[daveharty]

I've noticed that GIB often plays an honor in second position and solves a guess for declarer. This was the worst example I've seen lately (trick 3):

http://tinyurl.com/7ydry63

Pretty sure I was getting that one wrong otherwise.

[manudude03]

The problem is that GiB simulates double dummy and so it expects you to finesse and so the play doesn't matter.

[HighLow21]

The problem with THAT is that it makes things double-dummy that are anything but. But I cannot imagine a more worthy recipient of some good fortune from bad programming than my good friend Dave!

Maybe on defense, GiB takes into account that the declarer may be a first-rate class act in his dealings with people.

[daveharty]

manudude03, on 2012-April-20, 11:30, said:

The problem is that GiB simulates double dummy and so it expects you to finesse and so the play doesn't matter.

Fair enough, but wouldn't it make sense to have the default card be a low one in situations like this? It seems like GIB consistently plays the highest card here, or is it randomized when the play doesn't matter DD?

Tate: thanks for the vote of confidence but I was definitely not guessing this one...had GIB played low smoothly.

[Stephen Tu]

It's randomized when it doesn't matter DD. It's kind of hard to teach GIB to give declarer guesses when fundamentally it believes that declarer is never guessing.

I think Ginsberg was thinking along the lines of getting GIB on defense to simulate how a *single dummy* declarer would play, rather than a *double dummy*, which would then in theory get the right card to be played and avoid this kind of thing. But solving lots of single dummy problems is horrifically slow.

[mike777]

If we accept, per other threads, that GIb is already in the top 50% bridge level of ACBL players maybe we should just shoot for the top 40% this year.
It sounds just increasing the computer power/speed will make Gib a better player over time.

[pigpenz]

the only way you might get around this is if you have an old version of BBO
you could set the thinking time to maximum, but yes GIB does do this at times
usually it seems to happen when he didnt do it against you but against someone else
and you notice it when you look at results.
I have always assumed it was from the different types of GIB versions out there.

[Bbradley62]

How about this second-hand-high play (on Trick 2)?

---

[Bbradley62]

Yet another, with lots of IMPs being thrown around as a result...

---

Like Dave, I'd like to see GIB's logic be "if you think it's irrelevant, play low".

[Bbradley62]

And yet another one... watch East play his diamonds...

---

[Bbradley62]

And West on Trick 10:

---

[semeai]

Are these the advanced bots or the basic bots (or a mix)?

[Bbradley62]

Everything I've posted is an advanced bot in tournament play.

[rwbarton]

Bbradley62, on 2012-September-09, 16:31, said:

And West on Trick 10:

---

GIB's play here is correct, there are two layouts (swap ♦9 for ♦8 or ♦7) where the unblock is necessary to hold declarer to two of the last four tricks.

[Bbradley62]

rwbarton, on 2012-September-10, 08:00, said:

GIB's play here is correct, there are two layouts (swap ♦9 for ♦8 or ♦7) where the unblock is necessary to hold declarer to two of the last four tricks.

That layout seems unlikely since declarer didn't run diamonds after West sacrificed his 10. But then again, he didn't cash the fourth club either, so it's hard to tell what's going on.