Response to "John Lowenthal Story--A Follow-up"

Someone brought this post on BridgeWinners to my attention: 

John Lowenthal Story--A Follow-up

In the comments, someone expressed doubt about whether the story was true. That all depends on what you mean by "the story." The deal did occur, and the auction occurred as reported, and John duly made six notrump. That much is true. Anyone who is surprised by this obviously never knew John. The story as I reported it in "Remembering John Lowenthal" is, to the best of my memory, word for word the way John told it himself.

The deal reported in the New York Times, however, as well as the ensuing dialogue, is a construction. Years after the event, Alan Truscott heard the story and asked John if he could remember the deal. John could not. Alan thought it was too good a story to pass up, and he couldn't report the story in his column without an accompanying deal, so he made one up. To amuse himself, he constructed a layout where six notrump made but diamond contracts would fail, though that wasn't true in the actual deal.

There were many more Linda stories, sometimes with the full deal, sometimes not. I wish I had written them down. One I do remember is as follows:

John had observed they were losing a lot of tricks on defense and decided they needed some carding agreements. So, before the session, he said to Linda, "If I take a card from my hand, wave it around in the air and thumb it on the table like this, and if you look at that card and see that it's a high card--like an eight or a nine--then that means I want you to lead that suit." Linda said, "OK."

At some point in the session, John found himself on lead against three notrump. He led something, and declarer won and established dummy's suit. As luck would have it, Linda was winning the trick and John was showing out. It was clear that declarer had the rest of the tricks as soon as he gained the lead. It was also clear that the only tricks the defense might be able to cash were in the spade suit. So John, with jack-nine fourth of spades, took the spade nine out of his hand, waved it around in the air, and thumbed it on the table. 

The opponents were, of course, upset. But Linda was delighted. Her eyes lit up. For the first time all night, she knew what to do. She put the deuce of spades on the table and declarer played the ten. John couldn't believe it. All he was trying to do was to get her to cash her ace if she had it. He never imagined she would underlead it. He took his jack and returned a spade. Linda won her ace, looked around awhile, then shifted, and declarer claimed. The opponents had intended to call the director about John's antics. But when they saw Linda had ace-king-queen fourth of spades, they took pity on John and decided not to. John, meanwhile, was delighted with the result. They had held declarer to four notrump. In the past, he would have made six.

Finally, I do wish to say something about another comment in the BridgeWinners post. Someone said I left BridgeWinners because of a discussion of COVID-19 vaccines. Technically that's true. But more precisely, it was a discussion of COVID-19 vaccine mandates and my opposition to them, for which I was called an idiot. I realize to many, opposition to vaccine mandates is indistinguishable from opposition to vaccines. But I do think the distinction is important. And I wished we lived in a society where freedom of choice was sufficiently valued that the distinction was obvious.

 

Zenith Daylong - Oct 14, 2021 - Board 10

Board 10
Both vulnerable

♠ Q 6 3  ♥ 4 3  ♦ K Q 2  ♣ A J 10 5 4

RHO passes. I open with one club. LHO bids two diamonds, weak, and partner bids two hearts. I bid two notrump, and partner bids four hearts. Everyone passes, and RHO leads the ace of diamonds.

	NORTH Phillip ♠ Q 6 3 ♥ 4 3 ♦ K Q 2 ♣ A J 10 5 4
	SOUTH Robot ♠ K 8 5 2 ♥ A K Q 10 7 6 ♦ 7 3 ♣ 6

West	North	East	South
Robot	Phillip	Robot	Robot
Pass	1 ♣	2 ♦	2 ♥
Pass	2 NT	Pass	4 ♥
(All pass)

One would normally have a six-card suit for a two-level pre-empt. But since East has a jack-high suit and since the robots are not particularly aggressive with their pre-empts, I suspect he has seven of them. I play the deuce from dummy, and East plays the four. The card I play makes to difference to East. since West can have either one of them. Can my choice make a difference to West? Probably not. But since East would never play his second-lowest diamond, the three is the card I'm known to hold. So I play it on general principles.

West shifts to the seven of spades. (Robots don't play suit preference.) I play low from dummy. East wins with the ace, and I follow with the deuce. I have a diamond discard for my long spade. So after West takes his diamond ruff, I have the rest--barring something unexpected.

East returns the six of diamonds. West ruffs with the heart five and returns the nine of spades. East doesn't ruff it, and everyone follows to the first heart. So I claim.

