Computer dealt hands - with a hiccup

[blackshoe]

That looks familiar, Gordon. In fact, it looks very much like what I said early in this thread. :lol:

[nige1]

LOL.  Echo, I think you should give up trying to get an answer.  Nobody wants to accept that there was a human error and fouled board(s). The presence (if true) of a hand record showing both hands as they were only means the human error occurred at a different point. Even if it was a computer malfunction, to allow continued play of hands where the other players' hands are known or knowable cannot be right.  Continued shooting of the messenger cannot be helpful, either.

I agree that Echognome behaved impeccably, confiding his suspicions to the director alone. I still believe that there remains a problem with pseudo-random hand generation. If you examine a set of boards and something seems peculiar to you, then where do you draw the line? Examples ...

• The North hands have a lot of tens.
  
• The West hands have a lot of voids.
  
• Two hands are all one suit.
  
• Several hands have four aces.
  
• There are lots of nine card suits.
  
• North-South consistently hold more HCP than East-West.
  
• Two consecutive West hands are the same.
  
• Two deals are the same.
  

Had you predicted any of these, you would have received good odds against their occurrence. After the event, however they are no less likely than any other specific set of boards. (Except in human hind-sight). When computer-deals were introduced, directors sometimes redealt boards that seemed anomalous to them. Unfortunately, however, if you retrospectively cull deals, you no longer have random boards.

[gordontd]

That looks familiar, Gordon. In fact, it looks very much like what I said early in this thread. :wacko:

Yes, you're quite right. It seems I'm in good company.

[mjj29]

Examples ...

• The West hands have a lot of voids.
  
• Two hands are all one suit.
  
• Several hands have four aces.
  
• There are lots of nine card suits.
  

Sounds like the boards I generate for the christmas party :wacko:

[pran]

LOL.  Echo, I think you should give up trying to get an answer.  Nobody wants to accept that there was a human error and fouled board(s). The presence (if true) of a hand record showing both hands as they were only means the human error occurred at a different point. Even if it was a computer malfunction, to allow continued play of hands where the other players' hands are known or knowable cannot be right.  Continued shooting of the messenger cannot be helpful, either.

I agree that Echognome behaved impeccably, confiding his suspicions to the director alone. I still believe that there remains a problem with pseudo-random hand generation. If you examine a set of boards and something seems peculiar to you, then where do you draw the line? Examples ...

• The North hands have a lot of tens.
  
• The West hands have a lot of voids.
  
• Two hands are all one suit.
  
• Several hands have four aces.
  
• There are lots of nine card suits.
  
• North-South consistently hold more HCP than East-West.
  
• Two consecutive West hands are the same.
  
• Two deals are the same.
  

Had you predicted any of these, you would have received good odds against their occurrence. After the event, however they are no less likely than any other specific set of boards. (Except in human hind-sight). When computer-deals were introduced, directors sometimes redealt boards that seemed anomalous to them. Unfortunately, however, if you retrospectively cull deals, you no longer have random hands.

This is why regular and comprehensive statistic testing of the outcome from such programs is important, and also why it is important to examine particularly what happened when such unacceptable results were produced.

 

It is possible to throw six sixes in a row with a dice (it can be expected once in 7776 trials), but if that happens daily in a gambling house any serious house would take that dice out of service and examine it very carefully before admitting it into service again.

 

Similarly a computer dealing program that occationally exhibits strange results should be examined (together with its logs) to see if there is some flaw waiting to be revealed.

 

WHAT! There are no logs available that can be examined afterwards to find out exactly what happened inside the program?

 

Oh dear.

[thomaso]

I believe that Thomas Andrews' dealing program will generate all possible hands (eventually). Richard Pavlicek's dealer is similarly excellent and fast. They both have all kinds of fascinating stuff!

My "Deal" program does not, in fact, generate all possible deals. My Impossible Bridge Book does, but "Deal" does not use that code. Instead, it is engineered to generate lots of samples quickly.

