[mycroft]

The difference between 5+ any major and 5+ guaranteeing a side 5-card suit is that 2D-p-p is a killer. In Multi (which, by the way, is only as low as it is (and in places where it is in "general play", weak-only Multi tends to be more restricted) because of familiarity - all over the world. Using multi as a reason to allow other similar things is likely to get you "that's an exception for historical reasons. Push too hard, and not only will the exception not be extended, it will be closed") 2D-p-p shows diamonds; it almost has to. In Wilkosz, it shows lots of diamonds or 1336 or the like - good gamble that partner has the pointeds, and if he doesn't, then the opponents likely have a massive spade fit (you lose if he's both majors, yes, but it's still a good gamble).

The difference between "We'll allow 5+ weak 2s, but not 5332, and not 5M4Mxx" and Muiderberg is that Muiderberg requires the 4 card side suit even when it's 6M - "standard" weak 2s will open on 6331 and 6322 and you're not "safe" pulling. And yes, if you claim to play judgement-restricted weak 2s that could be 5, but not 5332, because that's insane - which I agree with, by the way - and it turns out that you're always happy to run with the kinds of hands that will blow up in your face opposite a 6=2=2=3, but will work out if partner really does have a 4-card side suit, even if it's 6-4, then eventually you will be found out.

I agree that regulation of conventions everywhere is a nightmare, and it's not right anywhere. I agree that the GCC is particularly bad because of its history. I agree it should be overhauled from the ground up. I know there's a better chance of me becoming part of Canada A than that happening.

I also realize it's a game, and with non-ACBL options for North Americans opening up dramatically, the option to play where your rules hold sway is there. So if you play under the aegis of the ACBL, and you play deliberately close to the yellow line, expect to cross it occasionally. Take your speeding ticket and don't complain about the traffic cop or the laws.

Michael.

[foo]

So now that it appears I am (finally!) free of the other regulatory thread chewing up all my time, I can participate here.

1= Every experienced Bridge player knows that Single suiters and Two Suiters are inherently different hand types. Calling a two suiter a subset of a single suiter catagory seems like a logic flaw.

2= While opening a Weak Two with a 5 card suit is sometimes done for tactical purposes, the classic definition of a Weak Two is that it is a 6 card suit.
There's math that proves that
=a 2 level preempt should be on 6 cards
=a 3 level preempt should be on 7 cards
=a 4 level preempt should be on 8 cards
due to the fact that the expected trump fit is 8, 9, and 10 cards long (LOTT level)respectively.

Thus saying that a Multi 2♦ opening promises 5+ cards in either major is a bit misleading. It's supposed to promise =a Weak Two in either major=. Which means except for tactical manuevers it should promise a 6 card suit in either major.


Other than these 2 logic flaws (which demonstrate just how easy it is to go from reasonable to silly when regulating anything), I think awm's efforts to create a sensical, logically consistent system for regulating conventions is to be highly applauded.
Anyone want to nominate awm for his SO's equivalent of the ACBL C&C or the WBF equivalent?

[DrTodd13]

Please show me the math that proves that 2 level preempts should be 6 cards.

[pbleighton]

Quote

2= While opening a Weak Two with a 5 card suit is sometimes done for tactical purposes, the classic definition of a Weak Two is that it is a 6 card suit.
There's math that proves that
=a 2 level preempt should be on 6 cards
=a 3 level preempt should be on 7 cards
=a 4 level preempt should be on 8 cards
do to the fact that the expected trump fit is 8, 9, and 10 cards long (LOTT level)respectively.

Thus saying that a Multi 2D opening promises 5+ cards in either major is a bit misleading. It's supposed to promise =a Weak Two in either major=. Which means except for tactical manuevers it should promise a 6 card suit in either major.

In the ACBL's Alert Definitions, Definition of expected length for natural bids for the Alert Procedure section:

"Definition of expected length for natural bids for the Alert Procedure are:
Suit bids:
3+ in a minor and 4+ in a major for opening bids, rebids and responses.
4+ for an overcall at the one level, 5+ for higher levels.
5+ for a weak two-bid.
6+ for a weak three-bid."

Also see Fantoni-Nunes multiple world championships playing supremely undisciplined two bids.

Peter

[pbleighton]

Quote

Please show me the math that proves that 2 level preempts should be 6 cards.

Hee hee.

Peter

[Echognome]

The math is pretty simple.

