[lamford]

---

Matchpoints. Opening lead ♠K. Table result NS +120

There was a lucky landing by NS in this hand from a North London club last night, and East, who looks and behaves like the Secretary Bird, was apoplectic. North-South were playing four weak twos and EHAA and South decided that his partner might well respond on a very weak hand as South did not have a strong opening bid available. He was also regretting not having opened 2NT now, and decided that he would give his partner some leeway and bid only 2NT. North guessed that South had dredged up an invite on some 15 count, and decided to pass, as his 1NT could be up to 10 and he only had 9. East led the king of spades and the traveller revealed that NS had a complete top. One poor soul had been doubled in 3NT by a loony East who found the miracle ♠T9 tight with North. SB started his tirade against North. "You breached Law 73C and we were damaged", he began, "in that you did not carefully avoid taking any advantage of the UI. The fact that South was as heavy as Eric Pickles does not matter. You played for South to be stretching and got lucky". "I disagree", replied North, "the slow 2NT does not demonstrably suggest anything at all". "Partner could have been thinking of bidding 3NT, or forcing to game. You seem to a member of the 'if it hesitates, shoot it' school."

How do you rule?

[campboy]

I think the difference between this case and the one that presumably inspired it is that in this case, without UI, passing would be clearly wrong. When you get the UI that partner is either very minimum or very maximum, passing becomes more plausible. So passing is suggested.

In the other thread the poll shows that the right action is not clear without UI (two people bid 3NT; two passed; two were unsure). The extra information that partner is either minimum or maximum does not help at all. So in that case the UI does not demonstrably suggest anything.

[lamford]

 campboy, on 2014-November-26, 07:03, said:

When you get the UI that partner is either very minimum or very maximum, passing becomes more plausible. So passing is suggested.

It has to be demonstrably suggested rather than more plausible. I presume therefore you would adjust. And I am not sure which "other case" you mean.

[lamford]

 campboy, on 2014-November-26, 07:03, said:

<snip>The extra information that partner is either minimum or maximum does not help at all.

Surely it is more helpful in a marginal situation, and game theory says that one should work out the gain from action A when partner is minimum and the loss from action B when partner is maximum and select the call with the highest expected value. However, you are not allowed to know that partner is either minimum or maximum, but not bog standard.

[campboy]

 lamford, on 2014-November-26, 07:09, said:

It has to be demonstrably suggested rather than more plausible. I presume therefore you would adjust. And I am not sure which "other case" you mean.

Yes, I should have said "demonstrably", and I would adjust. I think the chance of pass being right goes up a lot once you know partner doesn't have a normal invite.

[campboy]

 lamford, on 2014-November-26, 07:14, said:

Surely it is more helpful in a marginal situation, and game theory says that one should work out the gain from action A when partner is minimum and the loss from action B when partner is maximum and select the call with the highest expected value. However, you are not allowed to know that partner is either minimum or maximum, but not bog standard.

No, it is less helpful in a marginal situation, because all the UI does is make the situation more marginal. As the hand in this thread illustrates -- without the UI it was clear what to do; with the UI it becomes unclear, and so passing goes from ridiculous to plausible.

[lamford]

 campboy, on 2014-November-26, 07:29, said:

<snip> No, [the UI] is less helpful in a marginal situation, because all the UI does is make the situation more marginal.

I cannot agree with that. Even in simple situations like 1NT (11-14) - 2NT (natural, but slow) we know that partner was almost certainly considering passing or bidding 3NT. All hands which are standard invites are eliminated, and the chance of partner being maximum for the invite has gone up to something like 50%, from an original value of perhaps 20%, depending on how many hands were in the medium range of standard invites.

A slow invitational bid can always demonstrably suggest one of bidding on or passing, depending on the expected value of both actions, and the change to that expected value by the illegal elimination of the middle.

[ArtK78]

I think it is as clear as this: Without any inference suggested by the hesitation, would North bid 3NT? I think the answer is clearly yes. The North hand is HUGE. 9HCP and good spots in every suit, including the T9 of partner's opening major suit. The hesitation put something else into the mix - the fact that South had something to consider other than bidding 2NT. Therefore, given that 3NT is the 99.9% action with the North cards in the absence of a hesitation, it is clear that the hesitation demonstrably suggested something else, and that is what North did.