[helene_t]

If you examine a set of boards and something seems peculiar to you, then where do you draw the line? Examples ...

[*] The North hands have a lot of tens.

[*] The West hands have a lot of voids.

The best advice is always to assume that the cards were dealt randomly, even in the presence of extreme statistical abnormalities. The thing is, people find patters where there are none.

 

Even in this case, where two boards were identical down to the last pip, I think there is a case for treating it as a coincidence and just let the boards be played:

- Most players didn't notice.

- A software bug could be considered a "random" event (as long as we aren't talking about well-known software bugs causing predictable abnormalities).

- We don't want to create precedence for directors with a bad intuition for probabilities (and most everyone has very bad intuition for probabilities) to reshuffle boards whenever he thinks something looks strange. For every genuine bug in a dealing program there will be literally thousands of false positives.

[blackshoe]

I take your point about setting a precedent, Helene, but as Gordon said, here we have a player who has extraneous information about board ten. Are we to simply to give an artificial adjusted score on this board, and hope that anyone else who recognizes the problem calls the TD so he can get the same, or do we let them play on, with that extraneous knowledge?

[helene_t]

I take your point about setting a precedent, Helene, but as Gordon said, here we have a player who has extraneous information about board ten. Are we to simply to give an artificial adjusted score on this board, and hope that anyone else who recognizes the problem calls the TD so he can get the same, or do we let them play on, with that extraneous knowledge?

As it turned out, the complete deal (including the two hands he couldn't see) were identical down to the last pip.

 

I think as a TD, I would just assume the much more likely case namely that the two deals were just very similar (maybe down to two or three pips) and that it was just a coincidence so that the two hands he couldn't see might be completely different.

 

And then tell the player to play on and say he could use information from the previous deal if he wanted, if he really thought the other two hands were also identical, which I would not know. (For all we know the software bug might have produced two deals that were identical w.r.t. two hands but not w.r.t. the other two).

 

I would then be proved wrong afterwards, and would be torn between canceling the board, or assigning a splitscore so that this player's opponents were not damaged, or whatever.

 

But I think in most such cases I would be right.

[blackshoe]

If I'm not mistaken, the table TD determined that the two boards were indeed identical down to the last pip before he made his table ruling. Given that knowledge at that time, would you still make the same ruling?

 

If you did so rule (let's say you didn't know they were identical) and it later turned out your assumption was wrong, is that not TD error, requiring you to adjust the score treating both sides as non-offending?

[helene_t]

If I'm not mistaken, the table TD determined that the two boards were indeed identical down to the last pip before he made his table ruling. Given that knowledge at that time, would you still make the same ruling?

No I don't think so.

If you did so rule (let's say you didn't know they were identical) and it later turned out your assumption was wrong, is that not TD error, requiring you to adjust the score treating both sides as non-offending?

Yes I suppose so.

[axman]

I take your point about setting a precedent, Helene, but as Gordon said, here we have a player who has extraneous information about board ten. Are we to simply to give an artificial adjusted score on this board, and hope that anyone else who recognizes the problem calls the TD so he can get the same, or do we let them play on, with that extraneous knowledge?

The problem is that once the player gives any indication that something is amiss there is the probability that the partner will draw inference. Certainly, once the TD is summoned I am confident that I would include that inference in the reasons for the call. And, if I will then so will others.

 

And thus is the situation:

 

Until the player gives any inkling of his suspicions, what he has are suspicions. These suspicions might eventually be born out as fact or fancy. If the hand was indeed legitimate then he is entitled to any outcome that he earns at his own risk. If not legitimate then no one is [or at least ought to be] entitled to earn any outcome.

 

However, once the inference comes available the defenders are not in a position to earn anything because there has been a communication to the PARTNER. Actually, the law approaches being unsuitable for remedying this situation- and to the best that I am able to discern the only applicable passages lie in L16/73.