You hold 6 cards in a suit. There are 7 outstanding. Assuming these 7 cards are equally likely to be in any of your partner's or opponents' hands, your partner expects to hold 7/3 = 2.333 cards on average. If the distribution of the 7 cards is a binomial distribution, then the distribution of your combined trump length is approximately (note rounding):

6 cards 6%
7 cards 20%
8 cards 31%
9 cards 26%
10 cards 13%
11 cards 4%
12 cards 1%
13 cards 0%

So you will have an 8+ card fit, 74% of the time. The rest of the argument depends on what you believe about the law... I will make no argument for or against that part of the analysis.

If you open a 5 card preempt at the 2 level, then there are 8 cards outstanding and your partner will hold 8/3 = 2.666 on average. Again, if it's a binomial distribution, then your combined trump length is approximately:

5 cards 4%
6 cards 16%
7 cards 27%
8 cards 27%
9 cards 17%
10 cards 7%
11 cards 2%
12 cards 0%
13 cards 0%

So, you will have an 8+ card fit about 53% of the time. Now there are two things to note. First, this assumes your longest combined fit is in the preempted suit and that is not necessarily the case. Against that, you might either not be able to find your best fit OR you might play in your best fit at a higher level.

By the way, I'm just providing the mathematics. (and hopefully correctly. If someone feels that the Binomial is inappropriate, I'm up for suggestions.) You can argue whether having an 8+ card combined fit at the 2-level is the appropriate measure. However, there is also the argument of whether a 4-4, 5-3, or 6-2 fit is better, depending also on the combined assets.

[foo]

DrTodd13, on Jul 3 2007, 03:00 PM, said:

Please show me the math that proves that 2 level preempts should be 6 cards.

This has to be asked appropriately in order for it to be solved as a math problem.

The assumption is that the LOTT is valid. That is to say a 2 level contract when holding 18-22 HCP between the two hands and only one suit fit requires a trump fit of at least 8 cards (and a working side shortness, but that's a twiddle).

So if you hold 6 cards in a suit, and pd's expected support is 2+ in the vast majority of cases, then the hypothesis is proven.

I'll let hannie or helene_t or one of the other professional mathematicians do the calculations.

EDIT: Thank you Echognome.

I read Echognome results as "it's at least 74/53= ~1.5x safer to open a Weak Two with a 6 card suit than a 5 card suit."
Particularly given his quite accurate comments about making sure you are playing in what the odds say is your best trump fit.

[DrTodd13]

You've drastically oversimplified the problem. You can't look at what you expect to make in a vacuum. You have to look at how the opponents will react. It may be best to bid more than you can make if it preempts opps from finding their optimum spot. This all depends on vulnerability, opponents available methods and many other factors. Some of the factors are psychological which makes anything approaching "proof" nearly impossible.

[foo]

DrTodd13, on Jul 3 2007, 03:28 PM, said:

You've drastically oversimplified the problem.  You can't look at what you expect to make in a vacuum.  You have to look at how the opponents will react.  It may be best to bid more than you can make if it preempts opps from finding their optimum spot.  This all depends on vulnerability, opponents available methods and many other factors.  Some of the factors are psychological which makes anything approaching "proof" nearly impossible.

Not if you accept the axiom of the LoTT.

That's the entire point of the LoTT. You !don't! know what the other hands look like. You are taking action based on the assumption that the HCP are more or less scattered evenly around the table and bidding to the level of your best scoring assumed fit in one go.

The LoTT is all about "bidding what you expect to score best in a vacuum".

Of course, if you don't believe in any variation of the LoTT, that's a different story
(In which case I assume you don't make preemptive raises to the 3 and 4 levels in the presence of an appropriate trump fit either. I have a hard time believing that.)

Given that history proves that the traditional suit lengths for "n" level preempts are reasonable, I think anyone trying to argue otherwise has a very steep hill to climb.

[awm]

Certainly it's traditionally been true that weak twos are six card suits. However, plenty of people will preempt freely on five, especially at nonvulnerable. There's a substantial set of people who use the multi specifically to distinguish between a "bad weak two" (often five cards, opened with 2♦ multi) and a "good weak two" (very traditional, typically six cards to two top honors, opened with a natural 2M). Virtually everyplace that allows 2♦ Multi at all permits the opening of 2♦ Multi with a five-card major, and states its requirements as "2♦ opening showing a weak preempt of 5+ cards in either major" or the like. So while one could argue that Multi should have to show a six-card suit, that's simply not the case.

