[blackshoe]

 GreenMan, on 2013-November-20, 07:29, said:

In my example, obviously if LHO opens e.g. 1♦ instead of passing and partner overcalls 1♠, everyone would agree that I have AI that he has spades and values, and UI that he is not in the 8-11 range (or whatever distinguishes opening and overcalling hands in our partnership). The overcall and opening bid convey different, overlapping, sets of information, and the parts outside the intersection of those sets are UI.

But I would think, and it seems others agree, that if the two calls convey the SAME set of information that the whole set is AI. But Ed, at least, seems to be arguing that some part of that set is UI, and I'm supposed to sort out exactly what it is, and thus what actions it suggests. So how does a player do that?

By not thinking to himself "I have AI that suggests the same thing as the UI, therefore I can ignore the UI".

[GreenMan]

 blackshoe, on 2013-November-20, 07:39, said:

By not thinking to himself "I have AI that suggests the same thing as the UI, therefore I can ignore the UI".

But what is the UI that cannot be ignored? Obviously from your previous post partner's inattention or impatience is UI, but that's trivially obvious; can there really be anything more? You've said a couple of times "There is still UI" regardless of partner's legal actions, but what is it? Should I have the Los Alamos National Laboratory of Hair Splitting on speed-dial?

[PeterAlan]

 aguahombre, on 2013-November-19, 23:09, said:

When a player has both, and they lead to the same conclusion, he is using the AI and not the UI.

What's the basis for this assertion?

 campboy, on 2013-November-20, 04:31, said:

If the UI is duplicated by the AI then the UI gives you no extra information. So in fact, while it may be U, it isn't really I.

I don't think this holds water. After all, you could just as well write it substituting 'A' for 'U' and vice versa. If I get the same news from both the BBC and a newspaper, it doesn't mean that the information content is nil, but that's what you'd get from the conjunction of both versions of your AI/UI statement. Of course, you're really talking about extra information in your second sentence, and what you're saying is essentially the same as aguahombre.

 campboy, on 2013-November-20, 05:52, said:

The issue is not whether it's an LA, the issue is whether it's suggested. And if the UI adds no new information then it can't suggest anything.

This is clearly wrong. Any information potentially suggests something, and its novelty has nothing to do with it. It may be suggesting the same thing as other information, but it doesn't cease to suggest it.

I thought - silly me - that the general consensus was that when UI was created then the UI Laws (16B and 73C insofar as it's UI from partner) kicked in, and that they did not somehow magically disappear (cease to apply) when the same information subsequently became available from authorised sources, since there's nothing in the Laws to suggest that they do so cease (BTW I am aware that the same reasoning also leads to the conclusion that the same consequences follow even if the UI follows the AI).

You seem to be suggesting otherwise. Is there in fact a clear consensus?

[campboy]

What do you think it means for the UI to suggest alternative A over alternative B? I think it means that the difference between having the UI and not makes alternative A more attractive relative to alternative B. But if you have the same information from an authorised source, the difference between having the UI and not is nil.

The information content of the AI given the UI and that of the UI given the AI are both zero, as you say (provided the AI tells us no more than the UI, which is not always the case). But only one of these matters. In applying the law we consider the hypothetical situation in which the player only has AI. It does not make sense to consider the situation in which he only has UI.

This is exactly why we are not normally constrained by partner's answers to questions. Provided the answer is what we expect, they give no extra information and so cannot suggest anything.

[GreenMan]

 campboy, on 2013-November-20, 08:20, said:

This is exactly why we are not normally constrained by partner's answers to questions. Provided the answer is what we expect, they give no extra information and so cannot suggest anything.

I imagine some might argue that partner's having remembered our agreements is now UI and we may not choose from among LAs any option made more attractive by this information.

[aguahombre]

 blackshoe, on 2013-November-20, 07:37, said:

How does he know that 4♥ was interpreted as Texas?

Because Partner shouldn't be bidding 2NT with a long self-sufficient Spade suit, and won't be super accepting a drop-dead natural 4H bid.