	NORTH Phillip ♠ Q 6 3 ♥ 4 3 ♦ K Q 2 ♣ A J 10 5 4
WEST Robot ♠ J 9 7 4 ♥ J 8 5 2 ♦ A ♣ 9 8 3 2		EAST Robot ♠ A 10 ♥ 9 ♦ J 10 9 8 6 5 4 ♣ K Q 7
	SOUTH Robot ♠ K 8 5 2 ♥ A K Q 10 7 6 ♦ 7 3 ♣ 6

56% for plus 620. 

This isn't much of a deal from our perspective. It's far more interesting from the opponents' perspective. What would you play as East at trick one on partner's lead of the diamond ace? 

First of all, what does your card mean? When you are known to have at least six cards in the suit led, your card should be a three-way signal. If you play standard carding, the eight (your middle card) would ask for a diamond continuation and the jack or four would be suit preference. If you play upside-down attitude, it's better to play that your lowest card asks for a diamond continuation and high or middle is suit preference. This is less accident prone than playing middle encourages. To play that your lowest card sometimes asks for diamonds and sometimes for clubs is begging for an accident. You may think you can choose some criterion to make it unambiguous. But what's the point? Since there is no gain to switching methodologies, why risk it? If your lowest card is always encouraging, there can never be an ambiguity.

Why not do the same thing playing standard signals? Why not play high is always encouraging and middle or low is suit preference? For some reason, I don't know of any pair who plays that way. Perhaps anyone who is sufficiently concerned about disambiguation to adopt such a method is already playing upside down.

In any event, what switch should you request with this hand? It's hard to see beating this unless partner's diamond ace is a singleton. So let's assume it is. We have two aces and one ruff. Where is our fourth trick coming from? Given South's jump to four hearts, it seems unlikely partner has the spade king. Might we take a club trick as our setting trick? Declarer can pitch one club on a diamond, so he will need three clubs for us to have a club trick. That makes him 2623. If that's his pattern, does it matter whether partner switches to a club or a spade? A club switch beats it easily. What about a spade switch? I win, give partner a ruff with my lowest diamond, and now he switches to a club. That kills the dummy entry, so declarer can't take two discards. So it doesn't matter which suit partner leads at trick two.

Can we find a layout where partner's switch does matter? What if partner has a natural trump trick? Maybe then we can beat this even if declarer has a doubleton club. Let's try this layout:

♠ K J x  ♥ A K Q x x x  ♦ x x  ♣ x x

If partner switches to a spade and I give him a ruff, we are out of tricks. If he switches to a club, we get two aces, a club, and a trump trick. Perhaps there exists some deal were a spade shift beats it and a club shift doesn't. But the fact that I'm having a hard time constructing one suggests that, if there is one, it's a very specific layout. And there is nothing unusual about this layout. So I suspect it's right to ask for a club shift.

But the problem isn't over yet. Let's say you ask for a club shift and partner shifts to the club nine. (His card should be attitude: high to confirm he's ruffing a diamond and low to say he isn't.) Declarer hops with the ace and cashes the ace and king of hearts. You drop the jack of diamonds, suit preference for spades. Declarer now leads the heart ten and partner wins with the queen.

Now what? As long as declarer has another club, he's down. But what if he doesn't? What if he has

♠ K J x x  ♥ A K J 10 x x  ♦ x x  ♣ x?

Now partner must shift to a spade to get his ruff. Is that clear? Are you sure--and, more importantly, is your partner sure--that the diamond jack necessarily shows the spade ace? And what happens if partner does lead a spade to your ace? If declarer has the hand above, you must give partner his diamond ruff. But if he has

♠ K J  ♥ A K J 10 x x x  ♦ x x  ♣ x x,

you must cash a club. 

Partner could solve this problem for you by giving attitude with his spade return: high to say he can ruff a diamond return and low to say he can't, like his club play at trick two. But it's not clear he'll do that. He might not be on the same wavelength about what your problem is. He might be afraid that if he leads a low spade, you will play him for the king.

Fortunately, there is a more reliable way to solve both your problems. You should discard the king of clubs on this trick. This discard makes sense only if you have king-queen tight. So if partner began with three clubs, he knows another club is cashing; and if he began with four, he knows it isn't. In the former case, he will return a club, solving your cash-out problem. In the latter case, he has no choice but to play you for the spade ace. You will win his spade shift and, knowing the club can't be cashing since partner didn't lead one, will give him his diamond ruff.

