Source: http://andrew-gumperz.blogspot.com/

I was thinking recently about the development of the modern major suit raise structure. In the 1950’s, the forcing major suit was a jump to 3M.

1M  —  3M  =  12+ and 4+ card support (Forcing raise)

This method was simple but even at that time expert player recognized it had deficiencies. The biggest problem was inaccurate slam bidding. In the 1950’s American experts missed major suit slams routinely. Their bidding woes could be traced to opener making poor decisions about whether to start a slam investigation because he lacked information about responder’s:

  • Distribution
  • Precise Range of Values

Distribution
1M  —  3M
?

Responder
Jxxx, Axxx, Axxx, x

Opener
AKxxxx, x, KQx, xxx

After 1M-3M it may be difficult for opener to envision slam because he has no idea that responder has a singleton opposite his 3 small clubs since the uninformative 3M response says nothing about side suit distribution. Without distributional information, opener can not judge whether his side suit cards are or are not contributing to taking tricks. 

In contrast, a modern splinter raise would describe both responder’s values and his most important side suit feature (the singleton club). With both these pieces of information, opener knows that his KQx of diamonds are valuable and further that two club losers can be ruffed in dummy. Even an average player can envision that slam is likely.

Principle 1: A forcing raise made on a shapely hand should provide enough information about bidder’s shape to allow partner to evaluate whether his high cards are working.

Precise Range of Values
1 —  3
?
AQT9x, AKT, JTx, xx

What should opener bid in response? Slam may be cold, or it may have no play:

  • Kxxx, xxx, KQxx, AKx  Slam is cold
  • KJxx, xxx, KQx, AKx    Slam has no play
  • Kxxx, QJxx, x, AKxx     Slam is cold
  • KJxx, x, Qxxx, AKxx     Slam has no play

A 4 call by opener would show a balanced minimum that is uninterested in playing 3NT–which opener has–however it cuts against the grain to make such a discouraging call with a good 14 HCP facing an unlimited partner. Today, some players might use a convention like Serious 3NT to provide more bidding options, but in those days 3NT was natural so the only alternative to 4 was a control-showing slam try of 4.

Unfortunately, if opener bids 4, responder may take him for a better hand and get too high when responder has a nice 13 HCP. So opener has a blind choice between an over and an underbid. Why does opener find himself in this awkward position? Opener has no idea how many values responder holds. Consequently, he can not tell whether the partnership is or is not in the slam zone. If the 3 call had promised:

  • 16-18 points and 4+ card support, opener could confidently bid 4.
  • 12-14 points and 4+ card support, opener could confidently bid 4.

Principle 2: When a forcing raise leaves little room for exploration, it should limit the bidder’s values to a narrow range so opener can recognize when the partnership is in the slam zone.

The Modern Splinter Bid
With these two principles in mind, let’s examine a modern splinter raise. Splinter raises were first described in print in the late 1950’s and quickly adopted by forward-thinking American pairs. Even BJ Becker who was a notoriously intransigent natural bidder immediately saw the benefit and adopted this radical artificial convention.

Originally, the splinter raise was an unlimited bid. However today, most experts have settled on 9-11 HCP as the range for the direct splinter. I.e.,

1M —  4X = 9-11 HCP, a singleton and 4+ card support.

Opposite this raise, opener can tell:

  • Approximately how many highcards the partnership has between the two hands and hence whether the partnership can possibly hold the values to make slam.
  • Which of his highcards are and are not working and hence whether slam is likely to make.

This is a good convention for finding slams. Opener can easily judge whether the partnership is in the slam zone and slam decisions will be accurate because he can tell whether his own values are or are not working. However, a highly descriptive splinter raise introduces a problem the old timers did not have using the forcing jump raise in a major. The same information which helps opener to decide whether to stop in game or search for slam may guide the defenders to defeat your game contract. Further, when responder has only 9-11 HCP and opener may have as little as 12 HCP some of which can be wasted facing responder’s singleton, a game contract could be in jeopardy which is when you most need to gain a trick on the lead. Lastly, when the splinter raise shows 9-11 HCP, at least 75% of the time the auction will stop in game. A splinter raise can only show a profit when opener is strong enough to participate in a slam search. If opener has a dead minimum, an uninformative forcing raise would get us to our best spot (game) without helping the defense. A better version of the splinter raise is called mini- splinters.   It operates in two stages.

  • Stage 1 reveals responder’s support and shows his range. With this information, opener can often decide immediately that the hands are not in the slam zone. He can immediately sign off in game without telling the defenders where dummy’s singleton lies.
  • Stage 2 reveals the location of opener’s singleton. Only if opener has the extra values/shape so the partnership is in the slam zone will opener ask for responder’s singleton.

1     —    3       =    4-card support, an unspecified singleton and 9-11 HCP
1      —    3NT!  =    4-card support, an unspecified singleton and 9-11 HCP

Opener only asks for the singleton with enough to think that slam is possible opposite the right singleton. His cheapest call asks:
1      —    3
3NT!  —    ?

1      —    3NT
4!    —    ?
!    =  asking for singleton.

Responder shows the location of his singleton in steps:
step 1 = lowest singleton (clubs)
step 2 = middle singleton (diamonds)
step 3 = highest singleton (the unbid major)

Principle 3: Reveal information to the opponents only when you know the partnership is in the slam zone–i.e., when the cost of revealing it will be compensated by better slam bidding. When the limit of the hand is known to be a game, bid game as uninformatively as possible.

Concluding Thoughts
This article was not written to convince you to adopt splinters although it may help you tune how you play them to get more benefit. The purpose was to call out the three principles which apply to all slam exploratory methods. To recap:
  • When the partnership has only game values, get to game without giving information to the defense.
  • One player must limit his values to allow the other to judge whether the partnership is in the slam zone.
  • One player must describe his shape to allow the other to determine whether his high cards are working.*

Can you see which principle(s) the following auctions violate?
Auction 1
1M  —  2M
3Diamond Suit —  4
4M
The partnership has stopped in game while revealing a lot of information to the defense. Further, opener probably only has a hand worth a game try, hence the 4C cuebid gave away the show when the partnership had no legitimate slam ambitions.

Auction 2
1M  —  2X
2M  —  3M!
!  =  forcing
Neither hand is limited. We can anticipate that slam explorations will be awkward without further methods to help us limit one of the hands.

Auction 3
1M  —  2NT (Jacoby. Unlimited and forcing raise)
Neither hand is limited. Further, on the 90% of hands where the partnership ends in game, we do not want to reveal opener’s shape to the opponents.

Auction 4
1m  —  1M
4M
Neither partner has described their shape. The partnership will have a hard time judging when their high cards are and are not working.

Auction 1 has been discussed on this blog my article on single raise constructive. I will discuss the other auctions in the next few weeks to show how the flaws in these common agreements can be repaired.

Footnotes
*  In splinter raise auctions, the same player describes his shape and limits his hand. In other sequences, these functions can be performed by different hands or by one hand using different bids.