This post has been edited by aguahombre: 2013-November-20, 17:28

[aguahombre]

 aguahombre, on 2013-November-19, 23:09, said:

When a player has both, and they lead to the same conclusion, he is using the AI and not the UI. If he can articulate the AI he is using, good luck proving he is not telling the truth.

 PeterAlan, on 2013-November-20, 07:54, said:

What's the basis for this assertion?

My basis for the assertion in the first sentence (which you quoted) was stated in my next sentence (which you didn't). If the UI does not provide any additional information to that which the AI provided, and he says he was using the AI ---I will believe he is using the AI and not the UI.

[blackshoe]

 aguahombre, on 2013-November-20, 09:54, said:

Because Partner should be bidding 2NT with a long self-sufficient Spade suit, and won't be super accepting a drop-dead natural 4H bid.

Huh? Which one is "Partner"? Or did you leave a "not" out of that sentence?

[PeterAlan]

 aguahombre, on 2013-November-20, 10:06, said:

My basis for the assertion in the first sentence (which you quoted) was stated in my next sentence (which you didn't). If the UI does not provide any additional information to that which the AI provided, and he says he was using the AI ---I will believe he is using the AI and not the UI.

Then we're not speaking the same language. You wrote (omitting two earlier short sentences that are irrelevant to this point, but including the next two):

 aguahombre, on 2013-November-19, 23:09, said:

Authorized information is usable. Unauthorization is not. When a player has both, and they lead to the same conclusion, he is using the AI and not the UI. If he can articulate the AI he is using, good luck proving he is not telling the truth.

To me, the "first" (third in the fuller quote above) sentence is a general, stand-alone assertion, for which you have given no basis. You didn't link the next sentence to it in any way that means anything to me, and in any event I don't think that it does provide such support.

You are essentially saying "If the player has the same information available from AI and from UI then (a) if he can say so, and that he's used the AI and not the UI, then I will believe him, and (b) for bridge law purposes I will then treat him as having used AI and not having used UI." This is fine as a statement of personal views; it is questionable as a statement of what the bridge laws actually are, as blackshoe said:

 blackshoe, on 2013-November-20, 07:23, said:

This is a flawed interpretation of the law. In particular, Law 16B1{a} does not mention AI at all — it says that when a player has UI, he may not choose from among logical alternatives one that could demonstrably have been suggested by the UI. The presence of AI "saying the same thing" is irrelevant.

I'm sorry to be so picky - not least, because my personal views on what the laws should say are probably quite close to yours and campboy's - but it does seem to be an area where imprecise use of language leads us astray. For example:

 campboy, on 2013-November-20, 08:20, said:

The information content of the AI given the UI and that of the UI given the AI are both zero, as you say ...

No, I emphatically did not say that, because this slurs the distinction that I was careful to draw. What I said was that the additional information content was zero, but the actual information content remains the same. If both the AI and the UI have the same information content X, Y & Z then that is what they both have, but neither adds additional information to the other. blackshoe is saying (and this is what I understood the generally accepted interpretation to be) that because the UI retains its information content of X, Y & Z that information is what needs to be taken into account when considering what is "suggested". You and campboy appear to be arguing that it is only any additional information (nil in this example). Slurring this distinction leads campboy to fallacies (or at best careless use of language) such as:

 campboy, on 2013-November-20, 05:52, said:

And if the UI adds no new information then it can't suggest anything.

the point being that whilst the null additional information indeed suggests nothing the actual information content of the UI (here X, Y & Z) can suggest plenty.

We use "UI" rather loosely. I think that a more fruitful line of argument is to focus on the definition of "extraneous information" in Law 16A3, which broadly speaking seems to correspond to what we might loosely express as UI-AI, any information that is available from unauthorised sources and not also from authorised ones, essentially what you're both talking about. Law 16B is couched in terms of "extraneous information", not "unauthorised information". Unfortunately, Law 74C, which is rather more general in its scope, uses "unauthorised information" and not "extraneous information".

