You're in the car, or standing in the kitchen, or lying in bed. The conversation ended hours ago, and you're still inside it. You said too much about your job. You laughed at the wrong moment. There was that pause, the one you're certain meant something. By midnight you've reached a verdict: they were being polite, and they won't be eager to see you again.
Here is the uncomfortable part. You're probably wrong, and wrong in a predictable direction.
The gap between how it felt and how it landed
In 2018, psychologist Erica Boothby and colleagues published a set of studies in Psychological Science on what they named the liking gap. Pairs of strangers were put in a room and asked to talk. Afterward, each person rated how much they liked their partner. They also guessed how much their partner liked them.
The pattern was consistent. People underestimated how much their partners liked them. The partner's rating was reliably higher than the guess.
The finding didn't stay in the lab. The researchers also studied people in workshops and first-year students living in the same residences. In the dorm study, the gap showed up and persisted for months. People who spent time together kept assuming they were liked less than they were.
This isn't a story about how you'd feel if you were more confident. It's about a measurable mismatch between what you believe happened and what the other person reported.
Why your memory of a conversation is rigged
The researchers pointed to a simple asymmetry. You have access to your own mind during a conversation, and the other person doesn't. You can hear your internal editor in real time: that came out wrong, why did I say that, they look bored. You're judging yourself against an ideal of how the conversation should have gone.
The other person sees none of that. They get the finished product, which is a person who smiled, asked a question and stayed. They're also busy with their own inner commentary. The study's authors found that people who were more critical of their own conversation, and who focused on their flaws, tended to show a bigger gap.
So the conversation is replayed in your memory with a harsh narrator and no access to the other side. The result feels like analysis, but it isn't. It's a one-sided transcript that you're treating as a verdict.
This is also why the replay at midnight never seems to land anywhere new. You're reviewing the same biased footage again and again.
What the gap costs you
The cost isn't just a bad evening. If you believe people lukewarmly tolerate you, you act on the belief.
You don't send the follow-up text. You decline the second coffee, because why impose on someone who was only being nice. You hold back in the next conversation to avoid repeating the first one. The other person notices a person who seems distant, and the story you were afraid of starts to come true.
A mistaken belief that sits on a friendship is quiet and slow. It's the reason some people go years believing they're harder to like than they are.
Replace the verdict with evidence
The fix isn't to tell yourself you were great. You wouldn't believe it, and it would be just as unfounded as the harsh version. The fix is to give your memory a second witness: what the other person actually did.
Behavior is a better source than your inner feeling. People who are bored leave early, check their phones and give short answers. People who like talking to you stay, ask follow-up questions and bring up things from earlier in the conversation. Those are facts you can write down, and they don't care how you felt.
This is the Evidence Page. It takes about five minutes, and it works best the same day, before the replay hardens into a story.
It has three parts:
- The verdict. One sentence on what your mind is telling you. "I think I talked too much and she was bored." Writing it plainly takes some of the power out of it.
- The evidence, for and against. List what the other person did. Not what you think they felt. "She asked about my sister. She stayed 40 minutes past when she said she had to go. She laughed when I told the story about the ferry. She looked at her phone once." Include the bad evidence honestly. This isn't a pep talk.
- The next small act. Based on the evidence, what would you do if the more generous reading were true? Often it's just one text.
The habit matters more than any single page. After a few weeks, you can reread old pages and check your verdicts against what happened next. Did she reply? Did they invite you again? Your own record becomes a calibration tool. You'll see how often the midnight verdict was wrong, and you'll see it in your own handwriting.
One caution
The liking gap is an average. It doesn't mean every person you talk to loves you, and it doesn't mean your worries are always unfounded. Some conversations really do go badly, and the evidence will say so. That's useful too. A page that says "Short answers, no questions back, left quickly" is a more actionable thing to learn than a vague dread.
The point isn't optimism. The point is accuracy, which your self-judging mind isn't providing.
Your next moves
- Pick one conversation from the last week that you're still replaying. Write the verdict in one sentence, then list three things the other person actually did or said. If you can't find any evidence for your verdict, notice that.
- Write the page tonight, not next week. The replay hardens fast, and details like who asked what fade within a day or two.
- Separate behavior from interpretation. Use two columns or two lines. Did: "asked me a follow-up question." Thought: "she was just being polite." Only the first column counts as evidence.
- Send the follow-up you were going to skip. If the evidence is mixed or good, send the text, even a small one: "Enjoyed hearing about the ferry story. Coffee again soon?"
- Keep a running tally for a month. At the bottom of each page, add one line later: what happened next. After four or five entries, check how often your verdict matched reality.
Where a daily page fits
A habit of writing the day down gives you what a midnight replay never will: the facts, recorded while they were fresh, in a form you can look at later. Inkdays is built around that kind of page, one a day, written by hand-sized thoughts in ink. If a conversation is sitting in your chest tonight, it's a natural place to put what the other person actually said and did, and to find out, a few weeks from now, how much kinder the record is than the replay. You can see how it works at inkdays.lumenlabs.works.