♣ ruff. Then it's up to East to find the right card to get back to partner for another ruff. My opponent found the killing ♦ exit, so I went down; others switched to ♥J and allowed it to make with an overtrick (one allowed two overtricks by later discarding a ♥). It all depended on the specific clubs that declarer played on the first 3 tricks.

Wouldn't it be cool if West gave suit preference by the card he led on Trick 3, and if East used this information to determine which red suit to return?

[barmar]

Bbradley62, on 2012-April-05, 14:31, said:

Wouldn't it be cool if West gave suit preference by the card he led on Trick 3, and if East used this information to determine which red suit to return?

Where do you come up with these crazy ideas?

[pigpenz]

barmar, on 2012-April-05, 14:33, said:

Where do you come up with these crazy ideas?

I tried to post this earlier that with the 235♣ in dummy
most humans leading the 4♣ back for partner to ruff would mean this as some sort of signal
God only knows what GIB means in these situations.

[barmar]

pigpenz, on 2012-April-05, 18:25, said:

God only knows what GIB means in these situations.

It doesn't mean anything. GIB doesn't make or read suit preference signals.

[pigpenz]

barmar, on 2012-April-06, 08:13, said:

It doesn't mean anything. GIB doesn't make or read suit preference signals.

unfortunately I know...I have been trying to signal him for over a year


Question:
on opening leads what does GIB do when partner plays a card that most of us humans
would recognize as count or unblocking?

eg
leading A against 3NT asking for unblock of Q
or leading Q from KQ109X asking for J unblock.

or maybe I'll just start another thread on this topic cause I see it happen all the time.

[barmar]

GIB doesn't read signals. That's it.

It takes very few inferences from the plays of cards by partner and and opponents. It can't even figure out that when honors crash, it means partner had a singleton.

The only inferences it knows how to take are when a player makes a standard honor lead -- if you lead Q through K to its A, it will usually figure ut that you have the J.

[pigpenz]

barmar, on 2012-April-06, 09:21, said:

GIB doesn't read signals. That's it.

It takes very few inferences from the plays of cards by partner and and opponents. It can't even figure out that when honors crash, it means partner had a singleton.

The only inferences it knows how to take are when a player makes a standard honor lead -- if you lead Q through K to its A, it will usually figure ut that you have the J.

one of the things that I continually see happen is blocking suits

[barmar]

pigpenz, on 2012-April-06, 16:11, said:

one of the things that I continually see happen is blocking suits

And humans never make that mistake, do they?