As far as subsets go, there is again an issue of style. Pretty much everyone will use their judgement when deciding whether to preempt. Some of this is even a subconscious thing -- you remember the last time you held a hand similar to the one you hold now and think about whether you preempted and whether your choice worked out. It seems likely that people who use multi as a bad major suit preempt might start to notice that:

(1) They seem to get "caught speeding" more often when they're 5332, because their hand plays worse offensively and there's no place to run if the major doesn't provide a good fit. Thus they might tend to pass the 5332 hands more often than not...

(2) After 2♦-pass or 2♦-dbl, sometimes it's right for responder to pass. They will come to some agreement (explicit or implicit) about how many diamonds this shows and what opener should need to remove 2♦-dbl-pass-pass to him. If the multi very often shows only five cards in the major, it is probably right to pass the 2♦ bid fairly often with only five diamonds and a singleton major, both because this puts a lot of pressure on the opponents and because a diamond fit of at least seven cards is probably more likely than a major fit. It's also quite possible to pass without even having a lot of diamonds when responder is desperately weak, in order to get the opponents out of their comfort zone.

(3) They may notice that the Multi seems to get better results when opener actually has some diamond length, because responder's passes hit a fit more often, so they may tend to open borderline hands with a diamond side suit and not without. Similarly, they may notice that having a second five card suit helps a lot when they get doubled in the major, and therefore tend to be 5-5 to open these hands.

The "subset" rule merely allows players to make these tendencies explicit and disclose them. SOs don't even have the authority to regulate judgement -- the laws are pretty clear that you can regulate "agreements" (see Law 40C) but not regulate "what people bid at the table" (see Law 40A). As long as it's possible to have agreements with overlapping definitions (and pass versus virtually any preempt have overlapping definitions in standard methods) this problem remains essentially unavoidable. This is why it seems necessary to permit opening bids which show strict subsets of a legal alternative, despite troubling aspects like legalizing Wilcosz (if Multi is allowed).

[hrothgar]

foo, on Jul 3 2007, 11:34 PM, said:

Not if you accept the axiom of the LoTT.

That's the entire point of the LoTT.  You !don't! know what the other hands look like.  You are taking action based on the assumption that the HCP are more or less scattered evenly around the table and bidding to the level of your best scoring assumed fit in one go.

The LoTT is all about "bidding what you expect to score best in a vacuum".

Of course, if you don't believe in any variation of the LoTT, that's a different story
(In which case I assume you don't make preemptive raises to the 3 and 4 levels in the presence of an appropriate trump fit either.  I have a hard time believing that.)

Given that history proves that the traditional suit lengths for "n" level preempts are reasonable, I think anyone trying to argue otherwise has a very steep hill to climb.

Comment 1:

You're using an extremely naive interpretation of the Law of Total Tricks. Lots of folks favor methods that are overly aggressive than the Law advocates preach. For example, MOSCITO systemically raises a 4+ card 1M opening to the two level with three support. MOSCITO raises a 1M opening to the 4 level with 4+ card support.

While I use this as a specific example, several other systems are adopting similar methods. For example, a couple years back the Bridge World published an article advocating raising a 5 card major opening to 2M with two card support.

In both cases, the rational is identical: If the auction starts

1M - (P) - 2M - (P)
P - ???

Its MUCH more difficult to make an accurate balancing decision if the opponents could be playing in a seven card fit. The benefits of concealing information may very well outweigh the loss in constructive accuracy.

Comment 2: In what way, shape, or form does history prove that the suit lengths that you quote are reasonable. Single suited openings are certainly popular in the United States. Of course, players are effective banned from using anything else so its not like this is a particular useful data point.

Over the past couple years, I produced quite some summaries of the methods being played in the Bermuda Bowl and other such events. One of the specific points that I focused on way the definition of the 2♦ / 2M openings. Take a look at the systems that are actually being used in top level play. You'll quickly find that most pairs from outside North America have abandoned traditional single suited preempts for methods based on either two suited openings or assumed fit type methods.

Their 2M openings typically show either 5+ cards or, in some cases, 4+ cards.

