[Bbradley62]

http://tinyurl.com/7zvj6u9

Shouldn't the description of 2N include "3- S", not "5- S"?

[barmar]

The rules for 2NT have lower priority than the rules that say to bid a new suit, so the max length comes from negative inference. But GIB isn't able to perform this negative inferencing automatically when generating descriptions. It's tricky going through all the 2NT rules and figuring out precisely what shapes they each exclude, because they can be reached via different paths. So they just all say that they're semi-balanced.

[calm01]

Diamonds must be longer than spades otherwise spades might have been bid before diamonds. The exceptions might include 5 rotten spades and 5 superb diamonds in what is likely to be a slam going hand. I could, in these circumstances, understand GIB treating 5 rotten spades as a four card suit.

Even if GIB had 5 rotten or a normal 4 spades it would now surely bid them or repeat the diamonds. So with at least 9, and a likely 10-11 cards in spades and diamonds, it is unlikely to have a balanced hand. So 2NT is an inappropriate bid unless described as a forcing but waiting bid denying a decent 6 card diamond suit.

So GIB's inferences must be faulty in still retaining the possibility of 4/5 spades.

[barmar]

calm01, on 2012-January-03, 20:15, said:

Diamonds must be longer than spades otherwise spades might have been bid before diamonds. The exceptions might include 5 rotten spades and 5 superb diamonds in what is likely to be a slam going hand. I could, in these circumstances, understand GIB treating 5 rotten spades as a four card suit.

Even if GIB had 5 rotten or a normal 4 spades it would now surely bid them or repeat the diamonds. So with at least 9, and a likely 10-11 cards in spades and diamonds, it is unlikely to have a balanced hand. So 2NT is an inappropriate bid unless described as a forcing but waiting bid denying a decent 6 card diamond suit.

So GIB's inferences must be faulty in still retaining the possibility of 4/5 spades.

The issue is that GIB doesn't describe negative inferences, i.e. hands it can't have because it would have bid something else. GIB's bidding logic is (much simplified) essentially:

If the auction so far shows A1 and our hand matches criteria C1, bid B1, which shows S1.
Else if the auction so far shows A2 and our hand matches criteria C2, bid B2, which shows S2.
Else if the auction so far shows A3 and our hand matches criteria C3, bid B3, which shows S3.
and so on.

When describing a bid, we show the conjunction of all the Sn for the bids that have actually been made. It doesn't filter out all the Sn for rules that were bypassed because the criteria didn't match.

Conceivably, when writing S2, we could include the negation of S1, and when writing S3 we could include the negation of S1 and S2. But you can see that as the sequence gets longer, this gets cumbersome. It's also not always true, because there can be overlap in the criteria. And often the same rule will be part of multiple sequences like this, so what is excluded depends on which particular sequence happened to occur. This makes it difficult to write the Sn to properly exclude things. The way bidding rules are written doesn't make it easy to pull in all the impliciations from bids that could have been made but weren't. We'd have to split up many rules into multiple rules that explicitly check for these differences in the preceding auction.

So while it may be possible to get explanations like you want, it's not very feasible. And not as high priority as the other things we're working on. We'll do it in certain cases -- that's how we improved the explanation of 1NT-2♣-2♥-2NT to say that there are 3- ♠. But this is a self-contained sequence, the Stayman rules aren't shared by any other bidding sequences.