Why do I keep rereading our messages?

Because the thing you are looking for is not in there. In a 2005 study published in the Journal of Personality and Social Psychology (Kruger, Epley, Parker & Ng), people wrote messages that were either serious or sarcastic and predicted how well they would land. Senders expected 78% to be read correctly. Readers actually got 56% — near a coin flip. Spoken aloud, the same lines were read correctly 73% of the time. The tone lives in the voice. Rereading cannot add it back.

Because there are two ways to go over something, and they do not do the same thing. Researchers separate brooding — going back over a thing and comparing it to how it should have gone — from reflection, which is going back to work out what actually happened. In the study that first split them apart (Treynor, Gonzalez & Nolen-Hoeksema, 2003), brooding predicted worse mood over time. Reflection did not. Rereading can be either one. The difference is whether you are gathering information or grading yourself.

Usually one of four things. Proof you did not ruin it. Proof they were already going. A better version of the thing you said. Or permission to stop thinking about it. Those are real needs and none of them are silly. The problem is that a text thread can only answer the first two badly and cannot answer the last two at all. Naming which one you are after is often enough to close the app.

No. Wanting it is normal. Going back for it after you have it is the part with a cost. A meta-analysis of 38 studies covering 6,973 people (Starr & Davila, 2008, Journal of Abnormal Psychology) found that repeatedly asking for reassurance you have already been given tracks with low mood (r = .32) and, across 16 further studies, with being pushed away (r = .14). That second number is small. These are correlations, so the arrow could point either way. Reading the thread again is one way of asking without having to ask.

Watch it instead of reliving it. Ethan Kross and Ozlem Ayduk have run this experiment many times: people asked to revisit an upsetting moment through their own eyes get upset again and learn little. People asked to watch the same moment from a step back, as though seeing themselves in it, can work out what happened without the feeling flooding back. Same event, same person, different seat. Ask what happened in the conversation. Not why you are like this.

Yours. Their side is the part you cannot verify and have already read twenty times. Your side is the part you can check and the only part you can do differently. So look for the specifics. Did you ask a question you did not want the answer to? Did you make the thing you wanted sound smaller than it was? Did you go quiet right after saying something real? Those are facts about the thread, not verdicts about you.

No, and that is the useful part. A move is something you did in the conversation. Rereading happens after it, alone, and leaves no trace in the thread. The moves are in what you sent next: the follow-up into the silence, the check-in that was really a check, the softened version of what you meant. Movecheck reads one conversation and names those. It shows the exact line each name came from, and it only ever names yours. No score. No type. No verdict about either of you.

• Kruger, Epley, Parker & Ng, Egocentrism over e-mail: can we communicate as well as we think?, Journal of Personality and Social Psychology, 89(6), 925–936 (2005). doi.org/10.1037/0022-3514.89.6.925
• Treynor, Gonzalez & Nolen-Hoeksema, Rumination reconsidered: a psychometric analysis, Cognitive Therapy and Research, 27, 247–259 (2003). doi.org/10.1023/A:1023910315561
• Starr & Davila, Excessive reassurance seeking, depression, and interpersonal rejection: a meta-analytic review, Journal of Abnormal Psychology, 117(4), 762–775 (2008). doi.org/10.1037/a0013866
• Kross & Ayduk, Making meaning out of negative experiences by self-distancing, Current Directions in Psychological Science, 20(3), 187–191 (2011). doi.org/10.1177/0963721411408883