Furthermore, given the NS system, North knows that he has to respond to South's opening bid on hands that others might pass, to cater to South having a hand that others might open 2♣. So, North knows that South will be careful about taking a second call such as 2NT with marginal values. These are inferences not available to others, and makes it more likely that North should bid game (in the absence of the hesitation).

Not only would I adjust, but I would consider some additional action against North - possibly C & E. This case is too much.

[Trinidad]

 lamford, on 2014-November-26, 09:22, said:

A slow invitational bid can always demonstrably suggest one of bidding on or passing, depending on the expected value of both actions, and the change to that expected value by the illegal elimination of the middle.

But the expected value of both actions is not UI.

In some cases it is easy to demonstrate what the UI suggests. E.g. after a slow sign-off, the UI suggests to bid on.

After a slow invite, the UI doesnot suggest what direction to take. The UI indicates that accepting or rejecting the invitation is right, but it is not possible to demonstrate which one. And that is not significantly different from where we would be without the UI.

Any theorizing about a shift in the ratio of the probabilities of "heavy invites" and "light invites" is futile. It assumes a much higher degree of estimation accuracy than is remotely realistic for a bridge situation.

Of course, you can talk about expected values and pull in the IMP table. But that is not what we are supposed to judge, because the IMP table is AI. We are supposed to judge what the BIT itself suggests. And after a slow invite a BIT doesn't demonstrably suggest accepting over rejecting the invitation, or vice versa.

Rik

[lamford]

 Trinidad, on 2014-November-26, 09:46, said:

But the expected value of both actions is not UI.

Because the expected value is the result multiplied by the percentage chance of partner having that hand, and one of the ingredients of that expected value is heavily affected by UI, the expected value becomes UI. The IMP table is indeed AI, as is partner's tendency to be aggressive or conservative, as these are both pieces of information that the player possessed before he removed his cards. However, if a BIT changes the likely percentages of different hand types to our advantage (and when game is in the equation it usually will change it upwards), bidding game tends to become demonstrably suggested by a BIT. Matchpoints is different of course.

[lamford]

 Trinidad, on 2014-November-26, 09:46, said:

After a slow invite, the UI doesn't suggest what direction to take.

So you would rule "score stands" in this example?

[campboy]

 lamford, on 2014-November-26, 09:22, said:

I cannot agree with that. Even in simple situations like 1NT (11-14) - 2NT (natural, but slow) we know that partner was almost certainly considering passing or bidding 3NT. All hands which are standard invites are eliminated, and the chance of partner being maximum for the invite has gone up to something like 50%, from an original value of perhaps 20%, depending on how many hands were in the medium range of standard invites.

A slow invitational bid can always demonstrably suggest one of bidding on or passing, depending on the expected value of both actions, and the change to that expected value by the illegal elimination of the middle.

True, but the chance of him being minimum for the invite has similarly gone up. These two things push in opposite directions, and in many cases they will about balance out.

What's the average height of a man in the UK? What's the new average if you eliminate the middle 50%? My guess is it wouldn't be very different, and I don't know if it would be higher or lower.

[Vampyr]

 ArtK78, on 2014-November-26, 09:28, said:

Not only would I adjust, but I would consider some additional action against North - possibly C & E. This case is too much.

I agree, and I don't even care what C&E means.

[barmar]

 Vampyr, on 2014-November-26, 10:21, said:

I agree, and I don't even care what C&E means.

Conduct and Ethics -- essentially charging North with cheating.

I think all this talk of what is demonstrably suggested is misguided. In this instance, I'm with SB, this was a 73C violation, not 16B. There's no good bridge reason for passing the invitation with 9 HCP, it can only be explained by taking inference from the UI.

Actually, I just thought of a bridge reason: maybe the state of the game suggested that they need a swing, so he took a low-percentage action. But if this had been the reason, I'd expect North to mention it. But I also expect the TD to view it suspiciously -- it's an incredible coincidence that he decided to do that at just the right time. If there's no tempo break, you get to keep your good luck, but when UI is around, self-serving explanations like that should be viewed suspiciously.

