[Winstonm]

This theme has occured to me many times before, and once again it was brought up in another post that Justin started. I've posted a similar hand for convenience of comparison sake only - don't take this hand as the end all of the discussion.

Scoring: IMP

1D-1H
2N-?

---

A couple of posters suggested an self-splinter of 4D. This begs the questions of what IS a splinter, what should it show, and what is ITS GOAL.

There seems to me only two ways to approach a splinter bid: 1) as a isolated-suit bid, i.e., announcing slam interest and control in one specific suit or 2) as a short-suit slam try, i.e., announcing values in 3 suits outside the short suit - more of an exclusion bid - this bid would increase the value of any cards held outside the splinter suit, allowing an increased valuation of secondary cards.

For example, partner opens the bidding 1H and you hold these hands:
xxx, KQxx, x AKJxx versus Axx, KQxx, x KJxxx

If you chose to splinter with both hand types, how is partner to evaluate this?
QJx, A10xxx, xxx, AQ

Seems to me that a choice is necessary - either a splinter highlights a control in a single suit or it highlights cards held in 3 suits, those outside the splinter suit.

What say ye? What does it do - what is its nature? (Marcus Aurelius as quoted by Hannibal Lecter.)

[bhall]

When we Humpty-Dumpties make a bid, it means exactly what we want it to mean, no more and no less... And woe betide the partner who guesses wrong.

Seriously, the context in which the splinter occurs is crucial. When both "patterning out" and splinter sequences are available, it is reasonable to use the splinter to show values in all other suits. But that is hardly a universal practice (see the 2/1 Gitleman thread). Generally, one is forced to prioritize the information to be shared, and the best use of space may be to start with a splinter, even though it is a jump. I think that would be the approach in most expert partnerships, but I await their comments.

Your example hand, if it can start with a forcing 3♥ rebid, might best be cue bid toward 6 or 7♥: 3♥-3N-4♣ or 3♥-4♦-4♠.

[pclayton]

This is a great question.

As responder, I'd love to be able to make a call that says: Hi pard, I have 6 hearts, a stiff diamond and no side four card suit. Go ahead and ask me for key cards please because I think we have slam.

Regrettably, Opener will frequently sign off in 4♥ and we are back where we started.
As I mentioned in the other thread, a la Meckwell, what if responder could answer keycards with his next response?. Only in a very rare case does Responder wish to follow-up and seize back control after making a splinter.

I know very little about Turbo, but doesn't Turbo more or less do the same thing, albeit in a different way?

edited: I watched a Lauria / Versace auction this afternoon. 4N seems to show an odd number of keys, while cue bids show an even #.

[Winstonm]

Quote

As I mentioned in the other thread, a la Meckwell, what if responder could answer keycards with his next response?. Only in a very rare case does Responder wish to follow-up and seize back control after making a splinter.

This, IMO, get back to a secondary issue of bidding and that is the captaincy approach. While captaincy works well when a hand is finitely describe, a la a NT bid, it lacks - again IMO - when a cooperative approach is needed.

I have trouble with the concept of keycards combined with cue bidding, as to me cue bidding is a combination of card-showing, interest showing, and judgement.

As I said before, IMO the best cue-bidding is more art than science - and the two do not mix very well.

But then I am biased against an overreliance on keycard showing as the best avenue to slam.

[Winstonm]

Quote

Seriously, the context in which the splinter occurs is crucial. When both "patterning out" and splinter sequences are available, it is reasonable to use the splinter to show values in all other suits.

What about hand #1 (xxx, KQxx, x, AKJxx) as in this auction:
1H-2C
2H-4D?

I have seen a secondary splinter as the choice of many posters with this type hand - but my question remains the same - how do you seperate the meaning of the bids when you actually hold, Axx, Kxxx, x, AQxxx?

Actually, this secondary splinter is a concept with which I have been mentally tinkering - is a splinter the best use of a secondary jump?

The key to any low-high-card slam is tricks - and the principle reason - IMO - to use a splinter is to investigate and find good-fit, low-high-card slams.

