The Obamacare Syndrome

Well, here we are again with congressmen grilling a government official about the failure of a project over which she had little control and about which the accusers understand very little. One wonders if these people could write even 5 lines of working code. Get rid of Sebelius, they imply, and the sun will shine once more across our fair land. Of course, it is a systemic error which they are trying to lay at the feet of an accused individual. Over 80% of the large government software contracts fail after completion, which requires an additional large expenditure to fix. To start, the government, in its role of welfare agency to large corporations, spreads the money between several contractors who provide pieces that are guaranteed to succeed in their allotted tasks, but once put together don’t often produce to required outcome. The government simply hasn’t the resolve or the resources to provide the necessary oversight. That’s how the blame game works.

We can apply the same analysis to the bidding and play of a bridge deal. The 4 players are like computer programs that are doing their assigned tasks according to the definition of their individual responsibilities. If the play ends in failure, the critics are out trying to lay the blame after the fact when it may not be any individual’s fault, but a systemic malfunction.  The key is how information is shared between the participants.

In the previous blog we discussed an auction during which a player missed the chance of making a takeout double of a natural 1♣ opening bid when holding modest values with 4 spades and 3 hearts. The cast has changed on this deal played recently for the Lederer Trophy by world-class experts during which a very similar situation arose. The end result was far from optimal. Who, if anyone, made a mistake, do you think?

 

Allfrey’s 1♣ opening bid was systemic, as a 1♦ opening bid would require at least 5 diamonds, so highly regarded is that lowly suit by some. Just because you can doesn’t mean you should. In years past players didn’t feel the need to advertise an average holding, because that was what your partner would assume you had. In fact, this is a below average holding that doesn’t contain the normal component of 3 controls and has 10+ losers. Its Zar Points (19) don’t come close to the requirement for an opening bid (26).

To most experts Zia’s collection barely qualifies for a takeout double of 1 of a minor, even though it does not include 4 cards in both majors. This is a moot point to which we shall return later. However, we can count 9 losers, and one of the suits is clubs, the suit named by the opening bidder. If NS can make even 8 tricks total, North would have to have a 7-loser hand, the normal requirement for an opening bid, as well as the cards to make up an 8-card fit. He might be tempted to overbid, but the mood of the times is to bid first and escape the consequences later.

A lesser player than Robson might bid 3NT as this point, but Robson is the wait-and-see type. He sees 13 HCP in his own hand, gives partner and Zia 12 each, leaving his LHO with approximately 3 HCP. If the opposition is going to hand out information, why not take advantage by listening quietly? Bakhshi shows hearts after which Robson takes charge and maneuvers Allfrey into being the declarer in 3NT, putting the strong hand on lead. The question is this: has the exchange of information been more helpful to Allfrey or Zia? Certainly it has ruled out a spade lead and invited a heart lead.

Zia led the ♥2 to ♥5-♥K- ♥8.  Bakhshi had already made a potentially fatal mistake, according to Deep Finesse. With no entry outside of hearts he has to duck the first trick in order to get back to his hand to cash winners. Of course, he had every right to expect Zia to hold 4 hearts. Or did he?

Allfrey now sees what Robson anticipated at the very start – Zia probably holds all the remaining aces and faces. Declarer’s mind turns to a strip and endplay. A heart has been lost, and there remain 2 aces to lose. The ♠Q may be another loser, and perhaps another heart. One too many it seems. So how to engineer an endplay to reduce them to 4?

Allfrey took the heart return with dummy’s ♥A and worked on diamonds. Zia won the ♦Q with his ♦A and cleared the hearts. Playing on the minors, declarer reached the dummy with the ♣ K with 6 tricks won and 2 in hand:

 

Allfrey’s choice was to play a club and duck Bakhshi’s ♣ T, hoping that Zia had been caught with the bare ♣ A.  The result was down 2, when he would have made his contract by leading a spade from dummy. He was playing Zia for his double to have 4=4=2=3 shape rather than 4=3=2=4. I don’t see there was any clue in the play of the heart suit to guide either side. On winning the ♥K Bakhshi returned the ♥4, not his original fourth highest ♥6. This would have been an occasion to inquire about partnership agreements. That was painful, as East at the other table passed initially and subsequently was able to play safely in 3♦ making 110. 

