The optimal stopping problem is about deciding when to stop and act. Facing a choice, we can wait in order to gather more information. But waiting has a cost. Given more time we might make a better judgment, yet at the same time the opportunity itself might disappear.

It’s true when buying a house, when weighing a job change, when investing. Wait longer and you learn more. But wait too long and the good listing is sold, the good position is filled, the good timing has passed. The heart of the optimal stopping problem is, in the end, this balance. Keep looking, or decide now.

A confession seems like a matter of feeling, but seen through the lens of decision-making it is simple. Speak now, or wait a little longer.

Confess too early and you lack information. It’s hard to judge what the other person thinks of you, how far the relationship has actually come, whether you’ve mistaken simple kindness for affection. In this case a confession is a kind of early investment. There’s potential upside, but the data is thin.

Confess too late and the information piles up. You get to observe more — their tastes, their reactions, their tone, how often they reach out, all the small signals. The problem is that the market shifts in the meantime. They might start seeing someone else, their feelings toward you might harden into friendship, and above all that vague but powerful variable called timing might vanish.

So a confession can be seen as an optimal stopping problem: gather more information, or act now? The classic example of an optimal stopping problem is the secretary problem, which I covered before.

The catch is that waiting, too, has a cost. Wait one more day and the information grows a little. But at the same time an opportunity cost accrues. The relationship might settle into a stable friendship; the other person might read your restraint as indifference. Or your courage simply evaporates. Human resolve is a more volatile resource than we tend to think.

So if the probability of success has risen high enough, and the risk of missing the timing has grown larger than the value of the information you’d gain by waiting more, then it is reasonable to act.

Of course, no one can say how much “enough” is. Confessions come with no probability table. No rule like “confess once the other person shows affection above a certain threshold” exists in real life. If it did, everyone would date with a spreadsheet open. The problem is that feelings aren’t quantifiable — and the bigger problem is that we keep trying to quantify them anyway.

Even so, this way of thinking has its uses. A relationship is not completed by observation alone. At some point the observer has to intervene in the experiment. A confession is less a problem of finding the optimal solution and more a problem of noticing the moment when further delay is no longer rational. When there have been enough signals, when the cost of silence is mounting, and when waiting has begun to be avoidance rather than prudence — that is when you should stop.

In other words, from the standpoint of the optimal stopping problem, a confession is a declaration that information-gathering has ended. So how about confessing like this from now on?

“Based on the observations so far, I’ve judged that the probability you like me too is sufficiently high.”

A truly romantic confession.