[mycroft]

It may be true in linear systems such as the integers, but it can certainly be true that A is preferred to B, B to C, and C to A. It's a perennial issue in voting systems.

And bidding systems, and Magic deckbuilding, and. Magic deckbuilding is the easiest to show how it works, because the designers work very hard to plan for it.

Assume there are 2 main strategies at point X in the "metagame". A generally beats B, unless luck or playing skill supersede. So more people play A, and B slowly dies out. There's this other strategy, D, that really sucks against B, or any other "rogue" strategy (not A or B), but has a decided advantage against A. Eventually there comes a point where there's enough A in the room that playing D is an advantage overall. If it succeeds enough, people drop A, and start playing B, say (because it's best against anything but A); and D fares badly. So it goes away, and now it's safe to start playing A again...

Note that in this case, D > A > B > D. And there's really nothing wrong logically or game-theoretically with that.(*) I think that there are strategies that will play well in the homogeneous bidding environment of American bridge - that are theoretically very poor, but since we can't be punished for it...

It's one of the reasons that in UI cases, you may have to group things into classes, and "this class is suggested over that class by the UI, so you can't do any of this class." I don't know if it helps here.

(*) (for Magic geeks only) Note that totally by accident (I just wanted to go to the tournament, and asked if a friend had another deck) I pulled this off. It was the depth of Necro hell; there were Necropotence decks, ErnhamGeddon decks and basically that was it. I showed up with a fast Goblin deck. Nobody played a Necropotence against me the entire tournament (what use is it when you're at 9 life?); only one Armageddon hit, and was reacted to by Disking the board. Eventually I ran into a deck that gained more life than I could hit for, and got their 1-ofs early in both games. Oh well.

[gnasher]

mycroft, on 2013-October-31, 18:44, said:

It may be true in linear systems such as the integers, but it can certainly be true that A is preferred to B, B to C, and C to A. It's a perennial issue in voting systems.

And bidding systems, and Magic deckbuilding, and. Magic deckbuilding is the easiest to show how it works, because the designers work very hard to plan for it.

It's also true when you're playing Rock-Paper-Scissors, and there are cardplay situations at matchpoints where each of three lines is better than one of the others.

However, I don't see how we can have a comparable situation when comparing Logical Alternatives under Law 16. Until someone actually comes up with an example where this type of circular relationship applies, I think we can assume that if A is suggested over B and B is suggested over C, then A is suggested over C.

[campboy]

gnasher, on 2013-November-01, 02:28, said:

It's also true when you're playing Rock-Paper-Scissors, and there are cardplay situations at matchpoints where each of three lines is better than one of the others.

Well, this can be true in the sense that if you play A and your opponents at other tables play B, you expect to win, and similarly with B vs C and C vs A (you also get this sort of situation with bidding borderline games in teams of 8 scored by IMPing the sum of the four scores). But an individual won't prefer A to B, B to C, and C to A; if he did he would get stuck in an endless loop of changing his mind. And it is the individual's preference between the strategies we are talking about.

If you play rock-paper-scissors then you presumably pick an option at random, because you have no preference between them at all. ("Good old rock, nothing beats that." -- Bart Simpson)

[Trinidad]

I think people are getting incredibly biased by seeing the actual South hand. People forget the main point:

North didn't know what South was holding.

North has observed the BIT. Now what does it tell him? Much less than we think. The only thing North could be sure of that South didn't have a dull balanced hand. That is the UI North had.

I find it frightening that everybody concludes that the BIT tells North that South must have a solid suit. That conclusion is far too accurate and is entirely based on seeing the actual South hand. If this discussion goes on, at some point someone will say that the BIT suggests that South holds:

---

Everyone forgets that South could have held:

---

or

---

These are hands where South would also think over 3NT. (If you think these are obvious passes, add some distribution. If you think these are obvious bids, make the hands more balanced.)

Given the enormous mix of hand types South could have had, it is really silly to discuss whether the UI says that a spade lead has an a priori better chance of success than a heart lead.

Rik

[billw55]

Trinidad, on 2013-November-01, 03:52, said:

I think people are getting incredibly biased by seeing the actual South hand. People forget the main point:

North didn't know what South was holding.

North has observed the BIT. Now what does it tell him? Much less than we think. The only thing North could be sure of that South didn't have a dull balanced hand. That is the UI North had.

I think it is slightly more than that - north can infer that south was considering a call other than pass. Which would likely include your 1st and 3rd example hands, but not the second.

[gnasher]

Trinidad, on 2013-November-01, 03:52, said:

Given the enormous mix of hand types South could have had, it is really silly to discuss whether the UI says that a spade lead has an a priori better chance of success than a heart lead.

If we think that a spade lead is suggested over a heart lead, why is it silly for us to say so or to try to explain our reasoning?

South was considering some action. That suggests that he has at least one long suit, and his values are in his long suits rather than his short suits.

