Sunday, February 26, 2023

Free Weekly Instant Tournament - February 24 - Board 3

Board 3
Opponents vulnerable

♠ A K 8 7   A Q 10 5   K 8 4  ♣ 6 4  

I open with one notrump in first seat. Everyone passes. I like this auction. One notrump is frequently a difficult contract to declare. And in this event, where your system is foisted upon you, everyone should be in this contract. So this deal should offer a good opportunity to outshine the field.

LHO leads the eight of clubs.


NORTH
Robot
♠ Q 9 6
7 3
Q 10 6
♣ K 10 9 7 2






SOUTH
Phillip
♠ A K 8 7
A Q 10 5
K 8 4
♣ 6 4


West North East South
Robot Robot Robot Phillip



1 NT
(All pass)


The lead is probably top of a doubleton. I have three spade tricks, one heart, and one diamond, bringing me up to five. I need two more to make this. There are prospects for more tricks in all four suits. That's what makes playing one notrump challenging.You frequently have lots of options with little information at the start. Your goal is to find a flexible line that leaves as many options open as possible until you get a better picture of the layout.

There's no point in thinking about this too hard now. The defense is in charge for the moment. Let's see what they do.

I cover the eight with the nine, and East wins with the jack. He shifts to the deuce of diamonds.

That was unexpected. A heart shift looks more natural. Diamonds must look like a promising source of tricks to East. He probably has ace fourth or fifth. He might have the jack as well, especially if he has five. In this layout, switching from ace-jack fifth gives me a second diamond trick I wouldn't have managed by myself. But his partner might have the king, or I might have king doubleton. Also, giving me a second diamond trick might gain in the long run if the key to the defense is running the diamond suit.

I play low, West plays the nine, and I win with dummy's ten. I'm up to six tricks now. My first instinct is to play a heart to the ten. If that drives the king, I've made my contract. If it loses to the jack, I can decide whether to use my entry with the spade queen to repeat the finesse or to try for four spade tricks.

But maybe I can do better. What happens if I play on clubs? If East wins with the queen and plays another diamond, he gives me a dummy entry. I can now drive the club ace, giving me two club tricks. Even if East began with five diamonds, the defense can take only three diamonds and three clubs, so I've made my contract. If he began with four diamonds, I'll make an overtrick.

More likely, East will win the club queen and shift to a heart. Then I'm no worse off than if I had played a heart myself. In fact, I'm better off, since I've cut the opponents' communications. Having stripped West of his club exit, I may be able to endplay him.

The way playing a club could work out badly is if East wins and shifts to a spade. Now I lose my chance to take two heart finesses. But, looking at that dummy, he isn't apt to shift a spade. If he does, I'll worry about what to do at that point. I'll still have a variety of options.

I play a low club from dummy toward my six. East hops with the queen, and West follows with the three. As expected, East shifts to a heart, the four. I finesse the ten. West wins with the jack and shifts to the three of diamonds. 

Here is the current position:


NORTH
Robot
♠ Q 9 6
7
Q 6
♣ K 10 7






SOUTH
Phillip
♠ A K 8 7
A Q 5
K 8
♣ --

I would like to win this in dummy so I can continue clubs. What's the best way to do that? If I play the queen and East began with ace fourth, he may duck to preserve communication. Then I can continue clubs. If, instead, he takes his ace, I can no longer set up clubs. But, thanks to my club play at trick three, I may now be able to endplay West.

Say, for example, East takes the diamond ace and continues diamonds. I win in my hand and play ace of spades (unblocking the nine) and a spade to the queen, reaching this position:


NORTH
Robot
♠ 6
7
--
♣ K 10 7






SOUTH
Phillip
♠  K 8
A Q 5
  --
♣ --

If West has played low spades, I play a spade to the king. If spades break, the long spade is my seventh trick. If West has long spades, I toss him in to force a heart return, reaping the reward from my club play at trick three.

If West has played an honor, I play a spade and, if East follows, finesse the eight. If I lose to jack-ten-third, West is still endplayed.

More likely, if East chooses to win the diamond ace, he will shift to a heart. I must then decide who has the heart king. If it's East, I must finesse. If it's West, I must hop, cash the diamond king, and execute the same endplay. I can make that decision later. Let's see what happens on this trick.

I play the diamond queen. East ducks. Driving the club should work now. Unless, that is, East started with five diamonds. Then the defense will take three diamonds, three clubs, and a heart for down one. If that's the case, I need to abandon the club suit and go for the endplay.

Is that possible? If East started with five diamonds, he knows the diamond suit is dead. The natural play is to win and continue hearts. It would be quite diabolical to duck, assuming that would entice me to continue clubs. A robot, who assumes declarer is double-dummy, would never do that. And there are few humans who would be capable of it. So I'll stick with my plan.

I play a club. East wins and cashes two diamonds. Making one.


NORTH
Robot
♠ Q 9 6
7 3
Q 10 6
♣ K 10 9 7 2


WEST
Robot
♠ 10 3
K J 9 8 6 2
J 9 3
♣ 8 3


EAST
Robot
♠ J 5 4 2
4
A 7 5 2
♣ A Q J 5


SOUTH
Phillip
♠ A K 8 7
A Q 10 5
K 8 4
♣ 6 4

+90 is worth 79%. Almost every declarer played a heart at trick three. Some went down, some made it, depending on how they proceeded after that. A heart is certainly the intuitive play. But I think a little reflection will reveal that attacking clubs and retaining your flexibility is a better approach.

1 comment:

  1. I've seen numerous reactions to this post. I think it is important to appreciate that ChatGPT is a text generator, a sophisticated auto-complete and nothing more than that. It is an attempt to teach computers on some level what words "mean." if I ask it "What color socks am I wearing?" we want it to give an appropriate answer. We don't want it to answer "peppermint" or "seven notrump." We want it to understand what colors are, what socks are, and, ideally, what color socks are likely to be. We don't expect it to answer accurately, since it has no way of knowing what color socks I'm wearing. Similarly, when I ask it a question about bridge, we are looking for an appropriate answer, not an accurate one. It does not know how to play bridge and couldn't possibly tell you what you should do in a given situation any more than it could tell you what color socks I'm wearing. All it can do is understand what I'm asking and give an answer that makes sense linguistically, even if it is not logical. This is a non-trivial problem--one we have been working on for decades. I think ChatGPT does an impressive job--much, much better than any program could have done only a short time ago.
    What is disturbing is seeing people ask it things like, "Come up with a plan for balancing the budget in 10 years." It can certainly write a proposal that, linguistically, is a plan for balancing the budget. But anyone who expects it to come up with a plan that is viable doesn't understand what the program is doing

    ReplyDelete