It seems that we're all giving the meaning of the wording rather more careful attention than the drafters did. It would be helpful if at the next revision they started out by deciding exactly what they wanted to achieve by the relevant laws, and then drafted them accordingly with a bit more care. Among the things to take into account is whether the AI was available before the UI with which it overlaps, or only afterwards.

[campboy]

 PeterAlan, on 2013-November-21, 07:38, said:

No, I emphatically did not say that, because this slurs the distinction that I was careful to draw. What I said was that the additional information content was zero, but the actual information content remains the same.

I think you misunderstand what I mean. When I talk about the information content of UI given the AI I am using a technical term which essentially means the extra information you get from A if you already know B. You seem to have read my post as though the words "given the AI" make no difference.

Quote

Slurring this distinction leads campboy to fallacies (or at best careless use of language) such as:

That is neither a fallacy nor a careless use of language. I have defined precisely what I consider "suggested" to mean, and from that definition my assertion follows trivially. Now you clearly are using a different definition of "suggested"; what definition?

[PeterAlan]

 campboy, on 2013-November-21, 08:04, said:

I think you misunderstand what I mean. When I talk about the information content of UI given the AI I am using a technical term which essentially means the extra information you get from A if you already know B. You seem to have read my post as though the words "given the AI" make no difference.

No, I knew exactly what you were saying. However, I had deliberately not phrased what I was saying in this kind of way precisely because I wished to be very clear about the distinction between the information content of a piece of A/UI and the additional information content that a piece of A/UI brought. Your formulation, whilst accurate in the technical sense in which you used it, was therefore one that I was anxious to avoid using myself, because it obscured rather than highlighted this distinction.

 campboy, on 2013-November-20, 05:52, said:

And if the UI adds no new information then it can't suggest anything.

 campboy, on 2013-November-21, 08:04, said:

That is neither a fallacy nor a careless use of language. I have defined precisely what I consider "suggested" to mean, and from that definition my assertion follows trivially. Now you clearly are using a different definition of "suggested"; what definition?

This is where you and I part company. Your definition of "suggested" - which, incidentally, you only posted well after you used the word here - is not the normal one, because you import a specific restriction - namely, to the additional information, not the whole information, contained in the UI - into its semantics. That is not its normal English usage.

Let's analyse your statement, by first removing its conditional clause. "It [the UI] can't suggest anything." This is patently false - if the UI has information content, then it potentially suggests something. Now let's add back the conditional clause "if the UI adds no new information". My point is that the UI does not lose its original information content, and continues to suggest exactly what it always did. You and I can agree that it doesn't add anything to what it (and the corresponding AI) already suggested, but that's not the same as saying that it doesn't suggest anything at all - not unless you start to re-interpret "suggest", as you later did, in a highly specific and restricted way that does not, in my view, correspond to its normal usage. In saying all this, I'm following your line of looking at the use of "suggests" - but what I really think is happening is that whilst you're using "UI" in the first part of the sentence, your use of "it" is not in fact referring to "UI" but instead to "any new information that UI adds", which is not the same thing.

Since this question - is it the whole information content of the UI, or only that information which it adds to the AI, which matters for bridge law purposes - is the crux of the matter, it is particularly important to use language clearly, simply and naturally, and in a way that clarifies rather than obscures the point at issue.

[barmar]

 GreenMan, on 2013-November-20, 09:00, said:

I imagine some might argue that partner's having remembered our agreements is now UI and we may not choose from among LAs any option made more attractive by this information.

I once raised this question in rec.games.bridge: if I make a bid using a rarely-used convention (e.g. some of our responses to Mexican 2♦), I might be unsure if partner will remember its meaning; if he alerts it and explains correctly, is it now UI to me that he remembered, and I should continue bidding as if he didn't?

The overwhelming concensus was that this isn't a problem, because unexpected alerts and their associated explanations are UI, but not expected ones.

