[akhare]

straube, on 2011-December-12, 10:59, said:

Moscito's 1C

1D positive
1H and 1N+ semipositive
1S negative

<snip>
I've thought that for a strong club system, it's expensive to use the 2 and 3 levels for semipositives. For example, Moscito uses 1N and higher as semipositives. They're all set up for relays. But the point is that frequently opener doesn't have the strength to continue to relay them. So they have 1C-2S, P sequences. Feels like a misfire.

To be fair, the most common Moscito SP response is likely 1♥ (bal or no 5CM), after which opener typically bid 1N.

Also, Marston played around quite a bit with the SP responses and in the later versions, the ones with 5+ major used something like an 2-under scheme with multiple meanings so that it was easier for opener to show a different suit or relay...

[straube]

Yeah, that's a good point. The nice thing about a bid like 1C-2S (semipositive 6 spades) is that opener can pass and use 2N as a relay and 3C for something else; this compared to 1C-2C (semipositive 6 spades) in which opener's bid of 2S is a sign off. The bad thing about 1C-2S (semipositive 6 spades) is that responder can be passed when he might light to make another try.

Other subtle differences...our 1H semipositive is 2-4 QPs whereas Marston is 3-5 QPs and awm's semipositives are 5-8 hcps.

[awm]

You may be missing the point somewhat on this strong diamond system. The responses to 1D are not particularly optimized -- it is certainly possible to do better and what I play now is substantially better on technical merit. You are right for example that 1d-2c showing 6+c is unlikely to be best.

The goal wasn't the "best" response structure though, it was a simple structure. The 1d-2m auctions are same as our 2m openings. The 1d-1s auctions are just 1nt "systems on." The 1d-1nt-2c relays are the same as 1c-1nt-2d+. The goal here was to minimize memorization. The most complicated part of the system is the 1nt structure and that comes up a lot because we use it opposite all balanced ranges.

I'm sure moscito-style semi-positive relays (or my current system) is "better"... but at a high memory cost.

[straube]

Makes sense. I appreciate the memory load aspect.

I think I'm still interested in the relay possibilities over 1D. +2 is bad, but I think we'd still be better off than natural bidding. At least one hand knows the other's approximate shape...and many times will know the exact shape (all of the balanced and 5431 patterns for example). On the more distributional hands, we can only give an approximate shape. We might even "lie" and describe AQJTxx Jxxxx x x as a 6421 rather than a 5521 pattern. Opener is still more likely to have a picture of responder's hand than if responder were bidding naturally because opener is making S1 bids instead of simultaneously trying to show his own shape. I mean, I could argue that lack of room is exactly why we need relays. Maybe we could use relay breaks to show opener's self-sufficient suit.

With a strong club of 16+ (17+ balanced), we've found that requiring only 5 QPs for a positive and lowering semipositives to 2-4 QPs means that we have more relay auctions...and these are our best game auctions. Your 1D is as strong as our club, so I'm thinking the same ranges for responses.

I've started to understand why 1C works best with a weak NT opening; for one thing, if opener retires to 1N (like after 1C-1S, 1N) one might easily be doubled otherwise knowing that say an 11-13 was opposite a max 8. Did you look at all at having 1C-1D be 9+ instead of 9-13? Would this violate the GCC as it seemingly starts a non-GF relay auction? Seems like it would save room for game/slam investigation and it would free up 1N for something else....maybe to play. If one has Kxx Ax xxxxx xxx one could respond 1N rather than 2C or 2D.

[awm]

straube, on 2011-December-12, 20:59, said:

I've started to understand why 1C works best with a weak NT opening; for one thing, if opener retires to 1N (like after 1C-1S, 1N) one might easily be doubled otherwise knowing that say an 11-13 was opposite a max 8. Did you look at all at having 1C-1D be 9+ instead of 9-13? Would this violate the GCC as it seemingly starts a non-GF relay auction? Seems like it would save room for game/slam investigation and it would free up 1N for something else....maybe to play. If one has Kxx Ax xxxxx xxx one could respond 1N rather than 2C or 2D.

In the 2004 version (and really the last few versions before that too) 1♣-1♦ is 6-13. It would be the normal response on your example hand. The 6-8 range is special because if you have exactly four cards in a major, it is better to respond 1M so as not to lose a major suit fit after 1♣-1♦-1NT. However, 6-8 without a major (or 6-8 with a five-card major, for that matter) normally bids 1♦.