Wouldn't it be better to use a secondary jump to show a good primary suit and the jump is in the lowest outside controled suit and not necessarily a singleton? In this way, a secondary jump in the same suit would deny outside controls. For example:

xxx, KJx, Kx, AQJxx with this auction:

1H-2C
2N-4D

versus this:

xx, KQxx, xx, AKJxx

1H-2C
2N-4C

A side benefit to this concept would be that a direct splinter bid, i.e., 1H-4D, would deny a source of tricks by way of a good secondary suit.

[kenrexford]

This discussion is interesting to me, and I have put a lot of thought into many sequences involving splinters. I believe that a lot of work could be done in this area.

In the context of a 2/1 auction, I like the "picture splinter" idea. To explain with a simple example, after 1♠-P-2♣-P-2♦-P-2♠, Opener could splinter with a 4♣ or 4♥ call, either of which identifies the stiff (never void), announces (as a direct action) strong trumps (two of the top three), announces strength in Opener's side suit (three of the top four), and denies a control in the four suit (could be headed by the Queen at best).

I wrote up a lot of my thoughts for other situations on my blog (archived if interested at www.cuebiddingatbridge.blogspot.com -- in the "inferences from redundancy discussion). For instance, it seemed to me that a direct splinter (e.g., 1♥-P-4♣) should be done with a hand that is wrong for cuebidding, meaning that a predicted cuebidding sequence would be problematic. This might mean poor holdings in the suit with which you would otherwise start a 2/1 auction, poor trumps, and/or a control in the side suit.

Whatever the nuances you elect, a cuebidding sequence that "proves" ability to splinter (but not a picture splinter auction) carries inference from the non-splinter, and the splinter carries inference from not cuebidding.

However, all that said, the fewer your alternatives, the fewer the inferences available and the more baggage that any one call must carry.

Also, note that the "depth" of the splinter is important. The auction 1♠-P-4♥ is a sequence where Opener either commits to the five-level or does not. In contrast, 1♥-P-3♠ allows Opener to bid 3NT, 4♣, or 4♦, Responder to bid 4♣ or 4♦ after 3NT or 4♦ after 4♣, and Responder to bid 4♦ after an initial 3NT and a 4♣ call from Responder. That's a lot of space with which to work out specifics. Accordingly, 1♠-P-4♥ should ideally be well-defined but 1♥-P-3♠ might be wildly nebulous.

[Winstonm]

Quote

Also, note that the "depth" of the splinter is important. The auction 1♠-P-4♥ is a sequence where Opener either commits to the five-level or does not

A quick, off the top of my head response would be to make 1S-3N a heart splinter.

[rbforster]

Winstonm, on Jan 13 2008, 12:40 PM, said:

For example, partner opens the bidding 1H and you hold these hands:
xxx, KQxx, x AKJxx  versus Axx, KQxx, x KJxxx

In the first case with 4+ trumps and a good side suit source of tricks, I make a strong jump shift, Soloway style, which could either be my own very strong suit or a GF+ hand with support and a good side suit. Often you will be able to show shortness as well. For example,

xxx KQxx x AKJxx

1♥-3♣ (Soloway)
3♦(1)-4♦(2)
...

(1) - waiting for clarification
(2) - heart support, a good club suit, and diamond shortness

New suit bids by responder after the jump shift confirm heart support and a good side suit and show outside shortness. Returning to hearts would again show a good suit and support, but deny shortness (with 3♥ being stronger than 4♥). Rebidding the jump shift suit would deny support and show a very strong single suiter.

Alternatively, without a good side suit, I prefer either Jacoby 2NT or a splinter with GF values and 4+ support. In this case, the suit isn't good enough to show a strong jump shift (min 2/3 of AKQ) so I would splinter.

Axx KQxx x KJxxx

1♥-4♦
...

I believe it is useful/common to have an agreement for the value range on direct splinters. For example, I play that direct splinters show either a min GF or a slam invitational hand. If you're in the middle where you won't be confident bidding on over partner's 4♥ signoff, bid Jacoby instead.

[Winstonm]

Thanks for the input, Rob. Is true there are ways to bid hands with good suits and support.

The basic question I am asking in this thread is this: should a splinter emphasize the control in the splintered suit or should it emphasize cards excluding the splintered suit.

x, AKxx, xxxx, AKxx
x, Kxxx, Axxx, AKxx

Are both of these splinter bids over a 1H opener? Should they be?