On the other hand, if either of us does forget the meaning, the explanation will be unexpected (if I forgot, I'll expect the wrong explanation, but hear the right one; if he forgot, I'll expect the right explanation, but hear the wrong one), and in that case it's UI.

[Zelandakh]

The problem with the position articulated by Peter above can be summed up with a simple example. I open 1NT and partner responds 2♦, which I announce as a transfer. Unfortunately partner gets confused and also starts to announce transfer immediately afterwards. Now I have UI that partner remembered the system. Clearly the UI suggests bidding 2♥ over passing now, especially if we have ever played weak takeouts or if partner forgot once (20 years ago). So we should pass, right?

Here's a better one. We recently switched from playing normal Stayman to Puppet. So I alert (or not according to regs) partner's club response and give the appropriate response when an opponent asks. Is partner now constrained by this UI to continue the auction as per normal Stayman even when they had not forgotten? I think everyone here would agree not and yet that is the logical conclusion from the idea that is being suggested.

The argument against is simple. If the UI is exactly the same as the AI then it cannot suggest one alternative over another. Can you give a counter-example where this is not the case without it also resulting in a logical loop back the the (silly) examples earlier in this post?

[barmar]

 Zelandakh, on 2013-November-21, 10:29, said:

The problem with the position articulated by Peter above can be summed up with a simple example. I open 1NT and partner responds 2♦, which I announce as a transfer. Unfortunately partner gets confused and also starts to announce transfer immediately afterwards. Now I have UI that partner remembered the system. Clearly the UI suggests bidding 2♥ over passing now, especially if we have ever played weak takeouts or if partner forgot once (20 years ago). So we should pass, right?

Not a problem. Just because partner once, many years ago, made a mistake, it doesn't become a LA every time to assume that he's repeated that mistake. The only LA is that partner has made the bid according to your agreements. Since there's only one LA, UI is not a problem, because that only constrains your choice among LAs.

Quote

Here's a better one. We recently switched from playing normal Stayman to Puppet. So I alert (or not according to regs) partner's club response and give the appropriate response when an opponent asks. Is partner now constrained by this UI to continue the auction as per normal Stayman even when they had not forgotten? I think everyone here would agree not and yet that is the logical conclusion from the idea that is being suggested.

This is a little more difficult, since it was a recent change and slip-ups may be common. But has partner been getting confused? If he hasn't yet forgotten the system, I think you can reasonable assume that he's remembering now, so it's not an LA to treat his bid as normal Stayman.

If he has actually had trouble remembering the change, it does seem like there's a UI problem. Every time you have a Stayman auction, you're unsure whether he remembers, so it's an LA to assume he forgot, and the UI helps you choose. Therefore, it seems like Law 16B forces you to interpret his bid in the opposite way that he explained it -- i.e. when he remembers, you have to forget, and vice versa.

The only answer in this case is to assume the lawmakers could not possibly have intended you to act like this, because it would make the game unplayable in conditions like this.

[mycroft]

There are 4 steps in delivering an adjusted score over use of UI:

• If there was UI
  
• that [edit: demonstrably] suggested one Alternative (or a class of Alternatives) over another from a list of Logical Alternatives
  
• and that Alternative was taken
  
• and there was a less successful Logical Alternative not suggested by the UI

then we adjust - with a definition of Logical Alternative.

That definition of Logical Alternative involves what would be considered and taken by the players peers *with only the AI* (or alternatively, without the UI).

So, "AI trumps UI" is wrong. "AI showing the same as the UI, so there's no UI" is also wrong. "AI shows the same as the UI, so there can't be damage" is also wrong, but frequently not a problem. "There's UI, but there's no LA to the call suggested given the AI only" - that we can do.

Yes, in almost all cases, this is semantic quibble only. But if I didn't worry about semantic quibble, I wouldn't style myself as I do, would I?

