[DinDIP]

A few years ago I played a MiniMeck variant in a couple of national championships. I suggested to partner that we not play the “standard” North American responses: lumping all (or most) 8-11 hands opposite a strong club seemed a poor idea to me. It’s bad enough that opener has made an amorphous, strength-showing bid without responder doing the same; while competition is not common, when it occurs it is likely to put our side at more of a disadvantage than if we had shown something useful about our shape.

In general, limiting responder’s strength is more advantageous in non-competitive auctions, while showing shape caters well to competitive auctions. So, why not make the strength distinction once both opponents have passed?

Consider a 1♠ response showing 5+H (part of a transfer set of responses). Opener can ask with 1NT. Now responder can describe his strength and, if maximum, show shape:
2♣ = minimum (~8-11), exactly 5H (opener can ask again with 2♦)
2♦ = 6+H, strength not yet defined (2♥ asks for strength and shape)
2♥ = maximum (good 11+), a four-card or longer minor (2♠ asks which minor and side-suit shape)
2♠ = maximum, 4+S (2NT asks for shape)
2N = maximum, BAL
3m = maximum, good 5+m

Note an important consequence: with minimum strength responder reveals nothing about his shape unless he has something extra to show (a six-card or longer suit). When I played this, my partner and I had lots of auctions that went 1♣-1♠-1NT-2♣-4♥, telling the opponents very little about either hand.

Revealing unnecessary information – information that makes the defence easier and doesn’t help your partnership bid more accurately – is a significant problem for strong-club players who use a symmetric relay structure.

Most symmetric players use unlimited responses. This is advantageous if the opponents compete. However, it mean that the partnership frequently leaks information because opener typically relays out responder’s complete shape before asking about responder’s strength. Even with transfer responses, some of the time responder will bid the strain the partnership ends up declaring, meaning the defence has a complete picture of declarer’s shape, and often some idea of strength. Experiments in the early 1980s (shortly after Roy Kerr had devised the symmetric structure) to divide responder’s bids by strength to avoid relaying when slam was unlikely – 1♥ showed any 12+ and 1♠ and higher were all 8-11 – were unsuccessful.

At one stage Paul Marston and Stephen Burgess (and other pairs) experimented with a “weak relay”: using step+1 after shape had been resolved to ask responder to show controls only with a good hand, otherwise to bid the first step. This proved useful on a small number of hands but the strength information usually came too late, especially if opener would have liked to revert to natural bidding if responder were minimum.

A significant improvement is to give opener the option of a weak relay after the initial shape-showing response. Consider a variant of symmetric where the responses have been ordered so that a 1NT response shows 5+H, either one-suited, with diamonds or with exactly four clubs. Much of the time opener will relay with 2♣ and responder will resolve shape as usual.

With a minimum-strength hand opener has the option of bidding 2♦ instead. This asks responder to bid the first step with a minimum and higher with a maximum, showing shape as usual.

At first glance it looks like this has pushed the partnership two steps higher for little, if any, gain.

The important consideration, however, is typically not the level at which shape alone is resolved, but both shape and strength.

Consider the example of a 7RP 1-6-3-3 hand (where RP = relay points, commonly A=3, K=2, Q=1, with kingleton = 1 and singleton Q = 0).

On the standard track the shape would be resolved at 3♥, then responder would show strength at 4♦. After the weak relay, the shape is resolved at 3NT and strength at 4♦ (or 4♥, if your partnership agrees to allow good 6RP hands, perhaps x AQJ9xx KJx JTx, to respond positively to the weak relay).

An important advantage of the weak relay is that direct auctions are now possible. For example, opener can bid hearts cheaply after responder shows a minimum expecting responder to bid game unless holding top of the range AND lots of extra shape, such as 6430 or 6511.

As well, the partnership can get out of relay mode. This can be helpful on those deals where knowing the location of stoppers and strong suits helps the partnership choose the right game, as well as those where an excellent fit is required for slam.

The downside, as keen readers will have noticed, is that the step to show a minimum over the weak relay after the auction starts 1♣-1NT-2♦ is 2♥, meaning responder will be declarer if the partnership plays in hearts. This is a price to pay but note that responder is unable to transfer to hearts in most symmetric structures, so there is always a degree of randomness as to which partner will bid hearts first.

