[perko90]

When I was first introduced to Stenberg, the rebids looked something like this after a 1M-2NT* start:
3♣ = minimum (might still be enough for game); 3♦ by Responder asks for shortness
3♦ = extras, no shortness
3♥ = extras, stiff C
3♠ = extras, stiff D
3NT = extras, stiff other M
4X = extras, void showing

That seemed a bit backwards to me. The weakest opening hands were allowed the most room for exploration. So, I changed it to this after 1M-2NT*:
3♣ = extras, with shortness; 3♦ by Responder asks for shortness (shows a full opener)
3♦ = extras, no shortness
-- Responder has 2 ways to show the LR after the above rebids to temper ptr's expectation, but not kill slam either: 3M = LR w/ good controls and 4M = LR w/ bad controls
3M = Dead min; sign-off attempt if opposite a LR
3 other M = min, but enough for game opposite a LR
3NT = balanced, bigger than strong 1NT (18-19 for many)
4m = Good 5-card suit (2 of 3 tops), decent or better trumps
4M = bad hand for slam, but 6+ in M

I'm pretty happy with the above modifications, but I'm just curious what others are doing.
I can't read all the Stenberg web pages because they're often in Swedish.

[akwoo]

The problem with your scheme is that there is no way for responder to ask for shortness over a minimum. The situation when opener has a minimum and responder has a minimum game force is when the shortness information is most useful. Consider ♠AQxxx ♥Kxxx ♦Kxx ♣x opposite ♠Kxxx ♥AQx ♦Ax ♣xxxx

So far I've only played this in a Precision context, where opener is limited enough that you don't need 3 ranges - so anything that accepts the LR is considered to have extras (with the exception of some hands with extra trump, which bid 4M).

I play the 4 level bids as you've defined them.

[nullve]

perko90, on 2020-December-18, 15:21, said:

That seemed a bit backwards to me. The weakest opening hands were allowed the most room for exploration.

Since 2N is LR+ only, the main objective must to be able to stop in 3M opposite a sufficiently weak Opener. So Opener's 3M rebid must have a range so that Responder knows what to do with a LR, and higher rebids must promise at least some extras since they are GF.

A second objective must be to minimise information leakage, and this can be done by putting all sufficiently weak hands into one rebid which can only be one of 3♣, 3♦, 3♥(M=♠) or, if the range is made narrow enough (at the cost of increasing the information leakage), 3M.

A third objective may be to introduce symmetries so that hands with the same shape can be treated similarly regardless of the range. (Symmetries also reduce the memory load, of course.) But then the ranges probably need to be of roughly equal size. (E.g. bottom vs. top half of 1M range.)

The modern "Swedish" structure over 1M-2N, which I take to include

3♣ = bottom half of 1M range, unsuitable for 4♣+
...3♦ = GF relay
...(...)
...3M = LR (NF)
...(...)
3♦ = top half of 1M range, no SPL
3♥/3♠/3N = top half of 1M range, SPL ♣/♦/OM, unsuitable for 4♣+
4♣+: rare hand types,

seems to meet these objectives very well, although I can't help thinking that something like

1M-2N; ?:

3♣ = bottom third of 1M range (including 12-14, 5M(332))
...3♦ = GF relay
......3♥ = 12-14, 5M(332)
......3♠+ = A(M)
...(...)
...3M = LR
...(...)
3♦ = middle third of 1M range (excluding 15-17, 5M(332))
...3♥ = GF relay
......3♠+ = A(M)
...(...)
3♥ = 18-19, 5M(332)
3♠+ = top third of 1M range, A(M)

A(M):

E.g. the simple

3♠/3N/4♣ = SPL OM/♦/♣
4♦+ = no SPL (but not strictly BAL),

would be better in a system like 2/1 which has a very a wide 1M range.