[blackshoe]

All this stuff kinda fits together.

Max Hardy made a pretty good argument that the normal range for a Jacoby 2NT response is too wide - when responder is minimum, the partnership has the wrong captain. So he suggested that a Jacoby 2NT response should show definite slam interest, possibly with a side shortage.

He also suggested "under jump shifts" and "over jump shifts", to split the range of splinter bids and also to hide in which suit responder is short until, and unless, opener wants to know.

So you might have an auction like 1S-3NT (over jump shift)-4S, where opener has no slam interest no matter where the shortage is. Hardy's recommended point range here is a good 12 to a bad 15.

What about balanced game going hands (which can no longer be shown by a Jacoby 2NT bid)? Simple: trump swiss; 4C showing a balanced 12+ to 15- with 4 good trumps, and 4D showing a similar hand not good enough for 4C.

I'm not at all certain this is the best possible raise structure, but I think it's better than Bergen (even the full-blown Bergen nobody plays when they claim to be playing "Bergen Raises" :-).

[BillHiggin]

Rob F, on Jan 13 2008, 05:59 PM, said:

I believe it is useful/common to have an agreement for the value range on direct splinters. For example, I play that direct splinters show either a min GF or a slam invitational hand. If you're in the middle where you won't be confident bidding on over partner's 4♥ signoff, bid Jacoby instead.

If (and that is a big if) you want the two way type of agreement for a double jump shift, then it may be useful to look at turbo. I find it almost criminal to make a splinter bid and then take control over partner's signoff (here is information about my hand that may help you evaluate yours, but I will ignore your evaluation!).

With turbo, responder can show the "really big splinter" over the signoff by bidding 4N with an even number of key cards (kickback turbo is useful when the splinter is in response to 1♥ - but not critical) or cue bid with an odd number. At least you are not excluding partner from the discussion.

Having said "this might help if you want ...", in reality I do not like the split range treatment at all.

[kenrexford]

Winstonm, on Jan 13 2008, 05:45 PM, said:

Quote

Also, note that the "depth" of the splinter is important. The auction 1♠-P-4♥ is a sequence where Opener either commits to the five-level or does not

A quick, off the top of my head response would be to make 1S-3N a heart splinter.

Well, of course. Any space savings increases the utility of the bid. Unless I am misunderstanding, I'm not sure how responsive that is to the point.

[Winstonm]

Quote

I'm not at all certain this is the best possible raise structure, but I think it's better than Bergen (even the full-blown Bergen nobody plays when they claim to be playing "Bergen Raises" :-).

You may be interested in my Nov 2007 Bridge World article: Better Bergen Bidding.

The basics are these and designed to seperate the quality of hands when supporting opener's major in non-contested auctions:
1M-3M=forcing raise, 12-15 and 2-4 aces or kings
1M-3C=limit raise or 16+ balanced raise or strong jump shift with support*
*3D then relays as an ask or 3M signs off opposite a limit raise.
1M-3D=mixed raise or poor-control (0-1 ace or king) game forcing raise. (Although it is also possible to eliminate the mixed raise and use 3C for a 4-card limit raise and 3D as a 3-card limit raise - note this would free the forcing NT and subsequent jump-support for something other than 3-card limit raise.)

Notice that all Jacoby-type hands are incorporated, so 2N is free to be used as a natural bid.

[Winstonm]

kenrexford, on Jan 13 2008, 06:31 PM, said:

Winstonm, on Jan 13 2008, 05:45 PM, said:

Quote

Also, note that the "depth" of the splinter is important. The auction 1♠-P-4♥ is a sequence where Opener either commits to the five-level or does not

A quick, off the top of my head response would be to make 1S-3N a heart splinter.

Well, of course. Any space savings increases the utility of the bid. Unless I am misunderstanding, I'm not sure how responsive that is to the point.

Not on point - thus, top of head.

More on point - the direct splinter is such a space-consuming bid that IMO it needs finite parameters to be effective. Perhaps in theory it is best to have 4C, 4D, and 4H as seperate concepts but in practical play the edge may not be worth the additional memory work required.

For argument, let's say we agreed to play 1M-3M as the old-fashioned forcing raise.
In that case, I submit that 1M-4-splinter should be the same hand but with a feature, the feature being the singleton.

