[rbforster]

akhare, on Feb 28 2010, 04:29 PM, said:

1♦  DN OR GF H+m OR GF ♦
1♥: SP bal OR ♠+m GF OR ♠ GF
      1N: S+♣
      2D: SS ♠
     2H+: S+♦ -> two suited module
1♠: SP majors OR bal GF
      2♣: SP with major (and possibly 4441 GF)
      2D+ -> Bal module
1N: GF Majors or SS ♣
2♣: SP (Single suited major?)
2D: GF SS ♥ -> Single suited module at 2S+
2H+ GF Minors -> Two suited module

Over 1♥, opener has the option of "predicting" the bal SP hand by bidding 1N. Responder can now resume relays with 2♣ following the bal module (possibly relay stayman).

I notice you don't list your unbalanced SP hands with minor(s) in the above structure. Are they in with 1C-1H, or somewhere else?

In general, I think it's hard to design good relay breaks that cater to SP hands while at the same time trying to get the benefits of relays. For example, in your suggestion

1C-1H SP bal / spade GF's

if you play that 1♠ relays and promises extras (GF values opposite SP bal), you'll be forced to break relay on all your minimum openers. Not all of them will be suited to bid a natural 1N instead; others will have to bid a suit, etc. This will disrupt your relays in other common cases like when responder has a GF and opener is minimum.

I suppose you could also play that 1C-1H-1S-1N reveals the SP balanced hand (so that 1S doesn't have to promise extras), which might work ok although you'll wrong-side NT sometimes. I guess if you break to 1N with opener's minimum balanced hands, but relay with unbalanced ones, you won't typically want to be in NT anyway so having the weaker responder bid it doesn't hurt too much.

I guess the big question is whether you're trying to save that extra step that shows the SP hand or not. If not, you can't fit in as many GF hands, but maybe that's an acceptable tradeoff. If you try to remove the SP response by relay breaking with minimums by opener instead, now you'll lose your relays whenever you've got (min vs GF) values which is a pretty common set of hands. Plus now you'll need to assign forcing bids over those relay breaks to cater to the GF responding hands, as well as trying to find the best partial with the SP ones. Not an easy set of goals to accomplish simultaneously.

[straube]

Rob, I just sent akhare this link...

http://www.users.on.net/~mabraham/systems/...sitives.html#1D

Memory load, but it accomplishes some neat things...

1. if there's a balanced hand opposite an unbalanced hand, the balanced hand always does the asking. The GF hands are always +0

2. the semipositives all respond 1H...so opener immediately knows that responder has something in case RHO interferes. They can be relayed with 1S at +2. Opener can start to relay and then relay break (sometimes) to offer to play.

3. Opener can always retain captaincy if he so chooses. 1C-1D, 1H-1S, 1N can be anything.

[akhare]

Rob F, on Feb 28 2010, 11:40 PM, said:

I suppose you could also play that 1C-1H-1S-1N reveals the SP balanced hand (so that 1S doesn't have to promise extras), which might work ok although you'll wrong-side NT sometimes.

Yeah -- the idea was that after 1♣ - 1♥ - 1N would show the min bal hand and 1♠ would be a hand unsuited for it (without promising extras).

Over this, 1N can be S / ♠+m (to line it up with 2♦ / 2♥) and the two suiter with ♠ and the other minor resolves immediately at 2♥+.

This effectively gives us 2♣ / 2♦ to show some SP hands, including the unbalanced hands with the minors.

After 2♣ / 2♦ (SP), there are no relays per se, and the thought was that opener just makes the cheapest bid as pseudo relay to get responder to describe shape.

Anyway, this was just a strawman -- more suggestions / criticisms are welcome ..

[akhare]

straube, on Feb 28 2010, 11:57 PM, said:

Rob, I just sent akhare this link...

http://www.users.on.net/~mabraham/systems/...sitives.html#1D

Seems very interesting.

BTW, I noticed a couple of things:

1) We seem to have consensus that the Moscito structure as it exists today wastes too many useful 2 level responses for immediate SP bids

2) This might be the longest thread on symmetric relays ever on the planet

[straube]

akhare said:

1) We seem to have consensus that the Moscito structure as it exists today wastes too many useful 2 level responses for immediate SP bids

I agree. Basically Moscito relays these semipositives before opener necessarily wants them relayed. I.e. opener hasn't had a chance to show a minimum or his own suit yet. As we've pretty much established, responder either 1) has to make a bid that could get dropped when his intention was to force game or 2) has to have less than GF values so that he doesn't mind getting dropped or 3) has to be promised a rebid. In order for 3 to work, opener has to have A) a relay bid available and B ) a misfit bid available (typically 2N) and these two things frequently clash.

akhare said:

2) This might be the longest thread on symmetric relays ever on the planet

Could be. I appreciate the help and I think I've learned a bit.