I’m interested in feedback on this idea and how best to implement it. The structure I’m currently experimenting with is:
1♥ = 4+S, not 4432 or 4333 or three-suited with both minors
1♠ = diamond one-suiter (not 5332) or diamond and club two-suiter or club one-suiter (not 5332) or two-suiter with 5+C and 4H
1NT = 5+H, one-suiter, two-suiter with exactly 4D or two-suiter with 4+C
2♣ = BAL (4333, 4432 or 5m332), 11+HCP or 7RP (so AAQ or AKK)
2♦ = BAL 8-10, 4-6RP
2♥ =4H and 5+D
2♠ = 5+H and 5+D
2NT = three-suiter with both minors and spades
3♣+ = three-suiter with both minors and hearts
After a 1♥ response and 1♠ relay, 1NT and 2♥/2♠ show the same shapes as above, but with spades instead hearts. 2♣ shows a spade-heart two-suiter (with either suit longer), 2♦ shows a three-suiter with both majors, and 2N+ four spades and five or more clubs.

All the suit-showing responses (in response to 1♣ and after 1♣-1♥-1♠) are designed to allow opener the opportunity to make a weak relay, except when responder has a three-suiter short in spades.

The tradeoffs compared with standard symmetric seem to me to be these:

• Many long minor suit hands respond 1♠, so opener will be declarer in the most likely game (3N) after relaying with 1N.
  
• Hands with a long minor and exactly four hearts bid 2♥ at some stage, making responder declarer if hearts is the strain we play in. Note that this no different from standard symmetric where these hands would bid 2♥ as the reverser.
  
• The 1♠ response is more vulnerable to competition than 1N showing D or D+C.
  
• The 1N response is less vulnerable than the 1♠ counterpart in standard because responder has promised a 5+card suit.
  
• Memory strain as a result of a complex set of responses compared with the simple standard ones.

[hrothgar]

Thanks for a very thoughtful post...

I think that the idea has merit. My big question is about the opportunity cost. What are we losing?

For me, at least, the auction

1♣ - 1NT (showing 5+ Hearts)
2♦+

would be a relay breaks transferring captaincy.
The strong club opener is showing a minimum hand with 0-1 Hearts.

The 1NT bidder can now either chose a step response as a relay OR transfer to natural bidding.

I think that I could easily be convinced that the weak relay makes sense (and that the shortness showing relay breaks should be pushed a step)

[straube]

1C-1N as 5+H (0-3S) is maybe a prime example of where a weak relay might work best; responder has a strong preference for hearts over spades and often opener will have a heart fit. Why not frequently offer to abort relays?

But after 1C-1H there's a lot more uncertainty as to the final contract. If I understand your proposal, a 1S rebid wouldn't necessarily promise extras. It would just preserve the normal relays, perhaps in an effort to conserve space so as to find the best fit.

So would using the weak relay pretty much announce a fit? Or that opener has an independent suit of his own?

I don't like lumping very distributional but minimum hands in with flat minimum hands. I don't know how to separate them either.

Too bad your structure uses 2C and 2D responses for balanced hands. It would be nice if the balanced hands took up less room so as to allow for reverse relays.

Thinking that IMprecision addresses some of your concerns. It lumps responder's invitational with light GF hands and opener can certainly blast after finding a fit....at the risk of not discovering that responder has a shapely maximum. IMprecision does use reverse relays (such as 1C-1D, 1S-1N asking) and it's possible that a weak relay here would be beneficial.

[Cascade]

Balanced hands are much more frequent that minor one or two suiters.

1♣ 1♠ Balanced
1nt ...

2♣ Balanced max
2♦ Balanced min

and there is room for unbalanced minor two suiters in 1♠

If you make

2♣ Balanced max

2♥+ Balanced min

then you can put an unbalanced one suiter in 1♠ (or possibly both unbalanced one suiters - clubs and diamonds)

[Kungsgeten]

One idea which I think could be used more is that both opener and responder could be the captain after a strong 1C opening. I think IMPrecision does this to some extent (with their weak/strong 1D response) but I haven't studied it much. The top-level Swedish strong club partnerships use a structure where I think both opener and responder can use relays, but I haven't found details on the continuations (only what is available on the Bermuda Bowl CCs).