We have tried without 1NT game force. However, the 1NT game force offers a number of fairly substantial advantages. It allows you play virtually all contracts from the hidden hand. It also caters to better slam bidding, because you can set the suit early (for example 1♣-1NT-2NT-3♥ sets hearts and starts cuebidding, whereas 1♣-1♦-1♠ and both 2♥/3♥ are NF, and even 1♣-1♦-1♠-2♣-2NT-3♥ would be invitational). Basically responding with 1NT gives you the option of having a cuebidding auction rather than being forced to relay all the time, which can be a big advantage.

[awm]

People sometimes think that relays are more efficient than natural bidding, but this is not necessarily really true.

Suppose we are in a game-forcing auction and partner bids at some point 2♦. The number of sequences to 3NT (for example) which start with me bidding 2♥ next is equal to the number of sequences which start with me bidding above 2♥. So to make maximal use of my sequences, I should make the cheapest call half the time, the next cheapest call a quarter of the time and so forth.

It is true that natural systems don't make the cheapest call anywhere close to half the time. But it's also true that relay systems have relayer making the cheapest call substantially more than half the time. Yes, I know that people play relay breaks but do you really break relay half the time, at every single relay opportunity that comes up in the auction? I certainly don't! In fact I would guess that less than half my relay auctions involve any relay break at all (prior to 3NT anyway).

Anyway, the point is that maximizing efficiency is somewhere between the two extremes. You will want to rearrange some bids rather than playing strictly "natural" (i.e. to increase the frequency of cheapest call) but you won't want to go all the way in the relay direction. A lack of space makes it more necessary to do this in order to get good results, and also greatly reduces the number of sequences you have to consider in order to do a reasonable job.

[straube]

awm, on 2011-December-12, 22:44, said:

People sometimes think that relays are more efficient than natural bidding, but this is not necessarily really true.

Suppose we are in a game-forcing auction and partner bids at some point 2♦. The number of sequences to 3NT (for example) which start with me bidding 2♥ next is equal to the number of sequences which start with me bidding above 2♥. So to make maximal use of my sequences, I should make the cheapest call half the time, the next cheapest call a quarter of the time and so forth.

It is true that natural systems don't make the cheapest call anywhere close to half the time. But it's also true that relay systems have relayer making the cheapest call substantially more than half the time. Yes, I know that people play relay breaks but do you really break relay half the time, at every single relay opportunity that comes up in the auction? I certainly don't! In fact I would guess that less than half my relay auctions involve any relay break at all (prior to 3NT anyway).

Anyway, the point is that maximizing efficiency is somewhere between the two extremes. You will want to rearrange some bids rather than playing strictly "natural" (i.e. to increase the frequency of cheapest call) but you won't want to go all the way in the relay direction. A lack of space makes it more necessary to do this in order to get good results, and also greatly reduces the number of sequences you have to consider in order to do a reasonable job.

This is all true. It reminds me something I appreciate about our strong club structure which awards slavery to the first hand that reveals an unbalanced pattern. We go up the line...1C-1D (GF, possibly balanced), 1H (possibly balanced)-1S (possibly balanced), 1N (balanced) and each turn (starting with responder's first bid) one or the other may spin off (not S1) and start to describe something distributional. It's almost perfect Fibonacci and it lets us be +0 regardless of who winds up captain.

But when one hand starts to pattern, we're back to S1 relays and we haven't even added relay breaks. I'm aware of a few relay breaks that are useful, but for the most part it's just very nice to have one partner with a complete picture of the partnerships assets.

[straube]

awm, on 2011-December-12, 22:37, said:

In the 2004 version (and really the last few versions before that too) 1♣-1♦ is 6-13. It would be the normal response on your example hand. The 6-8 range is special because if you have exactly four cards in a major, it is better to respond 1M so as not to lose a major suit fit after 1♣-1♦-1NT. However, 6-8 without a major (or 6-8 with a five-card major, for that matter) normally bids 1♦.

We have tried without 1NT game force. However, the 1NT game force offers a number of fairly substantial advantages. It allows you play virtually all contracts from the hidden hand. It also caters to better slam bidding, because you can set the suit early (for example 1♣-1NT-2NT-3♥ sets hearts and starts cuebidding, whereas 1♣-1♦-1♠ and both 2♥/3♥ are NF, and even 1♣-1♦-1♠-2♣-2NT-3♥ would be invitational). Basically responding with 1NT gives you the option of having a cuebidding auction rather than being forced to relay all the time, which can be a big advantage.