[Free]

straube, on Mar 1 2010, 08:44 AM, said:

akhare said:

1) We seem to have consensus that the Moscito structure as it exists today wastes too many useful 2 level responses for immediate SP bids

I agree. Basically Moscito relays these semipositives before opener necessarily wants them relayed. I.e. opener hasn't had a chance to show a minimum or his own suit yet. As we've pretty much established, responder either 1) has to make a bid that could get dropped when his intention was to force game or 2) has to have less than GF values so that he doesn't mind getting dropped or 3) has to be promised a rebid. In order for 3 to work, opener has to have A) a relay bid available and B ) a misfit bid available (typically 2N) and these two things frequently clash.

I must object. The whole point of semi positives is that responder gets in and shows his shape on hands where the part score battle is most important. These are the most common hands, and they are vulnerable to preemption. Starting with 1♣-1♥ as ANY SP ruins this entire principle, you just tell your opponents "we have half the deck, we definitely have something to play, but we still need to start looking". Combining SP with GF in 1♦ ruins this idea as well because you can hardly play forcing passes anymore, so you make your GF hands less accurate. Putting all GF hands in 1♦ is a disadvantage because of the SP, but it's playable. You don't have to make it worse.

After a SP at 2-level opener can just decide what to do. He can relay further (makes it GF), he can invite naturally, or he can bid his own suit which is NF and shows misfit. Obviously you can miss your best fits, but that's pretty rare.

I think if you start with 1♥ with pretty much all your SP's (or even put some SP's in 1♦), then you'll probably get to worse part scores than when you'd just start with 1♦ negative.

The reason why SP's are setup for relaying is now pretty obvious: you need a way to make the auction GF, and a way to show opener's hand. Using 1 bid for GF and other bids natural NF is the easiest, and still keeps our awesome slam bidding method.

Personally, I think the only thing that can be improved about the MOSCITO structure is the SP responses 1NT to 2♠. You can find numerous ways of doing this, every year Marston has a new structure, so he's still looking for the best solution. This means it's not 100% perfect, but it must be close since he doesn't change the rest of his structure.
The fact that he toyed with the idea of using 1♦ or 1♥ as SP but then abandoned the idea and never tried it again surely tells us he has very good reasons not to play this way. Either the method is aweful, or the alternative is so much better.

[hrothgar]

Free, on Mar 1 2010, 11:30 AM, said:

straube, on Mar 1 2010, 08:44 AM, said:

akhare said:

1) We seem to have consensus that the Moscito structure as it exists today wastes too many useful 2 level responses for immediate SP bids

I agree. Basically Moscito relays these semipositives before opener necessarily wants them relayed. I.e. opener hasn't had a chance to show a minimum or his own suit yet. As we've pretty much established, responder either 1) has to make a bid that could get dropped when his intention was to force game or 2) has to have less than GF values so that he doesn't mind getting dropped or 3) has to be promised a rebid. In order for 3 to work, opener has to have A) a relay bid available and B ) a misfit bid available (typically 2N) and these two things frequently clash.

I must object. The whole point of semi positives is that responder gets in and shows his shape on hands where the part score battle is most important. These are the most common hands, and they are vulnerable to preemption. Starting with 1♣-1♥ as ANY SP ruins this entire principle, you just tell your opponents "we have half the deck, we definitely have something to play, but we still need to start looking". Combining SP with GF in 1♦ ruins this idea as well because you can hardly play forcing passes anymore, so you make your GF hands less accurate. Putting all GF hands in 1♦ is a disadvantage because of the SP, but it's playable. You don't have to make it worse.

After a SP at 2-level opener can just decide what to do. He can relay further (makes it GF), he can invite naturally, or he can bid his own suit which is NF and shows misfit. Obviously you can miss your best fits, but that's pretty rare.

I think if you start with 1♥ with pretty much all your SP's (or even put some SP's in 1♦), then you'll probably get to worse part scores than when you'd just start with 1♦ negative.

The reason why SP's are setup for relaying is now pretty obvious: you need a way to make the auction GF, and a way to show opener's hand. Using 1 bid for GF and other bids natural NF is the easiest, and still keeps our awesome slam bidding method.