I personally think that something like below could work:

1C--
1D = Negative.
1H = (12)13+ relay.
1S = Balanced or two-suiter without 5M. 8-12.
..1NT = R.
....2C = 5+C.
....2D = Balanced.
....2H = 5+D, 4+C.
....2S = 5+D, 4H.
....2NT = 5D440.
....3C+ = 5+D, 4S.
1NT = 5+H, 8-12.
2C = 6+m or 4441, 8-12.
..2D = R.
....2H = 6+C
....2S = 4441.
....2N+ = 6+D.
2D+ = 5+S, 8-12

The idea is that shape starts to resolve at 1S, instead of 1H. Its nothing new, but I like the theory behind it. After 1C-1H, opener shows his shape with the same principle as 1C-1S+. It also allows opener to use natural bidding after many of the 8-12 responses, since we're still pretty low.

[Cascade]

 Kungsgeten, on 2019-November-08, 12:56, said:

One idea which I think could be used more is that both opener and responder could be the captain after a strong 1C opening.li

I recall talking to Roy Kerr fifteen or so years ago and he had ideas of the following response structure to a 1♣ 16+:

1♦ any Positive 9+
1♥ double negative 0-4
1♠ and up semi-positive 5-8 with some relay structure.

Over the 1♦ positive opener could relay with 1♥ with the identical relay structure as the semi-positives or opener could describe with the same structure.

You lose a step with the relays with positive hands but gain a step if you relay with the semi-positive hands. You also gain the flexibility with the positive hands.

If you play forcing pass then this is even more efficient with 1♣ any positive; 1♦ negative; and 1♥ up for semi-positives and you can play standard Symmetric Relays or some transfer response front end.

[straube]

 Cascade, on 2019-November-09, 08:07, said:

1♦ any Positive 9+
1♥ double negative 0-4
1♠ and up semi-positive 5-8 with some relay structure.

imo it just isn't going to work. The semipositive responses just take up too much room for opener to comfortably suggest a part score. The DN similarly handicaps opener showing shape and making forcing bids. Pretty sure I have a thread somewhere in which I looked at 1C-1H as a DN and it wound up on the dustheap. Think my idea was to use the 1D response as semipositive, though, which at least would allow for lower part scores (some obvious tradeoffs here).

I've found it expensive to try to cater to allowing one partner to decide whether to be master or slave. It is usually less expensive to set up a structure such that one has to tell or has to ask....and then one wants to set up the structure to favor captaincy for such hands as 1)initiate the GF 2) are balanced 3) are stronger 4) are likely to declare.

Some other commentary...one problem of relay structures with positives is how do you know you're going to be permitted a relay auction? Like you start 1C-1H showing hearts and next hand preempts 3D. You don't know if partner has 7S/4H or 6H/4C or what...but you've spent bidding room communicating GF and 4+ hearts. So you've got the GF in (which is a very good thing) but that tight package of distributional information that you hoped to unwind won't be unwound. But if say 1C-1H is a semipositive showing hearts....and if hearts is the most interesting feature of the hand, well that 3D interruption may be disrupting a part score contract. Maybe there wasn't going to be a relay auction.

Another thing is that maybe David was looking for something such as....

1C-1H (GF, spades)
1S-1N (I have a minimum hand and uninteresting shape)
...........2C-tell me anyway
...........etc-natural?
.....etc-interesting or non-minimum hands

in other words, at first go responder sets a GF and communicates useful (major suit) information but after it appears that a relay auction will be permitted, we figure out whether we really want to have one or not.

Could be interesting. IDK, but figuring out "the math" of this is daunting to me....additionally because it wouldn't be for just 1C-1H. It would be for 1C-1S etc.

And it seems like sometimes we might get too high for finding partner's shape. I.e. we ought to come out the same ultimately when we're slamming because we've just inverted the order of information that we're getting (strength first, then distribution) but sometimes we'd just be happy knowing complete shape (maybe find out if partner has tolerance for our suit) in which case we've done a lot of "math" to set up our relay structure for not much.

[PrecisionL]

I have been experimenting with the following strong ♣ design at local clubs:

Responses to 1♣ (17+ bal, or 16+ unbalanced)

1♦ = 4+♥ and less than G.F. (8+ hcp)
1♥ = 4+♠ and less than G.F. (8+ hcp)
1♠ = No 4-cd major and 0-7 hcp
1NT = G.F. (a) 5-4 or better in the majors or, (b) any 5-5
2♣ = 5,6-cd minor (either)
2♦ = G.F., 5+♥
2♥ = G.F., 5+♠
2♠ = G.F. any 4441 hand

[foobar]

 PrecisionL, on 2019-November-10, 20:09, said:

I have been experimenting with the following strong ♣ design at local clubs:

Responses to 1♣ (17+ bal, or 16+ unbalanced)

1♦ = 4+♥ and less than G.F. (8+ hcp)
1♥ = 4+♠ and less than G.F. (8+ hcp)
1♠ = No 4-cd major and 0-7 hcp
1NT = G.F. (a) 5-4 or better in the majors or, (b) any 5-5
2♣ = 5,6-cd minor (either)
2♦ = G.F., 5+♥
2♥ = G.F., 5+♠
2♠ = G.F. any 4441 hand

Maybe we need a c) in the 1♣ - 1N response to handle the GF balanced hands (unless that goes in 2N)?