This deal is a good example of the technique of "drawing a box around partner." If you can play your cards in such a way that partner is forced to do the right thing, do so. Also, if partner knows the layout and could have forced you to defend in a particular way (e.g., cashing your club for you above), assume he will. If he doesn't, then assume that what he could have forced you to do is not the right defense. This is a far more effective method of defense than relying on signals, since signals are always in danger of being misinterpreted.

Zenith Daylong - Oct 14, 2021 - Board 9

Board 9
Opponents vulnerable

♠ K J 4  ♥ J 6 2  ♦ Q 7 3  ♣ 10 7 5 4

Partner opens with one spade. I raise to two spades. Partner goes on to four, and West leads the heart ten.

	NORTH Phillip ♠ K J 4 ♥ J 6 2 ♦ Q 7 3 ♣ 10 7 5 4
	SOUTH Robot ♠ Q 10 7 6 2 ♥ A Q 9 5 ♦ A K 4 ♣ 3

West	North	East	South
Robot	Phillip	Robot	Robot
			1 ♠
Pass	2 ♠	Pass	4 ♠
(All pass)

That's quite an aggressive four-spade bid! A short-suit try in clubs looks more to the point. But with all my high cards outside clubs, I would have accepted despite my minimum. So no harm done.

Assuming the lead is from shortness, I can take four heart tricks if I guess the suit. I start by covering with the jack and taking the king with my ace. If the lead is from stiff 10, 104, or 103, I can take a double finesse against the 87. If the lead is from 108 or 107, I need to cash the queen then get to dummy to finesse against the remaining spot. 

The double finesse works against more layouts, but it's hard to execute without running into a a heart ruff. So my initial plan will be to draw trump, cash the heart queen, hoping to drop the eight or seven, then reach dummy with the diamond queen for a finesse.

I cover with the jack--king--ace. Now six of spades (retaining the deuce on principle)--three--jack--nine. I play the spade king from dummy, and East pitches the diamond deuce. The spade deuce doesn't seem important anymore--if it ever was--so I play it. 

West takes the ace and shifts to the four of hearts. I don't think that was his best play. I wasn't picking up the heart suit without his help. I cover with the six, East covers with the eight, and I win with the queen. I draw trump and finesse the five of hearts to make five.

	NORTH Phillip ♠ K J 4 ♥ J 6 2 ♦ Q 7 3 ♣ 10 7 5 4
WEST Robot ♠ A 8 5 3 ♥ 10 4 ♦ 9 6 ♣ A Q 9 6 2		EAST Robot ♠ 9 ♥ K 8 7 3 ♦ J 10 8 5 2 ♣ K J 8
	SOUTH Robot ♠ Q 10 7 6 2 ♥ A Q 9 5 ♦ A K 4 ♣ 3

450 is worth 77%. Obviously West can hold me to four by not playing the second round of hearts, but the defense went the same way at every table. Luckily for me, not all declarers took advantage. The five was apparently too lowly a card to merit their attention. But the overtrick wasn't the only reason this was a good result. Quite a few players responded one notrump, corrected two hearts to two spades, and played it there.

This is a ten-loser hand with no aces, so I can see how some might judge it isn't worth a single raise. But the jack of spades is a big card. And I don't share the common view that 4333 is necessarily a bad pattern. It can be, but a lot depends on the pattern it faces. Opposite a five-five, for example, a 4333 is actually a plus--not a minus. Doubletons are useful when they are opposite partner's three-card suits. But holding three cards in partner's side five-card suit can be better than holding a doubleton, since it makes it easier for him to establish his long cards. And we don't need to go to such extremes. Even opposite a 4432, a 4333 can be better than a 4432 with matching doubletons.

Perhaps 4333 is a net minus if we know nothing about partner's shape. But, even so, it's not as much of a minus as most people think or as loser count suggests. Evaluating 4333 as a full loser more than 4432 is too pessimistic. So, in my judgment, this hand is a sound minimum raise. And the fact that four spades is a fine contract even though partner overbid with his four-spade call is a testament to that.

Zenith Daylong - Oct 14, 2021 - Board 8

Board 8
Neither vulnerable

♠ A K 6 3  ♥ 9 4  ♦ K Q 9 5  ♣ Q 9 7

Partner opens with one notrump. With 29 HCP, there is often no advantage to playing in a four-four major-suit fit. The cards you can ruff may be winners anyway. So I'm going to raise to three notrump without looking for the four-four spade fit. This has two ways to win: it might be the right contract when partner has four spades, and it might make the defense harder when he doesn't, since the opponents will have less information about declarer's shape. 