Personally, I think the only thing that can be improved about the MOSCITO structure is the SP responses 1NT to 2♠. You can find numerous ways of doing this, every year Marston has a new structure, so he's still looking for the best solution. This means it's not 100% perfect, but it must be close since he doesn't change the rest of his structure.
The fact that he toyed with the idea of using 1♦ or 1♥ as SP but then abandoned the idea and never tried it again surely tells us he has very good reasons not to play this way. Either the method is aweful, or the alternative is so much better.

Thanks for posting this (Didn't have the energy)

[hrothgar]

Here's a bit of background information regarding why MOSCITO adopted semipositives:

A few year's back, when I was investing a lot more time/enery in bridge, I spent a lot of time playing around with MOSCITO. I was playing a lot of boards and running an awful lot of simulations.

It quickly became apparant that the MOSCITO's very light strong club opening was stressing the system in couple of ways

1. The 1♦ "denies a game force" response was severely overloaded

The auctions after 1♣ - 1♦ were ugly, non intuitive, and very very common

2. The 1♦ "denies" a game force response was extremely vulnerable to preemptive. (It's bad enough when folks starting crashing your strong club opening in direct seat, however, folks were doing the same after 1♣ - (P) - 1♦ and we weren't any better positioned)

Devoting six separate bids to show semi-positive hands was a deliberate design goal. We wanted a system in which responder was able to immediately clarify strength and show information about shape.

[straube]

Free,

I appreciate your input.

Do we have the right understanding of Marston's structure? I read that the semipositives were defined as 3-5 QPs and that stiff kings and queens weren't counted. The only latitude I'm aware of were that 5/5 hands could make a positive with 5 QPs. Perhaps other hands could shade?

The problem I see are bids like 1C-2S showing a spade single-suiter. I've given the hand before...
KQJxxxx xx x Qxx and I'd like to know what Marston does with that. It only has 4 QPs. Or it could be even worse, like KQJxxxx xx K Kx which would count as 5 QPs.

He must bid 1D (GF) with these hands, but that's not what the notes I read said.

Atul and I came up with..
1H-all else
1N-H, 5H/m
.....2D-H
.....2H-5H/4C
.....2S-5H/4D
2C-5M/4M
2D-5S/4C
2H-6S
2S-5S/4D
3C-5S/5D

which was organized to let responder get a rebid. There's advantages to this, but sometimes the relay bid is up a step (for instance 1C-2H, 2N is relay) so we're back to +2

I think the trouble with semipositives at the 2-level is that they've relayed shape before opener has really asked them to do so and now opener is often endplayed into relaying again instead of bidding his hand naturally.

Now I know that this structure wasn't Marston's invention. It was just an attempt to let responder get another chance to bid. If he uses semipositives that really ARE semipositives then responder could bid 2S with 6 spades and wouldn't mind hearing opener drop him there.

I read on your site that semipositives were about 60% of the hands, positives 30% and negatives about 10%. First point is that the negatives (we've found) are more like 19%...maybe a little less because the computer counted ace-only hands as DN. Second is that if the 60% is true, then Marston is combining still semipositives with minimum GF hands because semipositives of the 5-7(8) variety are about low 30s%.

[akhare]

hrothgar, on Mar 1 2010, 08:31 AM, said:

Devoting six separate bids to show semi-positive hands was a deliberate design goal.  We wanted a system in which responder was able to immediately clarify strength and show information about shape.

Thanks Free and Richard.

What's your perspective on relaying SP hands? Personally, I think that it's essential to have relay breaks to allow us to get out in a sensible partscore. One obvious example of this is this the 1♣ - 1♥ (bal or no 5CM), which could be made on as little as AXX XXX XXXX XXX and out.

Do we really want to relay over this with say a 15-17 NT hand? IMO, it's far more important to be able to break out by allowing by opener to describe minimum hands that don't want to relay using something like:

1N: Min bal hand
2C: Majors?
Etc.

Similarly, we need the ability to play at in a partscore at the 2-level *while* allowing some sort sort of GT.

So, I guess what we are missing is a "user manual" for SP responses and any pointers in this regard will be greatly appreciated...

[rbforster]

In my experience, it's better to have some flexibility in letting responder do hand evaluation and not tie his hands by mandating that he must have 6 QPs to make a GF response. You have silly hands like

KQJx
QJxx
Qxx
xx

that are solid balanced ~10 counts with soft values and only 4-5 QPs that somehow don't count as a GF under these rules. Your example hands with long good majors are similar and might be worth "upgrading" similarly.