[PrecisionL]

 foobar, on 2019-November-11, 16:51, said:

Maybe we need a c) in the 1♣ - 1N response to handle the GF balanced hands (unless that goes in 2N)?

Whoops! 1♠ response = (a) 0-7 hcp & no 4-cd major, or (b) 8+ hcp and balanced

[Cascade]

 PrecisionL, on 2019-November-11, 17:47, said:

Whoops! 1♠ response = (a) 0-7 hcp & no 4-cd major, or (b) 8+ hcp and balanced

This can possibly be improved upon.

After 1!c 1!s 1NT denying a major and weak I would like to be able to bid 2!c and 2!d to play. Maybe you can do that and do something else with 8+ and a four-card major.

But 1!c 1!d and 1!c 1!h showing four of the major and 0-7 or 8+ balanced might work even more efficiently so that 1!c 1!s always denies a major.

[DinDIP]

 straube, on 2019-November-03, 11:19, said:

IMprecision does use reverse relays (such as 1C-1D, 1S-1N asking) and it's possible that a weak relay here would be beneficial.

Indeed! Testing in an IMPrecision-like context shows a weak relay to be valuable. When responder has 7+RP it's usually right to relay away but when the 1♦ responder has a BAL 8-11 you typically only want to relay when opener has significant extra strength (roughly 21+). Breaking relay here allows the partnership to conceal unnecessary information, especially as teller is often going to be declarer. It also allows the bidding to shift into stopper-search mode when there is no 8+card major-suit fit. This means forgoing a shortness-showing break but the BAL 8-11 hands are more frequent than the 7+RP ones with shortness.

[DinDIP]

 straube, on 2019-November-09, 09:25, said:

Another thing is that maybe David was looking for something such as....

1C-1H (GF, spades)
1S-1N (I have a minimum hand and uninteresting shape)
...........2C-tell me anyway
...........etc-natural?
.....etc-interesting or non-minimum hands

in other words, at first go responder sets a GF and communicates useful (major suit) information but after it appears that a relay auction will be permitted, we figure out whether we really want to have one or not.

Could be interesting. IDK, but figuring out "the math" of this is daunting to me....additionally because it wouldn't be for just 1C-1H. It would be for 1C-1S etc.

And it seems like sometimes we might get too high for finding partner's shape. I.e. we ought to come out the same ultimately when we're slamming because we've just inverted the order of information that we're getting (strength first, then distribution) but sometimes we'd just be happy knowing complete shape (maybe find out if partner has tolerance for our suit) in which case we've done a lot of "math" to set up our relay structure for not much.

My initial testing suggests that the weak relay works best when there is greater clarity about teller's shape, as David suggests. Rather than a weak relay immediately over a 1♥ (=4+S) response, it seems to work better if asker waits until teller has said something more about shape. For example, 1♣-1♥-1♠-1N shows similar shapes to an immediate 1N response but with hearts rather than spades. Now the weak relay works much better.

The other scenario where it works well is when teller is known to have a single-suited minor. Now a weak relay allows a stopper search.

[DinDIP]

 Cascade, on 2019-November-09, 08:07, said:

If you play forcing pass then this is even more efficient with 1♣ any positive; 1♦ negative; and 1♥ up for semi-positives and you can play standard Symmetric Relays or some transfer response front end.

Another alternative is to play 1♣ as negative, 1♥+ as semi-positive and 1♦ as all GF, with opener showing shape rather than responder. Strength is more evenly distributed when responder has a GF so the problems that can occur with weak hands relaying opposite strong ones occur less frequently.

[DinDIP]

 Cascade, on 2019-November-03, 21:39, said:

Balanced hands are much more frequent that minor one or two suiters.

1♣ 1♠ Balanced
1nt ...