I bid three notrump. Everyone passes, and West leads the heart deuce.

	NORTH Phillip ♠ A K 6 3 ♥ 9 4 ♦ K Q 9 5 ♣ Q 9 7
	SOUTH Robot ♠ 8 7 ♥ A K Q 7 ♦ A 10 3 2 ♣ K 6 5

West	North	East	South
Robot	Phillip	Robot	Robot
		Pass	1 NT
Pass	3 NT	(All pass)

I have five tricks in the majors. If I have four diamond tricks, that makes nine--ten after knocking out the club ace. What are my chances for eleven? One possibility is to take four heart tricks. I have two ways to attempt to do that. If West has led from jack-ten, I can hop with dummy's nine and win the trick. If he has led from eight third, I can play low, forcing an honor, then finesse against the other honor. 

What are my chances for an eleventh trick if I don't have four heart tricks? If either hand has four hearts and the club ace, I have a strip squeeze. I lead a club through the ace, then cash four diamonds, coming down to seven cards. The targeted hand has to keep two clubs (else I can duck out the ace) and three hearts (else my seven is good), so he must come down to a doubleton spade. Now I cash the ace and king of spades and toss him in with a heart to score my other club honor. 

That possibility solves my trick-one problem. If I play the nine and it's covered, I can go for the endplay. If I play low, I may lose that opportunity. Say East plays an honor at trick one and my later finesse loses. Now I no longer have a heart loser to throw an opponent in with. So playing the nine at trick one gives me two chances while playing low gives me only one.

I play the nine, and East covers with the jack, in principle denying the ten. It's usually right to conceal the card immediately above the one third hand plays, since that is a card that either opponent might hold from his partner's perspective. So I win with the king.

Which hand should I target for the endplay? The robots like leading from three-card suits against notrump, but ten third isn't especially appealing. Given West has the heart ten, I think it's likely he has the four-card suit, so I'll try leading clubs through him. I play the club five--ten--queen--three. He wouldn't insert the ten gratuitously. So either he has the jack also or he has ace-ten doubleton.

Now I can run the diamonds and squeeze West. All I have to do is decide which opponent is more apt to have a stiff diamond, so I can cash my honors in the right order to pick up jack fourth. As a general rule, when someone leads a four-card suit against notrump and you need to guard against a side-suit singleton, it's right to guard against the singleton on your right. That's because the only way the opening leader can have it is if he is 4-4-1-4 or if he has a five-card suit he chose not to lead. Against traditional opening leaders, playing East for the singleton diamond is clear. Robots are less inclined to lead their longest suit, so that changes the odds somewhat. But I doubt it changes them enough actually to reverse the normal play. Accordingly, I cash the diamond king--eight--deuce--four. Now the five of diamonds--seven--ace. West pitches the eight of clubs. 

Oops. And things were going so well. Eleven tricks is now out of the question. I still have nine tricks. Is there any way to find a tenth? I need to start by ducking a spade--both to correct the count for a possible squeeze and to isolate the spade stopper if spades are four-three. If East wins the spade and plays a club, I won't be happy. But why should he do that? Clubs is the suit I was attacking. If I duck a spade to him, he will probably continue hearts, his partner's suit.

I play the spade seven--four--three--nine. East, as expected, shifts to the five of hearts. I win with the ace as West plays the three. This is the current position:

	NORTH Phillip ♠ A K 6 ♥ -- ♦ Q 9 ♣ 9 7
	SOUTH Robot ♠ 8 ♥ Q 7 ♦ 10 3 ♣ K 6

To take another trick, I need to cash my winners and throw West in to lead away from his ace of clubs. Unfortunately, I don't have the communication to throw him in in hearts. I started out hoping West began with four hearts. Now I have to hope he began with three--along with at least four spades. I can cash the queen of hearts, then lead to dummy's queen of diamonds. West will be down to five black cards. If he has two spades, dummy's spades are good. If he has three spades, I can toss him in with a spade to lead away from his club ace. And if he has four spades, the club ace is singleton, so I can duck a club.

I cash the heart queen. I know West has the heart ten left. If I don't see it on this trick, I have no chance left. I do. West follows with the ten. I pitch a diamond from dummy and East follows with the heart six. I play a diamond to dummy's queen, and West discards the club eight. 