The UI tells us, for example, that partner is more likely to have a five-card suit containing three honours. If that's what he has, it's not in hearts but it might be in spades. Therefore this possibility suggests a spade lead over a heart lead.

The UI also tells us that partner is more likely to have a five-card suit containing two honours. If that's what he has, it's more likely to be in spades than in hearts. Again, this possibility suggests a spade lead over a heart lead.

A similar argument applies to all the possible suits that might have encouraged partner to consider action rather than inaction. Hence the UI suggests leading spades over hearts. If a heart lead is an LA, a spade lead is illegal.

[Trinidad]

You are stuck in partner having a solid suit, without looking at the consequences.

You say:
- It is most likely clubs (say 65%), that bans a club lead.
- It could also be diamonds (say 34.9999%), that bans a diamond lead.
- It is not completely impossible that it is spades (say 0.0001%), that bans a spade lead.
- Hearts are impossible. A heart must be led.

Instead you should think along the lines of:
- It is most likely clubs, that bans a club lead.
     If South has clubs, what other leads are than suggested by the UI?

If you ask yourself that question you will see that if South has clubs, dummy most likely has raised based on a diamond suit to set up. What lead does that suggest to you: the active heart lead or the passive spade or diamond lead? To me it suggests an active lead: a heart.

Only after you have done this, you go to the next likely hand for partner: long diamonds

The reasoning is exactly the same: Diamonds are definitely barred. A heart is suggested over a spade or a club.

Only after this, you go and look at the unlikely scenarios:
It is not completely impossible that partner has solid spades: This would suggest a spade lead over a heart.

The conclusion should be: A club is out, a diamond is out, to decide between the majors, it is to close to call.

Rik

[blackshoe]

You seem to be saying that either a heart or a spade should be allowed.

[gnasher]

Trinidad, on 2013-November-01, 09:35, said:

You are stuck in partner having a solid suit, without looking at the consequences.

You say:
...

Who is "you"?

[karlson]

It hasn't been mentioned explicitly, but I think that playing the double of 3n as asking for a lead of partner's worse major is pretty common, hence I would think the spade lead is suggested. Not sure if those arguing with Trinidad are assuming this and not making it clear. I also cannot find anywhere in the thread if N-S had an actual agreement about double.

These situations are very difficult, because in practice many people don't pause over 1n-p-3n. Whether it's your side pausing as required and declarer perceiving a BIT, or the opponents taking 10 seconds when you know from previous experience that they usually pass instantly, it can be extremely difficult to convince the director that the tempo was normal/abnormal. There doesn't seem to be any great solution for this.

[blackshoe]

karlson, on 2013-November-01, 11:39, said:

It hasn't been mentioned explicitly, but I think that playing the double of 3n as asking for a lead of partner's worse major is pretty common, hence I would think the spade lead is suggested.

The question is not what's common, but what methods the players concerned are playing.

[Trinidad]

gnasher, on 2013-November-01, 11:25, said:

Who is "you"?

Well.. err.. you, Gnasher. And quite a few others too.

Rik

[mycroft]

... and what methods they play with other partners, or what methods they know their partner plays with other partners.

I will admit I don't have agreements on this with most of my partners; but if I'm playing with a Korbel, double says "lead spades", I know that, even if we haven't played for 10 years and we didn't have that agreement then. If I'm playing with J. (but I wouldn't), I know it would mean "find my suit" - because that's what he plays with other people. I don't think I'd ever dream of double meaning "find my minor" - but it seems to be common enough that some just assume that would be it. I can see "pause, then pass" saying "I want to double for an unusual lead, but not the unusual lead partner would normally find" (that might get me to thinking a club is suggested :-).

I find this case interesting, because I think it hinges on whether, under the circs, a heart lead from an effectively barren hand (so I get two heart tricks, that I probably get even if I don't lead it, if I can get partner to - maybe 3 if I don't pooch one on the go) is an alternative at all, not what the UI suggests (because I agree that it suggests not trying to develop tricks in my hand but to go looking for partners').

[Trinidad]

blackshoe, on 2013-November-01, 09:43, said:

You seem to be saying that either a heart or a spade should be allowed.

Yes.

When South has a minor (whichever), arguments can be made for a heart lead and for a spade lead. I think that we all agree that if South has a minor, either major suit lead might work. Nevertheless, if we would really be able to count accurately and deal all the possible hands where North, has the given hand, West has a 1NT opening, South has a good minor and East has a raise to 3NT, and we would analyze all possible outcomes of all leads, we would be able to say whether a heart or a spade would have more success in that case.

Let's say, for argument's sake, that out of the millions and millions of possibilities we see that a spade lead would be succesful in 49% of the cases and a heart lead in 51% of the cases. Would we then disallow a heart lead based on that analysis?

