Why it feels like you’re always the one who reaches out

You remember every message you sent, the ones you drafted and deleted, the afternoon you nearly rang. Of your friend’s side you have only what arrived. The research on how people misjudge their standing with others points at exactly this asymmetry: you hear your own self-criticism and none of theirs. The same goes for effort. Their nearly-messaged-you shows up nowhere you can see.

The liking gap is the clearest measurement of the distortion. Across several studies, with strangers paired in a lab, dorm-mates followed over an academic year and adults in workshops, people consistently underestimated how much their conversation partners liked them. A misjudgement on first meeting would be unremarkable; this one survived months of living alongside someone. The samples were mostly students and workshop groups, liking was self-reported on both sides, and the size of the gap varied between studies. The direction did not.

A second set of experiments looked at the moment of reaching out. People who sent an unprompted message predicted how the recipient would feel, and the recipients were then asked. They appreciated it more than the sender had guessed, and the gap widened the more surprising the contact was. These are experiments about single moments, not friendships over years. Still, the entries you log as "me again, unanswered" are valued at the other end more than your ledger allows.

Suppose the count were accurate. There is still the question of what keeping it does. Clark and Mills spent a research programme separating two kinds of relationship by the norm they run on. In exchange relationships, a benefit is given to incur or repay an obligation, and members keep track of individual inputs. In communal relationships, benefits follow from concern for the other person’s welfare, and nobody is counting.

Tracking who initiated is diagnostic of the first kind. Where closeness is wanted, promptly repaying a favour reduced attraction rather than increasing it: the ledger behaviour itself signals the wrong relationship type. Most of this comes from laboratory studies of short interactions, and real friendships mix the two norms more than the theory does. But once "I always text first" is a tally you consult, the friendship is in the frame where tallies belong, and it feels like an exchange because you are treating it as one.

A related finding comes from a different field. In six experiments, people did an ordinarily enjoyable activity, walking or reading, with their output measured or not. Measurement increased how much they did and decreased how much they enjoyed it, and the drop outlasted the measuring: afterwards they read less, and in one experiment avoided reading altogether. Those experiments were about walking and reading, not friendship, and the extrapolation is ours. A friendship you keep score of starts to feel like work.

The usual next move is the test. Stop initiating, say nothing, and see how long it takes them to notice. It feels like the only way to get the other side’s data.

The evidence on how friendships decay says it is a test most friendships fail, and not because of what you are trying to measure. An 18-month study tracking closeness and contact across the people in someone’s life found that closeness to friends fell steadily once regular contact stopped, while closeness to family held up over the same gaps. The design is correlational and the two almost certainly feed each other. It cannot say that silence causes decay. It can say they arrive together.

So the silence does not isolate the variable you care about. It removes the thing the friendship was partly made of and then measures what is left. If they do not reach out, you have learned that a friendship without contact drifts, which was already known, and paid for the lesson with the friendship. If they do, you have learned something you could have found out by asking.

Whether your friend values you. All of the above is about averages: the average sender underestimates, the average ledger is lopsided, the average silence is followed by drift. None of it describes one person, and some friendships genuinely are one-sided. A friend who never asks about you, cancels and does not rebook, and is warm only when it suits them is not a prediction error.

What the research supports is narrower. The feeling that you are the only one trying is produced, in part, by the fact that your own trying is the only trying you can see, and the specific belief that your messages are tolerated rather than welcomed is, on average, wrong. The count is real. Its meaning is less certain than it feels.

The honest test is a direct sentence, not silence. Saying you have noticed you are usually the one who gets in touch, and asking whether that is just the shape of their life right now, is a deeper conversation than most people want to start. In experiments on deep conversation, people found them less awkward and more connecting than they predicted. Those were assigned conversations, often between strangers in a study, so the stakes were lower than yours. It is still better evidence than a month of saying nothing.

Stop counting, deliberately. Not because the count is wrong, but because keeping it is the behaviour that turns a friendship into a ledger, and because it is built from one side’s entries anyway. When you catch yourself tallying, that is the moment to put the tally down.

Replace it with what the other side can actually register. The relationship literature’s working definition of intimacy is feeling understood, validated and cared for, and the operative word is perceived: what moves a relationship is what the other person registered, not what you did. That framework rests on mostly correlational work, so it describes what goes together, not what causes what. Even so, a message that carries a specific thing they told you, the interview, the scan, the flat, is attention made visible, and it reads as warmth rather than as another entry on the ledger. If you want to know whether they value you, ask them in those words.

None of this needs an app. A line on paper with their name and the thing they mentioned, and the decision to send it rather than wait and see, is the whole method. What does not work is the silent experiment, because the one result it reliably produces is the one you were afraid of.