 

From personal experience there was the occasion in a shuffle and play swiss when I held a hand in round 4 that I would have been willing to wager a large sum was identical to a hand from the same board that I held during R3 [having retained the same table and thus the same boards for those two rounds]. However, while willing to wager on my hand I was not nor would not be willing to wager any sum on the other hands sight unseen.

 

On the first occasion I deviated from system which led to an apparently thin game. On the second occasion I duplicated my efforts to the same effect. After completion of play I informed the TD that the board was identical to the previous round. This had several effects [a] he accused me of cheating he ascertained that the other table had in fact not shuffled it (the other boards were ok) prior to play [c] the result was canceled and a substitute was played.

 

As for my personal example I feel that [with the exception of the TD behavior] it was handled as it ought to be handled. Though in law that assertion is questionable. I am unable to find justification for correcting the shuffle once the board is played. [L6D2 specifies that the result be canceled <the board was played in the previous session/round> but does not specify a basis for correcting the shuffle].

 

Notably L15B specifies:

B. One or More Players Have Previously Played Board

 

If any player plays a board he has previously played, with the correct opponents or otherwise, his second score on the board is cancelled both for his side and his opponents, and the Director shall award an artificial adjusted score to the contestants deprived of the opportunity to earn a valid score.

 

That the result be canceled but does not provide a mechanism for correcting the shuffle.

[bluejak]

When a board is cancelled in a team game it is normally a matter of regulation not Law what is done to the actual board.

[jdonn]

Helene has nailed it. Why is a director adjusting a board based on what he considers an extreme coincidence? There is no way to know where to draw the line.

 

What if the 4 hands were the same as the other board but completely rearranged (north and east hands from the other boards are partners on this one)?

 

What if just one pip is different?

 

What if the hands were the same as the previous board but with the hearts and diamonds of all players reversed?

 

These are questions that can't be fairly answered. The fact is even if the board was completely identical to a prior board, then even though it seems like a computer error that layout is as likely as any other particular layout occuring. There is no reason for the director to do anything.

 

The player doesn't have extraneous information. All we can prove is he thought he had extraneous information but may have made a lucky guess.

[pran]

If I'm not mistaken, the table TD determined that the two boards were indeed identical down to the last pip before he made his table ruling. Given that knowledge at that time, would you still make the same ruling?

 

If you did so rule (let's say you didn't know they were identical) and it later turned out your assumption was wrong, is that not TD error, requiring you to adjust the score treating both sides as non-offending?

In order to answer your question here you will first have to consider the following situation:

 

In a Mitchell or Howell event all deals have been dealt after solid shuffling by the players at the respective tables. There is just one copy of each deal and there is no reason to suspect any improper handling of the cards by any player.

 

Eventually it is discovered that two different deals are almost identical, for instance only differing in the allotted direction for the respective hands. And the way the dealing has been performed this similarity has evidently happened just by chance.

 

Do you cancel one of the boards?

 

Whether yes or no, on what legal ground?

[Echognome]

Helene has nailed it. Why is a director adjusting a board based on what he considers an extreme coincidence? There is no way to know where to draw the line.

 

What if the 4 hands were the same as the other board but completely rearranged (north and east hands from the other boards are partners on this one)?

 

What if just one pip is different?

 

What if the hands were the same as the previous board but with the hearts and diamonds of all players reversed?

 

These are questions that can't be fairly answered. The fact is even if the board was completely identical to a prior board, then even though it seems like a computer error that layout is as likely as any other particular layout occuring. There is no reason for the director to do anything.

 

The player doesn't have extraneous information. All we can prove is he thought he had extraneous information but may have made a lucky guess.

Let's be somewhat practical. The reason the director's knew that the board had been fouled is because the second board did not match the hand records when they checked. It had, in fact, matched the first board, but rotated. I am not 100% sure, but I'm pretty sure it was due to human error, not computer error.

 

So taking it as a fact that the board was fouled, how do you rule? Play on because the board is not identical? Doesn't that sound silly?

[gordontd]

Why is a director adjusting a board based on what he considers an extreme coincidence?