We have a bad result that is not attributable to any of the individuals, so it comes down to considering whether there was a systemic fault. The quality of information available to North and East was not good enough for either of them to make the correct plays. Let’s take a top-down approach by setting the goal and then consider ways it can be reached. The objective is to maximize the gain, not reach the safest part score. Two balanced hands with 24 HCP and no 8-card major fit are the stuff of which a vulnerable 3NT is made. This is the essential information which EW must transmit.

The system should provide information in a way that gives the best chance of avoiding the killing lead. If one considers the East hand as one that demands an opening bid, then I suggest that Allfrey return to his ACOL roots and open an upgraded weak 1NT (12-14 HCP). This will most likely get him to 3NT easily made on a spade lead. Of course, the vulnerability makes the weak NT dangerous, so many would avoid that bid.

Opening 1 of a minor is not safe, either, and it will be less likely to shut out South. A wide-ranging takeout double that contains little information with regard to distribution is a nearly risk-free way to get into the auction cheaply. Over this takeout double 3NT looms large on responder’s horizon. Redouble should be limited to lesser hands that do not merit a game contract. Does it make sense that the player with the best hand at the table is constrained to pass? That is a system fault that needs fixing.

Better methods would allow East to announce game-going values while shutting out the advancer. One could define 2♦ as game forcing Stayman, just as if the opening bid had been a weak NT. It will be easier to play the hand with all the defensive power in the doubler’s hand, so one can afford to be bold. It is best if West does not describe his hand with a spade bid, as he would if South passes. It is better to seek a major fit by asking about West’s major suit holding without revealing his own. If East bids 2NT, West will raise to 3NT leaving South without input from North with regard to the best opening lead. In this way EW neutralize the effect of the takeout double and turn the interference to their own advantage.

There are other tasks we expect a constructive system to accomplish in an optimal fashion, but one of the hardest things to teach a computer is commonsense. Here is an example provided by Mike Lawrence in the November issue of the ACBL Bulletin. West opened 1♣ and East, acting like a bad piece of software, jumped to 3NT. Six heart tricks were lost off the top. Unlucky! Mike suggests the patch given below.

The system should provide a way to explore for a 3NT contract, failing which, a route to 5♣ . Let’s assume a matchpoint background. First, if the responder decides to minimize the loss if they reach the wrong contract, he may jump to 3NT. The blind odds favor the move, but if this contract fails, his loss will be small, as many HCP addicts will bid the same way. Fishing around for stoppers can lead to the loss of an overtrick. At IMPs scoring, the overtrick doesn’t matter that much, but still one wants to minimize the chances of a killing defence. The start of 1♣ – 1♦ is a fine way to invite the opponents into the auction, but is that what responder wants here?

If responder looks to maximize his gain if he gets it right while others get it wrong, he will consider 4♣ or 5♣ as the alternative to 3NT. Making a club contract when 3NT fails will score well. East is willing to exchange information towards that end, although he realizes it may help the defenders if, after all, EW end up in 3NT. Lawrence copes with this possibility by having responder lie about diamonds then lie about hearts. When opener avoids 3NT, is 4♣ a cue bid looking for a slam with diamonds the agreed suit? I find the sequence ambiguous and would fear ending up in 5♦ or 6♣ . Can responder expect opener to cooperate sensibly?

Lawrence himself supplies the solution: initially responder should bid 2♣ as a forcing inverted raise, denying a 4-card major, after which ‘you can expect a sane auction.’ (That tells the reader how Mike rates his own sequence.) There are great advantages to supporting with support at the very start. First, the defenders are shut out of the auction at the 1-level. Opener can jump to 3♥/2♣ to ask for a heart stopper in 3NT. When responder bids 4♣ , denying major suit stoppers, opener can take a shot at 5♣ , knowing that clubs represent a good fit, or he may suggest playing in a 4-3 spade fit, just in case responder holds ♠JT7. With a less promising hand, opener can pass 4♣ . So, we conclude that system does matter… a lot. The information that is exchanged should bear directly on the target contracts. Diamonds are not relevant, so there is no need to exchange information directly with regard to that suit, unless you want to add a bit of razzle-dazzle to distract the opposition.