[Vampyr]

 Trinidad, on 2014-November-26, 09:46, said:

After a slow invite, the UI doesnot suggest what direction to take. The UI indicates that accepting or rejecting the invitation is right, but it is not possible to demonstrate which one. And that is not significantly different from where we would be without the UI.

OK, well do you place any weight on the fact that, after the UI, North did a truly bizarre action?

[barmar]

The difficult thing about this case is that the actual reason for the hesitation is the exact opposite of the reason that would suggest passing. If North knew that South was actually heavy for his bid, he would bid 3NT eagerly. It's just really bad luck that they can't make 3NT with 28 HCP. That's what makes it difficult to believe that North is "taking advantage" of the UI -- how can you be taking advantage of an inference if your inference is wrong?

[lamford]

 campboy, on 2014-November-26, 10:04, said:

True, but the chance of him being minimum for the invite has similarly gone up. These two things push in opposite directions, and in many cases they will about balance out.

However, the reward when partner is maximum rates to be 10 IMPs for a making a vulnerable game and the loss when partner is minimum might be only 6 IMPS (-100 instead of 140, say). If both have gone up, because of the UI, the player would be right to bid game. Which is why TDs routinely, and correctly, adjust when a slow invite leads to a successful game provided that biddding game is not the only LA. You often argue that something is demonstrably suggested when its chances go up from those absent the UI, even though it is still a massive underdog. There was a wild leap to 7S in another thread you may recall, where you were a lone wolf wanting to adjust, so I think your stance on this point is inconsistent.

[campboy]

 lamford, on 2014-November-26, 10:41, said:

However, the reward when partner is maximum rates to be 10 IMPs for a making a vulnerable game and the loss when partner is minimum might be only 6 IMPS (-100 instead of 140, say). If both have gone up, because of the UI, the player would be right to bid game. Which is why TDs routinely, and correctly, adjust when a slow invite leads to a successful game provided that biddding game is not the only LA. You often argue that something is demonstrably suggested when its chances go up from those absent the UI, even though it is still amassive underdog. There was a wild leap to 7S in another thread you may recall, where you were a lone wolf wanting to adjust, so I think your stance on this point is inconsistent.

You seem to be making the assumption that game is definitely making if partner is maximum and definitely going off if he is minimum. In that case, with UI you want to be bidding game (in the other thread at least; this one is matchpoints). But how likely do you think game is to make when partner's hand is in the middle? About 50% seems consistent with your assumptions, and so we want to be bidding game even without UI.

I don't think I'm being inconsistent at all. IIRC in the case you refer to, bidding 7♠ would be absurd without UI, and much more plausible (though still an underdog) with UI. Therefore it was demonstrably suggested, and I would adjust. In this thread, passing is absurd without UI, and much more plausible (though IMO still an underdog) with UI. As I already said, I would adjust here too. Wellspyder's case is not comparable IMO. There the choice between passing and bidding 3NT was close without UI, and I believe it is still close with UI; if the odds have changed then I'm not sure in which direction they've gone.

[Trinidad]

 lamford, on 2014-November-26, 09:53, said:

Because the expected value is the result multiplied by the percentage chance of partner having that hand, and one of the ingredients of that expected value is heavily affected by UI, the expected value becomes UI.

Sure... but this is a misrepresentation of what is going on: The expected value is not heavily affected by UI. It is slightly affected by UI in a direction that is indeterminable, because it depends on too many unknowns (what hands will partner tank with, on what hands will he just bid game, how aggressive does partner normally invite). The idea that a regular partner can figure out what kind of hand partner must hold for his BIT in an invitational situation is simply wrong.

So, you are trying to quantify something that you can't quantify. It is like measuring the diameter of a cell with a yardstick.

Rik

[blackshoe]

 lamford, on 2014-November-26, 07:09, said:

It has to be demonstrably suggested rather than more plausible. I presume therefore you would adjust. And I am not sure which "other case" you mean.

The wording is "could demonstrably have been suggested". So to demonstrate, you have to show the logic by which whatever action you're examining could be suggested by the UI.