Can a good time end a good thing?

Thursday nights

Someone suggests something new to do together. A friend suggests a Thursday dance class instead of your usual dinner. A colleague suggests a weekly working session instead of the usual check-in. A partner suggests a new hobby. The paper calls the person who suggests it the proposer, and the person who decides whether to join the responder. Here the responder is you.

Take the friend and the class. Some Thursdays you want to dance. On others you would rather have dinner, and you go anyway, because your friend wants to. Your friend cannot tell which Thursday is which, so every yes can mean “I came for you.” You like that it can.

Then comes a Thursday when you laugh through the whole class and hum the music all the way home. Your friend sees how much you enjoyed it, and you know they saw.

After that you start saying no, on the Thursdays you would rather have dinner and on the ones you would enjoy too. Nobody argues, and the class is as good as ever.

We usually read an ending like this as a verdict: the class was not worth it. Here it ended because you enjoyed it, and your friend could tell.

Why a good night ends it

You enjoy the class, your friend enjoys it, and now your friend can see that you do. Learning what someone enjoys is supposed to help two people do more of it together. Here it ends the class, and nobody has made a mistake.

You go on Thursdays you would rather have dinner because you care about your friend’s evening. While your friend cannot tell which Thursdays those are, every yes can mean “I came for you.”

Once your friend can see that you enjoy the class, a yes starts to look like something you would do anyway, and it gets taken for granted, even on the Thursdays you really do want to go. It can sting to have a yes taken for granted, whether by a friend, a partner or a colleague. Your usual dinner carries no such risk: it is still yours together.

Your yes on the Thursdays you wanted to go depended on your yes on the Thursdays you did not.

So the class can end while it is still worth having. Going on both kinds of Thursday would still leave your friend better off, and you better off on average. It stops because it can no longer be held together, not because it stopped being worth it.

Every stop comes too soon

When a class stops, it seems to show that it could not work for that pair. Watch a few dozen pairs like you and your friend, one tube each. The level in a tube is how sure the proposer is that the responder wants to go; each class leaves a small clue that moves it, and above the line the class stops and the tube closes. The tubes are split by something neither of them knows at the start: whether the responder usually wants to dance or usually prefers dinner. As with anything new, you know how you feel tonight long before you know whether it is for you, and your friend has only the clues.

Look at the pairs whose responder usually prefers dinner: some of their classes stopped too. If their friends knew that, a yes could again mean “I came for you,” and the class could go on. Nobody misjudged anything, and every clue was read correctly. A few clues that pointed to enjoyment carried the guess over the line before these pairs could be told apart from the rest, and once a class stops, no more clues come.

Every stop comes too soon, and that is the paper’s main result. When the two decide one Thursday at a time, each class reveals only a little, and moods carry little from one week to the next, the class stops with the guess only a little past the line, never far enough for stopping to help either of them. Every Thursday lost was still worth having for both on average, so both are worse off, on average, than if the class had told them nothing.

Some classes that stop could go on, if your friend knew what you usually prefer.

What brings it back

After a stop, your friend’s view of how often you want to go stays where it was. If the trouble is that your friend cannot tell what you usually prefer, the obvious fix is to make each class show more clearly how you felt. What would that do?

Clearer clues can end more classes, not fewer. Each clue that points to enjoyment now moves the guess further, so fewer of them carry it over the line, even for responders who usually prefer dinner. Clues that point the other way cannot undo this: they can keep a class going, but they never reach one that has stopped.

A stop that waiting cannot end needs the opposite of what caused it: a sign that you do not always want to go. Only a clue that arrives without the class can bring one. A song you both know comes on at dinner and you do not feel like dancing; your friend sees it, and on Thursday a yes can again mean “I came for you.” Clues like this can bring some pairs back, sometimes only for a while.

In simulations guided by couples’ daily diaries, stops that waiting alone cannot end account for at least 95 percent of the opportunities lost over the first twelve weeks. In the paper’s main simulation, the same weak clues arriving outside the class cut the pair’s combined loss over those weeks from 14.8 to 8.4 percent of what they would get with clues of neither kind.

A good night can end the class. A night you would rather not dance, if your friend sees it, can bring it back.

What to take with you