The point is that it's better to let responder judge this for himself. If this means that when you get done resolving shape relays for GF hands, your control asks need to cater to strong hands with fewer controls (and maybe missing some ambitious slams), well, I think that's a better problem to have than missing some games when partner undersells his hand and opener fears a misfit and stops low.

For me, 8+ HCP is enough to GF if you like your hand. Eventually when there's a relay ask for QPs, the first step is <=6 (5-6 are most common, 4 is rare). DCBs continue as usual, but opener can't make as many inferences about what missing honors "must be" since he's not sure of the total count (5 vs 6). You could also solve this problem with a weak relay/strong relay system for starting DCBs as Echognome has mentioned previously.

[hrothgar]

Rob F, on Mar 1 2010, 06:54 PM, said:

In my experience, it's better to have some flexibility in letting responder do hand evaluation and not tie his hands by mandating that he must have 6 QPs to make a GF response. You have silly hands like

KQJx
QJxx
Qxx
xx

that are solid balanced ~10 counts with soft values and only 4-5 QPs that somehow don't count as a GF under these rules. Your example hands with long good majors are similar and might be worth "upgrading" similarly.

The point is that it's better to let responder judge this for himself. If this means that when you get done resolving shape relays for GF hands, your control asks need to cater to strong hands with fewer controls (and maybe missing some ambitious slams), well, I think that's a better problem to have than missing some games when partner undersells his hand and opener fears a misfit and stops low.

For me, 8+ HCP is enough to GF if you like your hand. Eventually when there's a relay ask for QPs, the first step is <=6 (5-6 are most common, 4 is rare). DCBs continue as usual, but opener can't make as many inferences about what missing honors "must be" since he's not sure of the total count (5 vs 6). You could also solve this problem with a weak relay/strong relay system for starting DCBs as Echognome has mentioned previously.

The restrictions on the number of slam points are a natural consequence of the auction termination mechanisms that MOSCITO uses after shape is resolved.

If you want to allow responder the option to GF with less than QPs, feel free to do so. However, you'll also need to adjust your base QPs to five, or four, or where-ever you decide to draw the line.

In turn, this is going to mean that your "normal" slam investigatory sequences are going to be pushed up one or more steps.

Simply put, you don't get something for nothing...

[Free]

Straube, yes, the percentages seem to be wrong, but the conclusion is the same: DN < GF < SP.

1♣-2♠ shows a singlesuited hand with ♠. This means 6+♠ (5332 is considered balanced). Obviously you can have hands where game is likely and opener won't invite, but they will probably be quite rare. KQJxxx-xxx-x-Qxx is a nice hand, but you just described it as it is. Opener knows what to expect, with lots of Aces or a fit he can easily invite. If he has a misfit with long ♥ he can bid 3♥ and you give him game. I don't see many problems with singlesuited hands.

What I do see as a problem is for example bidding 2♥ to show 4♠ and 5+♥. Here opener may have a 2-1-5-5, you can easily have a nice m fit,...

Hands with singleton K or Q are always a problem for AKQ-points. Note that in Marston's document about responses to strong ♣, and his CC, he always mentions HCP, never AKQ-points. I think he might play it more flexible, but I'm not sure.
Borderline SP/GF hands (borderline because of a singleton K/Q) are not that big a deal because Marston's philosophy is to let opener describe his hand when minimum. So if you have a stiff K or Q, you can just bid it as a GF and hope partner is minimum. If he keeps asking, you'll have to lie a slampoint...

[straube]

I know most hands that would respond 2S fall between extremes, but I'm still uncomfortable making a nf bid that could be Jxxxxx Ax xxx xx or KQJxxxx xx x Qxx.

Now the nice thing about 2S nf is that opener has pass available and this keeps the relay mechanism at +1, but the semipositive strength then needs to be more narrowly defined.

I think RobF has the right idea about using a little judgment to upgrade (or perhaps downgrade) the semipositives and then the base QP step for positives is 6 or less. This is a similar approach taken by Mark Abrahms (sp?).

The selling point of the semipositive approach seemed like it had a lot to do with the high frequency of the semipositives...but I feel that this is because the semipositives included a lot of hands that were GF worthy.

If you keep the semipositives as 3-5 QPs irrespective, then you have the problems we've been discussing. If you make the semipositives GI hands opposite a 1C (something like 5-8 hcps depending on shapeliness) then their frequency drops and you have GF>SP>DN...at which point you have to start asking whether it's worthwhile to show these SPs at the expense (+1 step) of the GF hands. I don't feel good about this trade.