2♣ Balanced max
2♦ Balanced min

and there is room for unbalanced minor two suiters in 1♠

If you make

2♣ Balanced max

2♥+ Balanced min

then you can put an unbalanced one suiter in 1♠ (or possibly both unbalanced one suiters - clubs and diamonds)

All true, but:
1. when responder bids 1♠ with all BAL hands we have an 8+card spade fit more than 20% of the time (based on a 100,000 deal sim) and we will have wrong-sided the contract a reasonable portion of the time given the disparity in strength;
2. it's not clear to me that putting BAL hands into the 1♠ response (so as to ensure that opener is always declarer in a notrumps contract) because of their increased frequency is more important than stand-alone responses that immediately alert opener to responder’s shape? (I recognise that when one hand is BAL the likelihood of competition is reduced, but it’s not eliminated.) And the problems of competition are exacerbated if the response includes minor two-suiters or a minor suiter or . . . ; and
3. another advantage of an immediate BAL hand signal is that it allows opener to know when to ask and when it would be better to tell, especially if responder is known to have a minimum GF.

[DinDIP]

 hrothgar, on 2019-November-03, 05:12, said:

Thanks for a very thoughtful post...

I think that the idea has merit. My big question is about the opportunity cost. What are we losing?

For me, at least, the auction

1♣ - 1NT (showing 5+ Hearts)
2♦+

would be a relay breaks transferring captaincy.
The strong club opener is showing a minimum hand with 0-1 Hearts.

The 1NT bidder can now either chose a step response as a relay OR transfer to natural bidding.

I think that I could easily be convinced that the weak relay makes sense (and that the shortness showing relay breaks should be pushed a step)

Richard raises the important issue: what is foregone if adopting a weak relay, and is that price worth paying? After nearly 40 years of playing symmetric (on and off) I’m still agnostic about how best to use relay breaks. For most of that time I've played them as Richard does but not with great success. Although very infrequent, I've always thought those in The Way Forward, a British system from the mid-90s, in which step plus one showed a void (hearts at first opportunity, then spades, then clubs then diamonds) and continued relays, catered for hands that are very poor for normal relay methods.

[Cascade]

 DinDIP, on 2019-November-19, 22:34, said:

All true, but:
1. when responder bids 1♠ with all BAL hands we have an 8+card spade fit more than 20% of the time (based on a 100,000 deal sim) and we will have wrong-sided the contract a reasonable portion of the time given the disparity in strength;

It should be possible to put balanced hands with four spades into 1!h. I can't remember our relays and not sure if I have my old notes but we played a 1!h opening as four or more spades including balanced hands and somehow got into symmetric style relays.

[awm]

I feel like there’s been a bit of a bait and switch here. The first part of the post seems to be arguing that instead of the standard symmetric order of suit(s)-shortness-pattern-strength it would be better to order as suit(s)-strength-shortness-pattern, with the idea being that starting with suits is best in case of competition but showing strength early can sometimes allow natural relay breaks and/or early sign off before shape resolves. This seems like a good case!

But from here you go into this “weak relay” idea which kills your cheap relay break and loses a bunch of steps, and also leads to some issues when responder has min values with crazy shape.

My question is, why not have responder show strength to the first relay? You end up at basically the same place if you wait for complete resolution but reordering gets you the advantages you want. This also saves your initial relay break for shortness showing if you want it.

[straube]

1C-2C strong, GF 6+C or 5+C/4+D
2D-2S ask, 8 QPs
2N-3S ask, 2245

Like that? Pattern resolution at various end points? I might set up a computer program that way, but I would have difficulty without the crutch of knowing that 5422s usually resolve at 3C. I bet others would have difficulty, too.

And this still doesn't solve the problem of teller having a shapely (freak) minimum when asker wants to blast after a disappointing strength answer.

[awm]

I was thinking more along the lines of:

1c-1nt (hearts or hearts and clubs)
2c...
... 2d = 4-6 RP
...... 2h relay at +2, but we can zoom into RP and 6RP can zoom to honor location
...... not 2h = natural relay break
... 2h+ shape at +1 and 7+ RP

You do have to remember +1 and +2 but you kind of have to do that in response to the “weak relay” or in competition anyway. I wasn’t planning to immediately unroll all the RP. You potentially save steps later by not having to separate min/max and/or being able to zoom more.

It also might pay to rearrange the hand types a bit; for example on one suited hands instead of:
2s high shortage
2n mid shortage
3c two-plus doubletons
3d 5332
Etc...

2s = no singleton/void
2n = 6(331)
3c = 7(330)
3d+ 7(321)

The idea would be that opposite the less distributional hands you can often blast something without resolving shape.