[DrTodd13]

All you have is anecdotal evidence. The "law" of total tricks is not a law. It is an observation that some do not agree with. If in any way the LoTT were true then it could only be true given perfect bidding and play by everyone. The only thing you can prove is that the expected trump fit is 8.33 (or whatever the number) but that is a long way from that to proving that 2♥ will yield the best result over the long term with 6♥. What you've also failed to mention is that HCP is an important factor. You don't open this way when you have 20 points because excepted value of partner's hand makes game likely. You have to limit HCP such that expected value of partner's hand makes points roughly split where LoTT applies most.

[TimG]

foo, on Jul 3 2007, 03:34 PM, said:

Given that history proves that the traditional suit lengths for "n" level preempts are reasonable, I think anyone trying to argue otherwise has a very steep hill to climb.

Reasonable? Perhaps. Optimal? Far from clear.

Twenty years ago, Bergen-Cohen were dominating national pair events with (amongst other things) a preemptive style that might have been described as "n-1" relative to traditional preempting. At the time, I'm sure people thought the "n" approach had been shown to be reasonable.

What the math so far presented in this thread has shown is that a 6-card weak-two has the expectation of an 8+ card fit about 74% of the time and a 5-card weak-two has the expectation of an 8+ card fit about 53% of the time. Nowhere has it been indicated how often we need to have an 8+ card fit to make a preempt worthwhile (or reasonable). I doubt math can answer that question.

Nor has the advantage to being the first side to get into the bidding been considered. It may be true that a five-card weak-two is unsound looking at all four hands, but that it works out OK because of the difficulty opponents have in coping with a weak-two.

[Echognome]

Strict subsets don't work, especially when combined with the ACBL's current definitions of natural. I mean, we can start with Frelling 2♦. But imagine the following strict subsets:

2♣ = 3 clubs exactly, and at least 4-4 in the majors.
2♣ = 3+ clubs, but either another unknown suit of 6+ OR 6+ clubs.
2♣ = Either both majors (44 min) and 3+ clubs OR both minors (44 length)

You can see where a lot of strict subsets of 3+ clubs can become a nightmare.

So you may restrict 2-level bids to show 4+. We can play around with that as well.

[foo]

awm,
My only contradiction with what you are saying is that I do not agree with any explicit or implicit assumption that a two suited hand is a subset of single suited hands.
Bridge-wise, they are completely different beasts.

I'm happy to go along with the idea that any single suiter in "n" is a subset of all single suiters in "n". Ditto that any two suiter in "n+m" is a subset of all two suiters in "n+m".
But a hand with a 2nd place to play is very different from a hand with only one suit to suggest is very different from a three suited hand.

Bridge "physics" has to trump context independent mathematical arguments.


As for Wilcosz, my only problem is that creating a defense vs it that allows the defending side a decent chance to achieve equity appears to be very difficult.
More than anything else, the Laws and Regulations are there to protect and insure a reasonable opportunity at equity. Unless or until we can do that with regards to a specific method, that method should not be allowed any more than we would allow unrestricted performance enhancing drug use or unrestricted equipment modifications in any other game or sport.

[awm]

Echognome, on Jul 3 2007, 04:12 PM, said:

Strict subsets don't work, especially when combined with the ACBL's current definitions of natural. I mean, we can start with Frelling 2♦. But imagine the following strict subsets:

2♣ = 3 clubs exactly, and at least 4-4 in the majors.
2♣ = 3+ clubs, but either another unknown suit of 6+ OR 6+ clubs.
2♣ = Either both majors (44 min) and 3+ clubs OR both minors (44 length)

You can see where a lot of strict subsets of 3+ clubs can become a nightmare.

So you may restrict 2-level bids to show 4+. We can play around with that as well.

But once again....

Suppose partner and I decide to play a very aggressive preempting style. We define 2♣ as showing a weak hand (say 4-9 hcp) with 4+♣. Our three-level preempt are natural, normally six-card suits.

After playing this for a while, we notice that every time I open 2♣ on a 3334 hand I seem to get a lousy result. Of course, no one is holding a gun to my head forcing me to open 2♣ every time I have four clubs and 4-9 hcp, so I decide that maybe occasionally discretion is the better part of valor and start passing the 3334 hands.

At some point I notice that we're still occasionally going for a number. The problem is that I open 2♣ and get doubled. Partner keeps running from 2♣X on a doubleton (who wants to play a 4-2 fit doubled?) but sometimes there just isn't a good place to go. I notice (perhaps even subconsciously) that we do a lot better when I have a second four card suit than when I have (332)5 distribution. So I start passing a lot of the (332)5 hands too (maybe not all).