Dummy's club nine is a fortunate card. It prevents West from dropping the club jack from ace-jack third, pretending to stiff his ace. I know he has ace-jack of clubs left. If he has the long spade, I have him. I cash the ace and king of spades and lead a third one. Unfortunately, it's East who wins the trick, as West pitches the heart eight. He was just toying with me when he dropped the ten. East cashes his diamond jack and leads a club to his partner's ace. Making three. 

	NORTH Phillip ♠ A K 6 3 ♥ 9 4 ♦ K Q 9 5 ♣ Q 9 7
WEST Robot ♠ J 5 4 ♥ 10 8 3 2 ♦ 4 ♣ A J 10 8 4		EAST Robot ♠ Q 10 9 2 ♥ J 6 5 ♦ J 8 7 6 ♣ 3 2
	SOUTH Robot ♠ 8 7 ♥ A K Q 7 ♦ A 10 3 2 ♣ K 6 5

9%. My three notrump bid backfired. If I bid two clubs and let partner show his four-card heart suit, it steers West off the heart lead. He chooses the club jack instead. Now I can make four--or five if I pick up the diamonds--without the bother of an endplay.

But the primary blame for this result goes to my play of the diamond suit. If I play diamonds the other way and make five, I get 90%. As I think about it more, not only am I pretty sure I made the wrong play against robots, but I'm beginning to believe my statement, "Against traditional opening leaders, playing East for the singleton is clear" was wrong as well. One thing I didn't consider at the time was that I know something about West's club honors. I know he has specifically ace-ten doubleton of clubs or some holding headed by ace-jack-ten.

For the sake of argument, let's take it as given that West won't have a five-card suit lurking. Let's also assume East doesn't have six spades. (He might have opened a weak two-bid if he did. And a six-one spade break is unlikely even without that consideration.) So West's possible patterns are 4414, 2443, or 3442.  If those patterns were equally likely, the odds would be two to one in favor of playing East for the stiff diamond.

Often similar patterns are close enough to being equally likely that assuming they are is sufficient for a rough, at-the-table calculation. But this time our constraints in the club suit make a big difference. There is only one way for West to hold a doubleton club or a tripleton club. But, a priori, there are four ways for him to hold four clubs, since he can hold any one of four small cards along with his ace-jack-ten. So 4414 is roughly four times as likely as either 2443 or 3442, making the odds roughly two to one in favor of West's having the stiff diamond.

We could make a more accurate calculation by taking into account the spade distribution. But since it isn't close, there is no need to do so. Even against traditional opening leaders, West is the overwhelming favorite to hold a stiff diamond. And against robots, who might easily have a five-card black suit, he is an even bigger favorite. I played by rote and didn't stop to consider that the general rule might not fit this specific problem. It was a costly error.

At least I judged correctly to risk my contract by ducking a spade. East could now beat me by shifting to a club. But going down isn't much worse than making only three. Having misplayed diamonds, I had to go all out to try to find another trick

Zenith Daylong - Oct 14, 2021 - Board 7

Board 7
Both vulnerable

♠ 3  ♥ A J 4 2  ♦ A 9 8 4 3  ♣ K 3 2

I open with one diamond in first seat. LHO overcalls with two clubs. Partner and RHO pass. One tries to avoid defending low-level contracts at matchpoints. But when you have shortness in an unbid suit, it's often right to make an exception. Either the opponents are in the wrong strain, or partner has the suit and the hand is a misfit. As Lowenthal liked to put it, "If you balance with this hand, two bad things can happen: (1) Partner has spades. (2) Partner doesn't have spades." 

I pass, and partner leads the diamond seven.

	NORTH Robot ♠ A Q 8 ♥ 9 8 7 6 ♦ Q J 6 5 2 ♣ 8
		EAST Phillip ♠ 3 ♥ A J 4 2 ♦ A 9 8 4 3 ♣ K 3 2

West	North	East	South
Robot	Robot	Phillip	Robot
		1 ♦	2 ♣
(All pass)

Declarer plays a sneaky five from dummy, perhaps hoping we'll forget what spots were played and get our signals mixed up. The robots love to lead low from doubletons in the middle of a hand, but I haven't seen them do that on opening lead, so I'll assume declarer has king-ten doubleton. How will the play go? I'll win with the ace and return a low diamond (suit preference) for partner to ruff. Partner will play a heart to my ace, and I'll play a third diamond for a possible uppercut.