Because he doesn't consider it an extreme coincidence - the likelihood is so much greater that there was a shortcoming in the dealing process. If one were completely satisfied that there was no such shortcoming, and the two (identical but rotated) boards were properly dealt, then the board results should stand. But it's so unlikely that it's reasonable to doubt it in the absence of very strong supporting evidence.

[jdonn]

Helene has nailed it. Why is a director adjusting a board based on what he considers an extreme coincidence? There is no way to know where to draw the line.

 

What if the 4 hands were the same as the other board but completely rearranged (north and east hands from the other boards are partners on this one)?

 

What if just one pip is different?

 

What if the hands were the same as the previous board but with the hearts and diamonds of all players reversed?

 

These are questions that can't be fairly answered. The fact is even if the board was completely identical to a prior board, then even though it seems like a computer error that layout is as likely as any other particular layout occuring. There is no reason for the director to do anything.

 

The player doesn't have extraneous information. All we can prove is he thought he had extraneous information but may have made a lucky guess.

Let's be somewhat practical. The reason the director's knew that the board had been fouled is because the second board did not match the hand records when they checked. It had, in fact, matched the first board, but rotated. I am not 100% sure, but I'm pretty sure it was due to human error, not computer error.

Sorry I didn't see that, if I had then that's of course different. That's what I get for entering a thread late.

[jdonn]

Why is a director adjusting a board based on what he considers an extreme coincidence?

Because he doesn't consider it an extreme coincidence - the likelihood is so much greater that there was a shortcoming in the dealing process. If one were completely satisfied that there was no such shortcoming, and the two (identical but rotated) boards were properly dealt, then the board results should stand. But it's so unlikely that it's reasonable to doubt it in the absence of very strong supporting evidence.

You make it seem like you don't know what a coincidence is...

 

You say essentially "he doesn't consider it a coincidence, he just believes what occured is so unlikely that it almost couldn't be true." That would be like me saying "this is not an online bridge forum, it's just an internet-based vehicle for people to discuss bridge-related topics."

[gordontd]

You say essentially "he doesn't consider it a coincidence, he just believes what occured is so unlikely that it almost couldn't be true."

No. I say that this explanation for what occurred is so unlikely that I believe there's another explanation. You think that's what a coincidence is?

[aguahombre]

this is like the word "preemptive" we discussed a while back.

"coincidence", of course. coincidental, due to error by man or machine, or due to a random miracle.

[barmar]

The computer dealt the same hand twice.

Apparently not, according to Echo's post.

[pran]

Let's be somewhat practical.  The reason the director's knew that the board had been fouled is because the second board did not match the hand records when they checked.  It had, in fact, matched the first board, but rotated.  I am not 100% sure, but I'm pretty sure it was due to human error, not computer error.

 

So taking it as a fact that the board was fouled, how do you rule?  Play on because the board is not identical?  Doesn't that sound silly?

This is not what we have been told:

 

In a follow-up OP told us that: The computer dealt the same hand twice.

 

(In another post we have been told that the computer printouts were checked.)

 

Of course if hand records reveal that the board in question is not according to the deal the computer had created then that board has not been dealt in accordance with law 6 (more precisely in this case: With Law 6E4). Thus the board is fouled, end of story.

 

(And the computer program is probably in the clear)

[pran]

Sorry I didn't see that, if I had then that's of course different. That's what I get for entering a thread late.

Relax.

 

You didn't see it because it wasn't there.

[cherdanno]

If you really think this is coincidence then you have no idea how big the number of possible bridge deals is. The fact that every other deal is equally unlikely doesn't matter: for the given deal there is an alternate explanation (same board dealt twice) that is not completely unlikely which does not exist for other deals. That alternate explanation is more likely than the coincidence by a huge factor.

 

Btw, I wouldn't let the earlier results stand. Even if these players didn't notice it explicitly, they may have benefitted without realizing it (e.g. "learning" how to play a certain suit combination on the previous board).