So what's happening here? All I've done is agree to play a 4-card weak two in clubs (discouraged maybe, but allowed). Then I noticed certain hands seem to get me lousy scores when I preempt them, so I started passing those more and more often. Nobody opens a weak two bid with every hand including the right number of points and cards in the suit anyway (people have requirements about suit quality or lack of outside tricks or lack of side voids or no side four-card major or at most one of four negative features or whatever, some of which are just "table feel" and not even formalized anywhere).

It seems like I haven't done anything "wrong" here, just agreed to play a legal convention and then used a little bit of discretion to not open the ridiculously flat hands. But suddenly if my partner discloses this to the opponents I'm playing those dastardly (and illegal in ACBL-land) assumed fit preempts! Where did we go wrong?

It's just this kind of issue that allowing subsets prevents. If we want to disallow assumed fit preempts, I think we have to require weak bids at the two-level or above to show five or more cards in the suit. This rule is restrictive perhaps, but a lot less ridiculous than some of the things ACBL has done in the past...

[TimG]

hrothgar, on Jul 3 2007, 04:04 PM, said:

You're using an extremely naive interpretation of the Law of Total Tricks. Lots of folks favor methods that are overly aggressive than the Law advocates preach. For example, MOSCITO systemically raises a 4+ card 1M opening to the two level with three support. MOSCITO raises a 1M opening to the 4 level with 4+ card support.

While I may agree with your point of view, citing MOSCITO methods neither proves the soundness of the method, nor demonstrates that the method is reasonable.

[Echognome]

Perhaps a better way to define it is to say that you cannot promise length in another suit. So, if by consequences of your system, a 4 card outside suit is guaranteed, then this is disallowed. E.g. in the modified EHAA we play, we play that 2x shows 5+, but not 5332. Thus it EITHER has a four card side suit OR has 6+ in the original suit, but certainly does not guarantee two places to play.

Then to tie the loose knots, there needs to be a disclaimer that it cannot promise a 4+ card side suit over X% of the time (to avoid the either 54 OR 9+ agreements). What is the best value of X? I have no friggin' clue.

[foo]

DrTodd13, on Jul 3 2007, 04:07 PM, said:

All you have is anecdotal evidence. The "law" of total tricks is not a law. It is an observation that some do not agree with. If in any way the LoTT were true then it could only be true given perfect bidding and play by everyone. The only thing you can prove is that the expected trump fit is 8.33 (or whatever the number) but that is a long way from that to proving that 2♥ will yield the best result over the long term with 6♥.

What you've also failed to mention is that HCP is an important factor. You don't open this way when you have 20 points because excepted value of partner's hand makes game likely. You have to limit HCP such that expected value of partner's hand makes points roughly split where LoTT applies most.

1= Results from 100's of thousands of boards over decades of play at all levels is !not! "anecdotal" evidence. It's experimental results. Lot's of them.

2= Since we must play SD but there is no objective way to build systems based on strictly SD work, it is unfortunately necessary to include DD analysis in the creation of system.
Besides, DD Bridge is the Holy Grail of SD play. "Beating baby seals" or "stealing" only works consistently against poor opposition and leads to poor Bridge habits and lax Bridge skills if over-indulged.

3= I explicitly mentioned the HCP component of the LoTT:

Quote

The assumption is that the LOTT is valid. That is to say a 2 level contract when holding 18-22 HCP between the two hands and only one suit fit requires a trump fit of at least 8 cards (and a working side shortness, but that's a twiddle).

Howard Schenken knew that hands in the 7-12 HCP range were the most common and made that the original range of Weak Twos based on that.
Later RW play ATT proved that the most effective range for Weak Twos was ~5-10 HCP even though that reduced their frequency.

[foo]

awm, on Jul 3 2007, 04:24 PM, said:

It's just this kind of issue that allowing subsets prevents. If we want to disallow assumed fit preempts, I think we have to require weak bids at the two-level or above to show five or more cards in the suit. This rule is restrictive perhaps, but a lot less ridiculous than some of the things ACBL has done in the past...

In fact, that was exactly what the ACBL's reaction was to Bergen & Cohen.

The creation of the infamous "5 and 5" regulation.
(A 2level or higher preempt must promise at least 5+ cards and at least 5+ HCP in the suit shown by the preempt.)