GIB bug

[greiman]

This was during one of the Robot Reward Tourneys, so unfortunately, I don't have the complete hand record. GIB and I had a quick auction to ourselves landing in 6NT, and after a small diamond lead this is what GIB sees:

 

AQJ862, J, AT6, AK4

 

K7, AQ942, K92, J75

 

The initial diamond lead went around to east's DJ which GIB wins.

 

The obvious line seems to be a club to the ace followed by a heart finesse. Assuming the spades come in, that's 12 tricks if it wins, and 12 tricks when it loses. Granted, there might be value in cashing a spade first to see if we need both red finesses (or a doubleton club) if we only have 5 spade tricks. But GIB didn't do that either. The play went something like this:

 

1. Small diamond to the jack and king.

2. Small diamond to the ace (!).

3-8. Lots of spades.

9. Heart finesse losing.

10. DQ cashed.

11. I don't remember and didn't check.

 

Sorry for the lack of specifics, but I didn't know there was a way to report a bug directly from the hand record.

[lycier]

Hi

Welcome to Gib robot forum.

Of course, it must need the complete hand record to show its issue, as a reader, you have no right to decide whether it is a bug, so you would better report to BBO.

 

http://i64.tinypic.com/1zq4hf9.jpg

 

 

There are two reporting methods.

 

First method :

You can click the hand " Options " on your right side, then click " Send to member...", send your hand to fred,uday or georgi.

 

Second method :

You can click the hand " Options " on your right side, then click " handviewer link ", eventually you post it on the Gib robot forum.

[1eyedjack]

Having enough information to confirm that it is a bug and having enough information to correct it are two entirely different things.

 

Looks to me that there was enough info there to conclude with confidence that it is a bug.

 

But to correct it the programmers would almost certainly need the full hand record.

 

That said, bugs in the card play algorithm appear to be hard to correct, if not impossible. I very much doubt that this is correctable, and we just have to live with the fact that GIB makes bad plays occasionally and all that this thread does is to confirm what we all already knew: occasionally GIB throws a trick.

[johnu]

Of course, it must need the complete hand record to show its issue, as a reader, you have no right to decide whether it is a bug, so you would better report to BBO.

 

Obviously a bug just looking at the 2 hands because 12 tricks are guaranteed as long as spades aren't 5-0 by taking the heart finesse. Playing a 2nd diamond is a hopeless play which will survive only if the heart finesse works or if the heart finesse loses and that hand is out of diamonds. I don't see any bridge reason to play a diamond to the ace at trick 2.

 

This assumes the OP correctly remembers the hand and the sequence of plays.

 

Under My BBO/Hands and Reports, select a hand and click on Export deal, and then Send robot report. There is a text box where you can describe the problem you saw when playing the hand.

 

That being said, the play engine is pretty much a black box written by the original programmer who is not longer available and changes to the play algorithms are apparently almost impossible to make, so don't expect any changes.

[lycier]

Obviously a bug just looking at the 2 hands because 12 tricks are guaranteed as long as spades aren't 5-0 by taking the heart finesse. Playing a 2nd diamond is a hopeless play which will survive only if the heart finesse works or if the heart finesse loses and that hand is out of diamonds. I don't see any bridge reason to play a diamond to the ace at trick 2.

 

This assumes the OP correctly remembers the hand and the sequence of plays.

 

You said " Obviously a bug ..."?

Really? We need the hand record, the evidence is very important.

If this matter happened before March 15, Greiman reporting would make no sense since it is a matter of version 34.

And I checked Greiman's hand records in a month, I have not found related hand. Would you tell us playing time?

[lycier]

If it is really a bug without the hand records, where is this bug? how to fix it ?

 

So I strongly think your comments above are empty words, useless and unfair .

[1eyedjack]

Put it this way:

Are you able to construct any complete hypothetical deal consistent with the information provided wherein GIB's play is not a bug? One example would suffice. Then it would be justified to seek the compete hand record. If you find it impossible to construct such a hand, then your criticism of its characterisation as a "bug" (in particular for want of the complete deal) is hot air.

 

Personally I cannot construct such a deal and doubt that anyone could. Hence my acceptance that it is a bug. But I do not possess the wisdom of Solomon. Maybe you can. I shall keep an open mind.

[Bbradley62]

Obviously a bug just looking at the 2 hands ...

 

This assumes the OP correctly remembers the hand and the sequence of plays.

You said " Obviously a bug ..."?

Really? We need the hand record, the evidence is very important.

Since JohnU has included the caveat "this assumes...", his statement is entirely correct.

 

If this matter happened before March 15, Greiman reporting would make no sense since it is a matter of version 34.

Although the announcement of Version 35 includes "Other various bug fixes and improvements", there is no mention of any change to card-play features and these generally get mentioned (such as cashing A when defending 7N). It is almost certain that any card-play bug that was in Version 34 is also in Version 35. In the extremely unlikely case that this was addressed in the Version 35 improvement, OP has done no worse than given developers a terrific opportunity to say "we fixed that".

 

And I checked Greiman's hand records in a month, I have not found related hand. Would you tell us playing time?

Greiman very clearly stated that this hand occurred in Robot Rewards Tourney. Why on Earth were you expecting to find it in his hand records?

 

 

If it is really a bug without the hand records, where is this bug? how to fix it ?

 

So I strongly think your comments above are empty words, useless and unfair .

If support staff is actually reading these posts (and I see no evidence that they are) they can probably find the hand (or reproduce the scenario) based on OP's information. Yes, it would more convenient if OP had posted the hand, but your characterization of the original post is off-base.

[lycier]

Bbradley62, you meant that my method is wrong, the characterization of your method is more reliable?

Here I take a example :

"Sb never know Gib CC since there is a Gib hand to show sb play rediculously in the MBC, so sb is a novice ."

How about it? According to your reliable method,this comment also includes hand records, this words is not off-base?

Then if sb as me what time it happened, what diagramm? the answer is " ignore it" .

How about it? is it not off-base?

 

In a word, if you strongly think your method is more reliable,please take out your way. A pity I have no idea what characterization you have, your goal is to raise objection? Take your statistical characterization instead of just yell .

[greiman]

Wow, these topics move faster than I expected.

 

As some of you have surmised, this hand was played during a Robot Rewards Tourney, which doesn't have accessible hand records afterwards. After the round, I pulled up the movie for the hand, and emailed a friend of mine bemoaning my poor luck. A day or two later, I decided to post here with the scant details that I had left from that email (since the movie was no longer available).

 

I'm aware that GIB isn't a world class player and will occasionally drop a trick here and there. But this seemed like an egregious error with nothing to gain and everything to loses, so it felt worth reporting. Apparently there's a fancy way to do that now. I'll do that in the future.