[jdonn]

I made up a puppet stayman about 6 years ago (that was similar to versions which had already been independently invented, though I didn't know at the time) which allows for responder to show both his 'smolen shapes' in the stayman bid while always having opener declare. The advantages are all 5-4 and 6-4 and 5-5 combinations can be shown by responder with the fit always being found, giving you more bids on the transfer auctions (2NT 3♦ 3♥ 3♠ and 2NT 3♥ 3♠ 4♥ can be used for other things) and that opener will declare 100% of the time. Disadvantages are complexity, giving them more artificial bids to double or not, and revealing (just a little more than most people) declarer's hand. I don't usually play puppet stayman, but when I do it's clear to me that this one is very good.

Immediate responses to 3♣:
3♦ = either no 4+ majors, or 4 spades
3♥ = 4 hearts
3♠ = 5 spades
3NT = 5 hearts

Over 3♦:
3♥ = describe your spade length
3♠ = 4-5 in the majors
3NT = to play

Over 3♥:
3♠ = 3 or 4 spades
3NT = 2 spades

Over 3♠:
3NT = passable (responder has 4 spades)
4♥ = transfer to 4♠ (responder has 5 spades, theoretically should be 5-4 in the majors)

There are more followups to all the bids, including some things allocated on 4 level over 3♣ 3♦ to show 5-5 majors and ways to show one or both minor suits after bidding stayman, but this is the general structure.

This negates 2NT 3NT having to be 5-4 in the majors as in some structures people have put forth. You could use it as a relay for even more minor suit hands, but I think for practical reasons I would have 2NT 3NT just be to play, and 3♠ relay to 3NT for whatever minor suit devices but with 3NT never to be passed.

Later note: Something like MikeH suggests, which is essentially this but with the ways to show 4 and 5 spades reversed, seems to me to be equivalent to this. Just off my head I think my way would make a 3♦ rebid more frequent, but would allow responder to sign off more often after the 3♦ bid instead of going through another ask, so I have no idea which is better. Probably there is little to choose between them.

[awm]

jmc, on Mar 24 2007, 02:46 AM, said:

Awm, what method were you considering? A partner of mine has recently suggested an artificial 3NT and so far it seems artificially complex to me.

jmc

I wasn't considering anything in particular. It's been a while, but I think I put up this thread in response to an email conversation where it was commented that 2NT-3NT artificial was very popular at the expert level.

For quite some time I played a method where 2NT-3NT was a 4441 slam try short in a minor. My general experience with these sorts of methods was:

(1) Giving the opponents a lead directional double of spades on the way to a normal 3NT (assuming signoff hands went via 3♠) was rather costly. The alternative of having signoff hands go via 3♣ didn't seem much better (combination of the less-frequently-useful double of 3♣ and the information to the defense about opener's shape).

(2) The forgets of 2NT-3NT tended to be very costly, and since the "3NT signoff" hand is much more frequent than any reasonable artificial use of 3NT, it does not take a very substantial "forget probability" before the forgets outweigh the value of the convention.

(3) I do not see very many top players using 2NT-3NT artificial from my experiences in the US nationals. Of course at my age this experience is limited (and 2NT openings are infrequent) so I could be wrong about this.

Currently I don't play 2NT-3NT artificial any more in any partnerships.