[p_t_red]

North: ♣Q8 ♦QT3 ♥T962 ♠A976
East: ♣J5 ♦K ♥J843 ♠KQT842
South: ♣AT976 ♦A87654 ♥A ♠3
West: ♣K432 ♦J92 ♥KQ75 ♠J5

The Auction:
West North East South
____ Pass Pass 1♦
Pass 1♥ 1♠ 2♣
Pass 2♦ 2♠ 3♦
3♠ 3NT Pass Pass
Pass

Okay, the auction wasn't great. 8♠ lead by East.
♠E8 S3 WJ NA
♦N3 EK SA W2
♦S8 W9 NQ ♥E4
The problem here is that East played the King on the first diamond.
Holding KJ the proper play on the first round is the Jack,
so North has a "proven" finesse against West for the Jack
on the second round.

[cloa513]

p_t_red, on 2011-September-29, 21:30, said:

North: ♣Q8 ♦QT3 ♥T962 ♠A976
East: ♣J5 ♦K ♥J843 ♠KQT842
South: ♣AT976 ♦A87654 ♥A ♠3
West: ♣K432 ♦J92 ♥KQ75 ♠J5

The Auction:
West North East South
____ Pass Pass 1♦
Pass 1♥ 1♠ 2♣
Pass 2♦ 2♠ 3♦
3♠ 3NT Pass Pass
Pass

Okay, the auction wasn't great. 8♠ lead by East.
♠E8 S3 WJ NA

♦N3 EK SA W2
♦S8 W9 NQ ♥E4
The problem here is that East played the King on the first diamond.
Holding KJ the proper play on the first round is the Jack,
so North has a "proven" finesse against West for the Jack
on the second round.

Read the thread and post the hand.
http://www.bridgebas...ting-gibs-bugs/

[cloa513]

cloa513, on 2011-October-01, 07:22, said:

Read the thread and post the hand.
http://www.bridgebas...ting-gibs-bugs/

Here is the hand

[barmar]

Was this a basic or advanced robot?

[cloa513]

barmar, on 2011-October-01, 22:38, said:

Was this a basic or advanced robot?

Why would either robot type simulate that East would false card with KJ of diamonds?

[barmar]

Not sure. But I know I've seen the robots get this right, and I only play with advanced robots.

[rduran1216]

http://tinyurl.com/3jj9aet

hahahaha

[barmar]

That's different. West could have ducked with Kx, trying to get an extra undertrick if declarer started with a doubleton. In general, GIB doesn't assume that just because a finesse worked the first time, it will work the next time. Of course, what the GIB declarer doesn't know is that a GIB defender will never duck like this, because it bases its defense on double dummy analysis, so it thinks declarer will play the Ace next time.

But there's unlikely to be a reason in the original hand for the defender to play K from KJ, unless it needs to avoid being end-played.

[cloa513]

When GIB simulates the cards for each play, does it simulate every single card or just the suit involved?

[barmar]

I'm not sure I completely understand your question, but I presume it simulates every LEGAL card to play.

[Bbradley62]

Here's another one, at matchpoints, with a basic bot:

---

[uday]

This one was interesting so I dug ( then scratched my head, then asked around for help )

The basic bot uses the weaker of GIB's two internal play-card engines ( bec. this engine is faster ). This engine simply deals out the enemy cards and takes the best DD action based on the form of scoring. It isn't smart about some things like this. It simply recognizes that an original Kx with north occurs more frequently than Kxx or Kxxx with south in the sample deals it uses to decide what to play for.

[Bbradley62]

retracted messy stats stuff... much ado about nothing... Thanks for your help, Uday. Just looks like basic GIB's small sample size produced a strange result.