## Thursday, July 19, 2007

### Inference from Redundancy?

A few troubling thoughts have kept me up at night over the years, and I believe to be approaching somewhat of a resolution of my angst. I thought I'd share this.
Angst #1: What is 1S-P-3NT? I'm not satisfied with a simple "13-15 balanced with three-card support." This begs the question of whether this is quacks or primes, or a blend. If Opener is balanced and interested in a notrump quantitative bash, this is not that relevant. If Opener is shapely, this is critical. Few answer this question.
Angst #2: When would you elect to Splinter immediately instead of bidding a 2/1 GF and then later cuebidding (or Picture Splintering)?
Angst #3: When is Jacoby 2NT best?
Angst #4: When is 3C, as a route to show a 12-14 Splinter, a better start than a 2/1 and cuebidding, or Picture Cuebidding?
The "definitions" of these calls (direct Splinter, Jacoby 2NT, 3NT, and 3C) do not seem to be tied to theory or strategy, or prepared auctions. The failure to elect these options is rarely, if ever, discussed as tailoring and further refining cuebidding sequences where these options are not elected. I'm not satisfied with that, so I plan to explore this a bit, in a series of posts, where I vent rambling thoughts as I myself think this through.
Delayed Splinters through 3C
The issue popped up recently as a result of my analyzing a specific bidding problem, which relates to the 3C call with an included 12-14 splinter option. So, I'll start with that.
As a recap, my technique uses 1M-P-3C as a multi-faceted bid, showing either 9-12 with four trumps or a 12-14 splinter. 3D asks. 3M shows a minimum limit (9-10 if you will). 3NT shows a maximum limit raise with good controls (primes), meaning 4+ controls. 4M shows a maximum limit raise with poor controls. Four of a new suit is a mini Picture Jump, with typically 5422 shape, a concentration of values in trumps and the side suit, and no controls in the other two suits, maximum limit raise ("maximum" in the sense of not good enough to GF). Three of the other major, the missing bid, shows a 12-14 HCP splinter, which Opener can ask about.
So, why splinter through 3C rather than cuebid?
This call should be a prepared call. Imagine a typical hand. You have QJx AJxx x KJxxx, and partner opens 1H. The alternative option is to bid 2C. You expect a likely 2D call from Opener, which would be nice because you can then set trumps at the two-level by bidding 2H. Life is good.
But, what will happen next? All times that it matters, Opener will bid 2S. You will then cuebid 2NT, showing poor trumps (not two of the top three honors), because you are not right for a Picture Splinter. Let's now assume that partner bids 3C, the cheapest cue possible, which could be based upon the Queen of clubs. You are still not right for a Picture Jump, because although you have clarified the poor trumps, your clubs are also poor. You cannot cue 3D, because you lack a diamond honor. You could cue 3H, though, and show that you have one top heart honor. Great! A cuebid!
Now, suppose Opener makes the cheapest cue available again, 3S. What next? You are too weak for 3NT. You cannot cuebid 4C, because you lack two top clubs. The best that you can do is to cuebid 4D, showing a diamond control that is not an honor -- shortness. Thankfully in this one auction you could at least show this! In any other auction, you could show nothing to partner except the one top heart. Not very impressive, eh?
So, why splinter through 3C rather than cuebid? This hand provides the solution. Responder will "shape bash" a 3C...Splinter auction when he lacks ability to make meaningful cuebids in a predicted auction. The features leaning one toward a "shape bash" 3C call might include:
1. Not two of the top three cards in the suit you would pick for a 2/1 start.
2. Poor trumps (not two of the top three honors)
3. No control in the side suit
On the other hand, you might tend to start with a cuebid with the stiff being the Ace, King, or Queen, because you might elect to show that card as the honor in support of a suit he bids rather than as a stiff.
This list is not a checklist, of course. The idea is to think through plausible auctions that will occur after a 2/1 and to elect the "shape bash" of 3C when cuebidding will be a mess. This will typically occur when perhaps three, or all four, of the criteria above are present (poor trumps, no side control, not two top in the suit you would initiate, and perhaps a stiff honor).
Thinking this through, and discussing this with partner, leads to some inferences. First, a "shape bash" of 3C...Splinter implies three or more of the four criteria above. Opener can better evaluate his holdings this way. Conversely, a 2/1 auction that suggests a likely stiff or void will be done with a hand that does not have three of the four criteria.
Consider the example, again. QJx-AJxx-x-KJxxx. This is a great hand for a "shape bash" auction. You start with 3C in response to a 1H opening. Partner bids 3D, asking. You bid 3S, partner asks (3NT), and you show (4D). Partner will now expect that you have only one top heart, that your diamond stiff is not an honor, and that you only have one of the top three honors in each of the back suits, one of which is the Queen. He will expect that one of these assumptions, but no more, might be inaccurate. Very nice, eh?
What about the alternative? Suppose something about your hand is different. So, you bid 2C. Partner bids the expected 2D, and you set trumps with 2H. Partner cuebids 2S, and you show poor trumps by bidding 2NT. Partner shows a club card by bidding 3C. You deny a diamond card by bypassing 3D to bid 3H, showing one top heart. Partner cuebids 3S. You bypass 4C because you lack a second top club and bid 4D to show the stiff diamond. What is different about your hand? Why did you not splinter through 3C initially?
We know that you have a stiff diamond and that this stiff is not an honor. That favors a "shape bash." We know that you have precisely one top heart honor, also favoring a "shape bash." We know that you lack two top clubs, again favoring a "shape bash." So, you must have more than 14 HCP's. The most you can have outside of spades would be AJ in hearts, J in diamonds, and AJ in clubs, for 11 HCP's. So, you must have 4 HCP's in spades. QJ does not do it. So, you must have, at a minimum, the King. Because Opener cuebid spades twice, he seems to have the stiff Ace or a void, but he can read your spade King by the "redundancy" inference.
What if, instead, the auction were more strained. Suppose that partner's rebid had been 2H. Now, you would have to bid 3H to support hearts. If Opener bid 3S, you would bypass 3NT (non-serious) and 4C (not two top clubs) to cuebid 4D. Now, Opener would have a somewhat vague picture of your hand, but he would be able to assess that you did not have a "shape bash" suitable holding, meaning something different. Whatever his hand is, and whatever his needs are, he would be better placed.
So, the 3C...Splinter should be reserved for "shape bash" hands, meaning three or more of the criteria from above. Not using a "shape bash" 3C approach suggests a stronger hand (15+) or that two or fewer of these criteria are present.