But what if declarer has a heart loser to pitch? If partner has the heart king, we need to cash however many hearts we can before playing the third diamond. If we try to cash too many hearts, however, we lose the second diamond ruff. If partner has king third of hearts, that's not a problem. Partner knows declarer can't have singleton heart, since that would give me five of them. So, after taking his diamond ruff, he will play king and another heart. But what if partner has king fourth of hearts? Now he doesn't know whether declarer has two hearts, in which case he must play king and another, or a singleton, in which case he must play low to my ace.

I know it's right for partner to return a low heart from king fourth. Is there any way I can get him to do that? I don't have a lot of options. I must return a low diamond; my only choice is whether to return the three or the four. Will my choice make a difference to partner? (I realize this question is rhetorical. My robot partner doesn't play suit preference period, so my card will make no difference. But I might as well try to solve the problem anyway, just for practice.)

I should point out that if I do return the four, it's a falsecard--not a nuanced signal. Partner should always assume I am signaling as loudly as possible. The four in principle denies the three, so partner should assume that declarer has it. But that actually solves his problem, doesn't it? If he thinks declarer has another diamond, there is no immediate uppercut. So there is no reason for him to lead the king from king fourth. The falsecard tricks partner into doing the right thing.

But hold on. It also tricks him into doing the wrong thing with king third. If declarer has a third diamond, there is no reason for him to lead the king from king third either. If I knew partner had four hearts, returning the diamond four would be a good play, and it is worth remembering for a different occasion. But I don't want to take control when I'm not sure about the layout. If I play an honest card, maybe partner can work out the position himself from whatever clues he has available.

I win with the ace, and declarer drops the king. I return the diamond three. Partner ruffs declarer's ten with the club six and returns the heart three to my ace. Declarer plays the queen. Is there any chance that's a foolish falsecard from queen doubleton and I can cash the jack now? No. I already decided partner would lead the king from king third. This is either a stiff queen or king-queen tight. 

I return diamond eight. Declarer ruffs with the seven, and partner overruffs with the ten. Partner tries to cash the heart king, and declarer ruffs with the four. We can't take any more tricks beyond my club king. Making two.

	NORTH Robot ♠ A Q 8 ♥ 9 8 7 6 ♦ Q J 6 5 2 ♣ 8
WEST Robot ♠ 10 7 5 4 2 ♥ K 10 5 3 ♦ 7 ♣ J 10 6		EAST Phillip ♠ 3 ♥ A J 4 2 ♦ A 9 8 4 3 ♣ K 3 2
	SOUTH Robot ♠ K J 9 6 ♥ Q ♦ K 10 ♣ A Q 9 7 5 4

Minus 90 is worth 70%. A surprising number of defenders shifted to a spade trick two, allowing declarer to make two overtricks. That's so narcissistic! Going after your own spade ruffs instead of giving partner his diamond ruffs.

Partner did have king fourth of hearts and solved his problem. How might he have solved it? Let's look at the problem from his perspective.

It seems unlikely I have more than one natural trump trick. One trump trick, the diamond ace, and two ruffs makes four tricks. We need two hearts to cash in order to beat it. So returning a low heart gives up on defeating the contract; it's playing to stop an overtrick. At IMPs, therefore, you should play the king. But at matchpoints, it's not so clear. Since we are vulnerable, some pairs may be buying the contract our way and going minus 100. So the difference between minus 90 and minus 110 may be greater than the difference between minus 90 and plus 50. If it's a 50-50 guess, playing to stop the overtrick is probably the percentage play.

Is it a 50-50 guess? Or is there some reason to think either a singleton or a doubleton heart is more likely? I can't think of a reason. But there is anoher argument for leading low. If declarer has a singleton heart and you lead the king, you have dropped a trick. But if declarer has a doubleton and you lead low, you still have a chance to survive. If partner happens to have ace-queen third, he can cash the ace and queen before leading the third diamond. 

Should he? I believe he should. If you were sure returning a diamond was correct, you would lead a high spot (attitude), not a low one. Leading low shows the heart king and shows that you are not so sure about that immediate diamond return. So if partner trusts you, he should do the right thing. He should cash two hearts with ace-queen third (but not with ace-queen fourth, since he knows you would have led the king with only three hearts).

What if you have king fifth of hearts? I believe you should lead a high spot to prevent partner from trying to cash the queen. Leading low is right only if partner has specifically ace-queen tight. You simply have to give up on that layout.

.