It appears the question I am asking is whether or not this bid should be an asking-type bid (are your values outside this short suit) or a telling bid (I have game forcing values with shortness in this suit.)

I believe it important to define which type of bid is being made.

[blackshoe]

Winstonm, on Jan 13 2008, 07:23 PM, said:

It appears the question I am asking is whether or not this bid should be an asking-type bid (are your values outside this short suit) or a telling bid (I have game forcing values with shortness in this suit.)

I believe it important to define which type of bid is being made.

Oh, it's definitely important.

Rosenkranz (and probably others) has argued that asking bids are more useful when one partner is likely to be significantly stronger than the other - for example, after a strong opening bid. Hardy has argued that splinters are most useful as an aid to hand evaluation, rather than necessarily an aid to getting to slam (although the splinters we've talked about here do that too). Rosenkranz also mentioned that the Romex folks tried mini-splinters for a whle (for example 1♥-2♠ with 4 trumps, 6-9 HCP, and a side shortage) but found them more useful to opponents, so abandoned them.

Looking at Hardy's comment, it seems a GF splinter is a bit of both asking and telling - for him, it says "I have a side shortage, if that interests you, ask and I'll tell you where" and "please reevaluate your hand in view of this shortage".

[kenrexford]

I think that most folks look at direct splinters all wrong. The question should not be, IMO, whether the bid is an asking bid or a telling bid. The question should be as to what works best in the context of a whole systemic approach.

Take, for instance, the two hands proposed in the initial post (responding to a 1♥ opening):

♠xxx ♥KQxx ♦x ♣AKJxx
♠Axx ♥KQxx ♦x ♣KJxxx

To me, the first is wildly inappropriate for an immediate splinter, because I have a better option in mind. I start with a 2/1 2♣ call, expecting a very high likelihood of a 2♦ rebid by partner, after which I can establish hearts as trumps (2♥). When it matters, partner will cuebid 2♠. If he cannot cuebid 2♠, and hence lacks a spade control, I will sign off and keep all things a mystery. If he bids 2♠, I can jump to 4♦, which systemically for me will show two of the top three hearts, a stiff diamond, three of the top for clubs, and no spade control. Exactly what I have. So, no splinter. (The alternative-auction scenarios play out relatively well, also.) This has nothing to do with "definitions," or "ask-vs-tell." It has to do with alternative options and avoiding redundancy.

So, what about the second hand? With this hand, I could also start 2♣. I still expect a 2♦ rebid to be frequent. If so, I bid 2♥ setting trumps. If partner can cue 2♠, that is nice. 2NT is no concern, as I have spade control. So, I think through what my most-likely-sequence options will be. I cannot cue 2NT because I have good trumps. I cannot cue 3♣ because my clubs are lousy. I cannot cue 3♦ because I do not have a diamond honor (with stiff Queen, I may very well opt 2/1 with this pattern). I end up with a redundant 3♥ cuebid, confirming a spade control (if Opener bypassed 2♠) and confirming the two top hearts. Little is shown of my hand, except negatives. If life is good, and partner bid 3♠ or 3NT, I have a tough time bidding the club control and the diamond shortness control, as they are touching controls. If partner cues 4♣, I cannot know whether it is the Ace or the Queen. I dislike the most likely follow-up, and the less frequent auctions are worse. So, I start with a splinter.

Again, the decision on the second option is not so much "asking" or "telling" or definitional, although a "definition" exists -- a "cuebid troubled splinter." Parameters are general, case-by-case specifics might be inferred by partner looking at this hand.

Now, one could define the immediate splinter by ask-vs-tell standards, as I do with picture splinters. But, the difference IMO is that the direct is too early, without any prep work. A specific picture splinter only makes sense in the context of a foundation of a few initial calls. The direct splinter, if too defined, would be extremely remote in occurrence. I consider the direct splinter to be the ugly stepchild, gaining its definition from inference of all other options. Not right for a picture splinter, and not right for cuebidding. The residue, defined by predicted problems.

[Winstonm]

Thanks for contributing to the discussion, Ken. Your points are well taken, and my own personal bias would be to bid 2C, as well - but that is because I have the luxury of doing so within the parameters of the system I use.