No, we wouldn't for the simple fact that, in practice -at the table, it would be impossible to do such an analysis. We just say that we (the TD and the player) don't know which of the two was suggested. It is hard to see the difference. This is like having an ocean of 49% black and 51% red balls. It is hard to tell what color balls is more abundant.

Now comes the next part. It is not completely impossible that South has solid spades. It is not at all possible that South has solid hearts. We need to adjust for this, so to our ball ocean we add one truck load of black balls and 0 truck loads of red balls. And then we mix everything up again.

Can you now tell which color balls are more abundant? Or does it still look like 50-50?

What is happening is that in the likely case that South has a minor, we say that we can't tell what major is suggested: The uncertainty is too large. This governs millions and millions of possible hands. And then for the few hands that South can have with solid spades, we say that spades are suggested. But this little bit of spade suggestion is much less than the margin of error for the far more likely cases where South has a minor.

Saying that a spade lead is suggested because South might just have solid spades is just as silly as saying that the universe is 3600 seconds older than we thought previously, because in their original evaluation the scientists overlooked the effect of daylight savings time. No matter how accurate we can determine the effect of daylight savings time, it is a tiny effect, much smaller than the error in the big picture. Similarly, we can accurately determine that should South hold a major that a spade lead is suggested, but it is insignificant compared to the question: what major lead is suggested when South holds a minor?

Rik

[lamford]

Trinidad, on 2013-November-01, 15:02, said:

I think that we all agree that if South has a minor, either major suit lead might work.

Why do you think we all agree? I think a top heart is much more likely to be better than a spade if partner has a minor. There might still be time to switch to a minor if we lead a top heart. And a top heart is going to be more successful than a spade when partner has something like Jx or Jxx in the suit we lead. However, we do not need to go through these arguments. We ask peers playing the same methods: a) what they would lead without the UI, and then b) what they think is demonstrably suggested by the UI. The evidence so far is that a heart is an LA, and that the UI demonstrably suggests leading something else. So we adjust. Just as with UI rulings in the auction, bidding on is often suggested over Pass, although what to bid is unclear. Bidding on is then an infraction. And it seems that you are not putting forward any new arguments in support of your view.

[ggwhiz]

lamford, on 2013-November-01, 15:54, said:

Why do you think we all agree? I think a top heart is much more likely to be better than a spade if partner has a minor.

I don't care if a poll of peers is split 50-50, a major suit lead is NOT suggested by the UI and a heart honor lead with a chance to switch to partners minor (or spades) in time is more indicated than a spade if you are looking to take advantage.

[gnasher]

Trinidad, on 2013-November-01, 09:35, said:

You are stuck in partner having a solid suit, without looking at the consequences.

Why did you say that, when I had just discussed the possibility that partner has a five-card suit headed by two honours?

[Fluffy]

lamford, on 2013-October-25, 18:08, said:

The possession of the nine is strong reason to think that a top heart would normally be led. Indeed, I recall a Woolsey article where he argued leading a top card from KQ8x(x) was theoretically best.

So what?, heart honor is normal, low heart is normal, which one is the offending side leading?, obviously not the succesful one.

I am talking in case you are adjusting. 9 tricks makes no sense, its either 10 or 7

EDIT: Actually once 9 tricks are assured, declarer might take a club finese for 11, so perhaps 11 is the answer.

[gnasher]

Trinidad, on 2013-November-01, 09:35, said:

Instead you should think along the lines of:
- It is most likely clubs, that bans a club lead.
     If South has clubs, what other leads are than suggested by the UI?

No I shouldn't. I should think along the lines specified in Law 16. That is, I should:
- Identify the LAs.
- For each pair of LAs, decide whether one is suggested over another.
- From this, identify the subset of LAs that are not suggested over any other.
- Choose one of these.

When, as part of this analysis, I compare the LAs of a small spade and a small heart, there are two relevant questions:
- How likely is it that partner was considering bidding spades or doubling for spades? I think it is quite likely, though not as likely as that he has clubs. Our length in spades makes it more likely that partner has a minor, but that is counterbalanced by responder's failure to investigate a spade contract.
- If partner was considering bidding a minor or doubling for a minor, how does that affect the probabilities of success for these two leads? I think the answer is that it reduces both to a near-zero chance of defeating the contract.

Hence the net effect is that a small spade is suggested over a small heart.

[Trinidad]

lamford, on 2013-November-01, 15:54, said:

Why do you think we all agree? I think a top heart is much more likely to be better than a spade if partner has a minor.

So what is your conclusion? That a heart lead could have been demonstrably suggested over a spade lead by the UI?

Is that a relevant question when North led a spade? (Other than that it says that the spade lead was not demonstrably suggested over a heart by the UI.)

The relevant question is whether a spade lead could have been suggested over a heart by the UI. Others have argued that it was, since South could have solid spades, but not solid hearts. I have said that the spade lead was not suggested over a heart lead. I assume that the two of us agree on that and that neither of us would adjust.

Rik