Is this 2004 also your current version? So 1C-1D, 1N and you use your NT structure (or I suppose one could use regular stayman and transfers). 1C-1D, 1M would show an unbalanced hand with that major? That would need to be forcing I think. What would responder's continuations be? And which are forcing on opener and which are not?

1C-1M is 0-8 with a 4+ major or 6-8 with a 4-cd major. I really like how bidding continues until some semblance of fit is found. We always risk a silly contract when we pass our nebulous diamond. It's infrequent enough, but still... Do you ever respond in a 3-cd major? I would dislike raising clubs with 3343 for example.

[mycroft]

The ACBL's treatment of relays is rather odd. But a descriptive bid - even "0-4 or big" tends not to be defined as a relay, even if it is the cheapest bid and requires explanation. Also, my guess is (and it's just a guess; I haven't looked) that after 1♦-1♥; whatever, that the weak hand doesn't continue to relay out the strong hand (which would be the issue that would get them into trouble). Maybe the gameforcing "weak hand" does, in which case...

Note that the idiosyncratic definition of "relay system" needs all its points; defined only slightly differently, the mama-papa 1NT-2C; 2H-4C; 4H-6NT auction would be banned as a relay system. Similar arguments can be made for other parts of the definition with, again, mama-papa auctions.

[awm]

straube, on 2011-December-13, 00:07, said:

Is this 2004 also your current version? So 1C-1D, 1N and you use your NT structure (or I suppose one could use regular stayman and transfers). 1C-1D, 1M would show an unbalanced hand with that major? That would need to be forcing I think. What would responder's continuations be? And which are forcing on opener and which are not?

We didn't change opener's rebids. 1♣-1♦-1♥ is a three-suiter short in spades with 4♥, 1♣-1♦-1♠ is a three-suiter short in clubs with 3-4♠ and 4+♦, 1♣-1♦-2♣ is a three-suiter short in diamonds with 4+♣, and so forth. These rebids are not forcing because opener is limited to 11-16 and responder could have as little as a 6-count. If responder tries to sign off in a suit at the cheapest level or bids 1NT (or passes) it's less than invitational; if responder jumps in a suit or bids 2NT that's invitational. If responder relays that's normally invitational-plus.

2004 is not exactly the current version, but the changes since then have been minor and generally not in the "core" of the system (stuff like responses to preempts) although we did make this notrump range change to 10-12/19-21. Through 2004 I played this system quite a lot in several regular partnerships; subsequent to then I haven't played much with Noble Shore and Charlie Garrod (they both live on the other coast; Noble has found other partners and Charlie has cut way down on his bridge playing) or with Greg Humphries (he basically gave up bridge for a while; he's now playing professionally more or less full-time but that implies other fixed partnerships for him). Sam Ieong and I have switched to our strong club methods. I've formed several other partnerships but haven't felt the need to inflict this system on them. The only partner I still play strong diamond with on a semi-regular basis is David Kempe, but we both have other partners and probably only play a couple times a year. Even that may decrease since he's based in Los Angeles and I have moved north to Santa Clara.

[benlessard]

Quote

So to make maximal use of my sequences I should make the cheapest call half the time, the next cheapest call a quarter of the time and so forth.

It is true that natural systems don't make the cheapest call anywhere close to half the time. But it's also true that relay systems have relayer making the cheapest call substantially more than half the time. Yes, I know that people play relay breaks but do you really break relay half the time, at every single relay opportunity that comes up in the auction? I certainly don't! In fact I would guess that less than half my relay auctions involve any relay break at all (prior to 3NT anyway).

This is really missing the big picture of relay bidding. If opener say that hes got 5S- and youve got the option of Responder show 4H or opener show 5S/4H ? Wich one would you pick ? The point of a bridge bidding system is not to give information it's to give information that help you make a decision. If opener showned Axxxx,Kxxx,Ax,Ax the fact that responder manage to show anything is a nuisance not a plus. Natural bidding = both showing something in the hope that one of the player will be able to make a decision, so in that case maximizing the information on both side make sense. Relay bidding = getting as much as information possible from one side so the other side can make a decison.

A compromise is often the best solution because some information are worth more than other. so responder job is not only picking the contract after getting the information but it's also finding ways to get useful information rather than getting the most information, so for that being possible he often need to transmit information of it's own.

[straube]

Thanks for explaining.

Wondering if 1C-1D, 1H could serve as a maximum.