Note that Questions asked by partner, and answers to those questions, are UI. So theoretically, after partner asks about an Alert, I'll have to at my turn as well, if I want the information. Yes, sometimes we shortcut around the Law to make the game playable :-)

[chrism]

 mycroft, on 2013-November-21, 10:54, said:

Note that Questions asked by partner, and answers to those questions, are UI.

I don't think so; it is UI that partner asked the question, but the answer to the question is AI.

Actually, though the Laws allow a question to be asked (20F1) they do not explicitly state that the answer is AI even to the questioner; however, assuming that it is, then the clause "The partner of a player who asks a question may not ask a supplementary question until his turn to call or play" makes sense only if the partner was allowed to be aware of the answer.

I realize that we are in yet another grey area where the Laws manage not to state clearly exactly what they mean, of course.

[WellSpyder]

 mycroft, on 2013-November-21, 10:54, said:

There are 4 steps in delivering an adjusted score over use of UI:

• If there was UI
  
• that suggested one Alternative (or a class of Alternatives) over another from a list of Logical Alternatives
  
• and that Alternative was taken
  
• and there was a less successful Logical Alternative not suggested by the UI

then we adjust - with a definition of Logical Alternative.

That definition of Logical Alternative involves what would be considered and taken by the players peers *with only the AI* (or alternatively, without the UI).

So, "AI trumps UI" is wrong. "AI showing the same as the UI, so there's no UI" is also wrong. "AI shows the same as the UI, so there can't be damage" is also wrong, but frequently not a problem. "There's UI, but there's no LA to the call suggested given the AI only" - that we can do.

Yes, in almost all cases, this is semantic quibble only. But if I didn't worry about semantic quibble, I wouldn't style myself as I do, would I?

In my view it is far from a semantic quibble only - it is a manifesto for making the game unplayable! Frankly, I am amazed that we are having this discussion, since what you and PeterAlan (and maybe others earlier) appear to be arguing just looks crazy. But I know from real life that PeterAlan is an entirely sensible person, and although I don't know you irl, I have plenty of evidence that you understand the laws pretty well. So presumably I must be missing something pretty fundamental.

My problem is that the only get-out clause you have allowed above in the presence of UI is when there is no LA to the call suggested. But there are LAs to most of the decisions we take at the bridge table, so this isn't much help. Take the earlier example of a 1♠ opening out of turn, that is repeated after a pass by the correct dealer. I might have two choices as responder, say a 1NT or a 2♠ response. I think 1NT is a better response, and 75% of people agree with me, but 25% would bid 2♠. So both bids are LAs. 1NT certainly appears to be suggested by the UI, since that is the response the majority thinks is correct. So you are saying I must raise to 2♠, even though I think 1NT is a better bid and I know absolutely nothing from the UI that I don't also know from the AI???? It just doesn't make sense to me.

[mycroft]

I missed a word, and it's a critical word. That word is "demonstrably", and it modifies "suggested by the UI". I don't think you can say that "because the AI suggests that 1NT is a better call, that shows that the UI suggests that 1NT is a better call" and meet the standard.

The problem with "AI trumps UI" is that it's used in cases where the information isn't, you know, completely the same. It just is good for the arguing.

I don't know what to do in the original case - what is 4♠ over a "I know your hand to within a queen or a card, and I want to play exactly 4♥"? Partner just realized his clubs are spades? But I've seen way too many "forget transfer" auctions that everybody got right, and there's no UI (of course) except, except and except...that are "but it's obvious from the AI that partner can't have that hand" (and it sure is with the UI as well, at least).

[nige1]

IMO

• Few players understand or comply with UI law.
• Blackshoe and Peter Alan correctly interpret the law-makers' intentions.
• Different directors interpret UI law differently so it should be clarified.
• The words logical and demonstrably add confusion rather than value to the law and both should be expunged.
• It might be more sensible to penalize the giving of UI rather than the use of it.

[blackshoe]

 nige1, on 2013-November-21, 19:07, said:

It might be more sensible to penalize giving UI rather than using it

I doubt that, given how hard it is to not give UI.