[rbforster]

hrothgar, on Mar 1 2010, 11:00 AM, said:

Rob F, on Mar 1 2010, 06:54 PM, said:

In my experience, it's better to have some flexibility in letting responder do hand evaluation and not tie his hands by mandating that he must have 6 QPs to make a GF response... 

If you want to allow responder the option to GF with less than QPs, feel free to do so. However, you'll also need to adjust your base QPs to five, or four, or where-ever you decide to draw the line.

In turn, this is going to mean that your "normal" slam investigatory sequences are going to be pushed up one or more steps.

Simply put, you don't get something for nothing...

I agree you aren't getting something for nothing, but I think there is a better solution than pushing lower your base QPs for a GF. Obviously if you switch from min 6 QPs for a GF to min 5 QPs, all your slam investigations with 6+ QPs will be one step higher. Since there aren't lots of 5 QP GF's, this isn't a good tradeoff. On the other hand, having an ambiguous first step with no base QP, so

4-6 QPs and enough to GF
7 QPs
8 QPs
...

keeps your slam investigations at the same level for all your 7+ QP hands. You only take the hit on the 4-6 QP hands, where opener will not be as sure about responder's controls when he hears that first (weakest) response. I think this is a sensible approach to dealing with these infrequent-but-control-poor GF hands.

[akhare]

straube, on Mar 1 2010, 12:24 PM, said:

The selling point of the semipositive approach seemed like it had a lot to do with the high frequency of the semipositives...but I feel that this is because the semipositives included a lot of hands that were GF worthy.

Well, as Richard and others have pointed out, the real design motive behind the specialized SP responses is to tackle 4th hand interference.

So, the real efficacy should be measured in ability to cope with situations in which 4th hand tries to throw in a monkey wrench (vis-a-vis 1♣ - 1♦ (0-7 any))...

[rbforster]

akhare, on Mar 1 2010, 10:41 PM, said:

straube, on Mar 1 2010, 12:24 PM, said:

The selling point of the semipositive approach seemed like it had a lot to do with the high frequency of the semipositives...but I feel that this is because the semipositives included a lot of hands that were GF worthy.

Well, as Richard and others have pointed out, the real design motive behind the specialized SP responses is to tackle 4th hand interference.

So, the real efficacy should be measured in ability to cope with situations in which 4th hand tries to throw in a monkey wrench (vis-a-vis 1♣ - 1♦ (0-7 any))...

I think there are two issues here.

1. interference by 4th hand after an ambiguous 1C-1D auction is bad in normal precision, and direct suit-showing bids for semipositives are clearly better at handling that interference.

this is separate from

2. some people like to mix up their hands based on "real playing value" vs the normal classification of positive or SP. Some example SP hands with lots of shape and few controls are still arguably game forcing, while other soft balanced hands are bid as SP despite having GF strength (~10 hcp).

The first issue is the real reason why people play direct semipositive bids. The second is a way to tweak your system to make your slam bidding easier (QP requirements) and involves shuffling around a few hands between the GF and SP categories at the margins. You don't design a system around #2, you design a decent system for other goals and then realize you can make some changes along the lines of #2 that seem to be a slight improvement (and that your system is structured in a way that can handle both GF hands bid as SP and upgraded SP hands bid as GFs).

[akhare]

Rob F, on Mar 2 2010, 03:46 AM, said:

1. interference by 4th hand after an ambiguous 1C-1D auction is bad in normal precision, and direct suit-showing bids for semipositives are clearly better at handling that interference.

A related question for the standard Precision 1♣ - 1♦ players -- how do you folks cope with 4th hand interference?

[lowerline]

akhare, on Mar 4 2010, 11:41 AM, said:

Rob F, on Mar 2 2010, 03:46 AM, said:

1. interference by 4th hand after an ambiguous 1C-1D auction is bad in normal precision, and direct suit-showing bids for semipositives are clearly better at handling that interference.

A related question for the standard Precision 1♣ - 1♦ players -- how do you folks cope with 4th hand interference?

(dbl)
pass = 19+ relay --> rdbl = 0-5; other = 6-8
rdbl = 16-18 takeout (4crd M)
other = 16-18 natural

(1♥)
pass = 19+ relay --> dbl = 0-5 4+♠; 1S = 0-5 <4♠; other = 6-8
dbl = 16-18 takeout (4♠)
other = 16-18 natural

(1♠)
pass = 19+ relay --> dbl = 0-5; other = 6-8
dbl = 16-18 takeout (4♥)
other = 16-18 natural

(1NT+)
pass = 16-18 balanced
dbl = takeout
other = natural NF