I brought it up as an example as I've seen similar type hands where other posters would splinter with the hand and wonder WTP?

To me the problem is with the partner of the splinter bidder - how does he evaluate his hand in light of the splinter - this is at the heart of the tell vs ask IMO.

You brought up the good point about a splinter in the 1-under suit. Suppose I had not discussed with my pick-up pard the nature of splinters but we both agreed to play splinters. Then I pick up QJ9, AJxxx, Axx, Qx and we have this auction with me opening the bidding.

1H-2C
2H-4D*

What am I to do? In this particular case, I cannot know unless we have a specific agreement - if it shows cards are held in all other suits excluding diamonds, my hand is quite good. Likewise, as you pointed out, if it denies a control in the unbid there is no strain on my choice of action. However, if all I know about partner's hand is that he is short in diamonds and may or may not hold a spade card, I am only guessing about what to do.

One cannot have it both ways and retain accuracy.

I am at a loss as to which method is superior - but I am confident that the undefined approach is more luck than skill when it works.

I really think direct splinters should be of the exclusion variety as the nature of these bids should be a hand without a good outside source of tricks - without a good, long suit.

The benefit of this approach is in the upgrading of secondary honors - Q10x, AKxxx, xxx, Kx becomes a much more slammish hand if a diamond splinter promises cards in all suits outside of diamonds.

[kenrexford]

I agree with your assessment, and I think we are reaching the same conclusion for the same reasons, just articulated differently. The direct splinter that describes "scattered values" and a "poor anchor suit" is also a cuebidding liability and inappropriate for a picture splinter after exchange of foundational information.

Whether the parameters for such a call are understood in great detail or understood on a more generalized feel basis determines the effectiveness of one's fielding of the calls, but we are probably talking rare occasion for critical nuance.

In contrast, I agree that the idea of point ranges for splinters being " 'nuff said" does not work well unless it happens to work well.

[miguelm]

The nature of the bid is quite simple actually. It asks partner to evaluate his hand as if you were playing with a 30 points deck

I have been using Splinters all my Bridge life, but only recently I started to use them with some rules.

After a Major opening bid, if you hold a GF hand with support for partner's suit and you play 2/1 you will have several ways to bid your hand.
So, IMHO it is just reasonable that you define the overall features that define each path.

As for Splinters, I like to:
- only splint with singleton, never with void (this is just a matter of personal preference...)
- have all side suits controlled.
- hold at least 4/5 italian controls.
- have always 4 card support.
- deny a good 5+ cards side suit.

So, attending those hands in the first post

xxx, KQxx, x, AKJxx
This one always starts with 2C. It holds a good side suit. Moreover, it fails to control the other.

Axx, KQxx, x, KJxxx
A splinter, although a minimum.

[Jboling]

miguelm, on Jan 14 2008, 06:24 AM, said:

The nature of the bid is quite simple actually. It asks partner to evaluate his hand as if you were playing with a 30 points deck

I share this view especially, Goren mentions in one of his books that with a 8 card trump fit and a shortness you only need on average 26 points outside the shortness suit for 12 tricks (this could probably be more accurately analyzed with todays tools). And based on this I am prepared to relax the requirement for controls in the side suits for the splinter-bid, if you have 26 points in the other suits you have all suits controlled. So if opener upon hearing a splinter bid from partner sees that we have about 26 points in the other suits, he should consider slam, often just check for keycards. The more narrow limits for the strength of the splinter bid the better of course.

I also prefer bidding good 5 card side suits before splintering, except if the suit is solid (AKQxx or better). A side suit bid tells opener that a honor in that suit should be upgraded, while a solid suit does not need any help, so a splinter is better in that case.

And I would like to add one more requirement for a splinter bid, one should not have a singleton honor in the splinter suit. Partner will strongly devaluate his holdings in the splinter suit, for example KJTx works much better against Q-singelton than x-singleton.

But what to do with hands with singleton honors? I think you have to lie about something, unless you want to invent something complicated for a rare situation; bid the hand as a balanced one, or ignore the singleton honor and splinter. When bidding the hand as a balanced one you can often get information about partners holding in the shortness suit, so that is typically my preference especially with higher honors; a jack might be better ignored, a queen is a borderline case.