Am I boring to talk to?
Probably you can't tell. That is the finding. Of everyone who was at your last conversation, you are the worst placed to judge how it went.
The liking gap
Strangers talked in pairs. Afterwards each was asked in private how much they liked the other, and how much they thought the other liked them. People rated their partner higher than they believed their partner had rated them. Both did. Both cannot be right.
Among students rating flatmates they had just moved in with, the gap was still there months later. It is not first-meeting nerves that wear off.
People might simply be modest, so trained observers were shown recordings of the same conversations. They could tell how much the two had liked each other. The signal was sent. Everyone picked it up except the person it was aimed at.
Run over text chat, the gap was still there. It shows only that typing did not remove it, not that it is the same size — and we read that paper's abstract rather than the paper.
What this does not say
It does not say you are interesting. It is a pattern across groups, not a verdict on you. Some conversations do go badly, and the person who thought so was right.
It also only covers conversations that happened. If what you are reading as a verdict is silence, none of it applies, because there was no conversation to misjudge. That is a different problem with a duller explanation.
What to do with it
Nothing during the conversation. There is no technique here.
The use of it is afterwards. If you replay conversations and reach a verdict, you now know something about the instrument producing it. It reads low, and observers watching the same conversation do not reproduce it. That is not a reason to believe the opposite. It is a reason to stop treating the post-mortem as information.
What that verdict usually causes is not starting the next one, which guarantees it never gets corrected.
Where this comes from
After a conversation, people underestimate how much the other person liked them — and an outsider watching the recording reads the room more accurately than the person who was in it.
What the study actually did
Five studies. Strangers paired in a lab; trained coders watching the recordings of those same conversations; conversations of varying lengths; 100 adults at public "How to Talk to Strangers" workshops in the UK; and 102 first-year students rating the suite mates they were living with, five times across an academic year. The gap turned up in all of them and lasted months among the students. The coders could predict how much people actually liked each other — so the signals were there to be read, and the person in the conversation was the one missing them.
What it doesn't show. That you specifically are liked. This is a bias in one direction across a group of people, not a verdict on any single conversation — some people read a bad conversation correctly. Shyer participants showed a bigger gap, but that is one moderation test in the smallest of the five studies, on a measure the researchers only started collecting partway through, so treat it as a hint rather than a finding. Self-esteem, rejection sensitivity and narcissism made no difference to the size of the gap. All of it was face to face, and the samples were university students and people who had signed up for a workshop about talking to strangers. Whether it happens over text is the next claim rather than this one.
Boothby, E. J., Cooney, G., Sandstrom, G. M., & Clark, M. S. (2018). The liking gap in conversations: Do people like us more than we think? Psychological Science, 29(11), 1742–1756 ·
10.1177/0956797618783714
It is not an artefact of being in the room together. The same gap turned up among people who met over text chat.
What the study actually did
Two experiments in which strangers were paired over text chat, audio or video and then moved to video chat — working through a shape-matching task in the first, talking about films and television in the second. People underestimated how much their partner liked them in every condition, including the ones that started in text, and which medium they had started in did not change that.
What it doesn't show. That this is what happens here. Everyone ended up on video, so nobody stayed unseen, and the pairs were assigned by a study rather than chosen off a roster. It also does not say the gap is the same size in text as in person — what was found is that the medium did not make it go away.
Oviedo, V. Y., Guydish, A. J., & Fox Tree, J. E. (2025). The liking gap online: People like us more than we think. Computers in Human Behavior Reports, 17, 100582 ·
10.1016/j.chbr.2024.100582
Read from the abstract. We have not read this paper in full, so what is above is what its own summary says and no more. A paper states its sharpest limits in its discussion, and we have not seen them.