1C-1D, 1H-1S relays
.....1N-strong NT
.....other maximums

1C-1D, 1S-club short, minimum
1C-1D, 1N-spade short, minimum

something like that

Very cool system. I like how you can stop as early as 1H.

[mycroft]

Relays aren't primarily designed for the most efficient exchange of information - but an efficient enough exchange of information such that:

• the try-notry decisions can be reasonably made by 3NT; the slam-noslam decision can be made before 5trump, and so on;
  
• if at all possible, all information is one-way, *and* the described hand is dummy.

Relay breaks tend to be set up to invert the one-way direction of information flow, and if designed well, also invert who becomes declarer, because having the described hand declarer is more damaging than having the lead go through the strong(er) hand.

Note that that "one-way" thing (and, of course, the "stop on a dime, out of nowhere" thing - but I'm referring to GF relays, as that's what I know) is what the RAs that regulate relay systems are going after, because it is such an advantage.

So, it's an optimisation problem, and "most efficient use of space" is *not* the optimising criterion (though, of course, knowledge of the way of most efficient use of space is important in cramming everything necessary into the first go-nogo decision space - up to 3NT).

Asking bids do similar things, but require more of the optimisation to be done at the table ("I need to know these three things, and if the answer to X is Y this one more - how do I get the information I need short of <bail point>?") with the exchange of not having to go through the motions every time you want anything on the track, and giving that information to the opening leader (and, of course, having different information available, and therefore, using different go-nogo judgement).

[straube]

How about...

1C-1D
.....1H-max
..........1S-min
...............1N-strong NT
...............2C-short diamond
...............2D-4S, short heart
...............2H-1-3-(54)
...............2S-3-1-(54)
...............2N-5m/5m
..........1N-GF ask (as for 1C-1N)
.....1S-max or min, 4 hearts, short spades
.....1N-max or min, short clubs
.....2C-min, short diamond
.....2D-min, 4S, short heart
.....2H-min, 1-3-(54)
.....2S-min, 3-1-(54)
.....2N-min, 5m/5m

or....

1C-1D
.....1H-max
..........1S-min
...............1N-strong NT
...............2C-short club
...............2D-short diamond
...............2H-1-3-(54)
...............2S-3-1-(54)
...............2N-5m/5m
..........1N-GF ask (as for 1C-1N)
.....1S-4H, short spade, max or min
.....1N-short heart, max or min
.....2C-min, short club
.....2D-min, short diamond
.....2H-min, 1-3-(54)
.....2S-min, 3-1-(54)
.....2N-min, 5m/5m

[awm]

It is generally good to be bidding suit-below-shortness here. The reasons are:

1. It gives partner the option to pass; bidding the shortness directly will almost never be passed.

2. The cheapest step as relay isn't a place you'd want to sign off anyway.

I also think your method seems rather memory intensive, and I am not sure what you think you are gaining by removing the least-troublesome hand from 1D in exchange for much worse 1C auctions.

[straube]

I don't understand that last part about removing the least troublesome hand from 1D...I didn't mean to remove any hand from the strong 1D opening. Perhaps you meant the strong NT? But that was only like 14-16 or so (can't remember exactly what 1C-1D, 1N range is).

I tried to keep it symmetric to reduce memory load, but I understand your point.

[Zelandakh]

mycroft, on 2011-December-13, 18:45, said:

Relay breaks tend to be set up to invert the one-way direction of information flow, and if designed well, also invert who becomes declarer, because having the described hand declarer is more damaging than having the lead go through the strong(er) hand.

Perhaps your relay breaks work this way but by no means all. I use relay breaks primarily for stopper asks, something otherwise difficult to do in a relay system, but also to convert to natural slam bidding on hands where that is more efficient, for example because it is best to check on a side suit control before passing game or because RKCB is going to be more efficient than relaying.

This effectively continues the one-way information exchange but converts the information being given from one form to another. Others feel that it is more important for the unbalanced hand to always describe to the balanced hand and in this case the information flow is indeed inverted. As a regular poster in this forum you must surely be aware that there is almost always more than one successful approach to any given bidding issue.

[mycroft]

Heh - I don't play a relay system at all :-).

And note the "tend to".

It is good to know that there are relay breaks that don't turn into reverse relaying, that's sensible, and I wouldn't have thought of it straight up. I'm sure that my point that the system is still designed to make the less-described hand declarer as much as possible still stands, though.

[gwnn]

BTW I'd be interested in playing a few sessions online with this system, anyone in?