A calm record

Built from 132,342 messages — held with care, not judgment.

A note from the author (Im 'A' for clarity). While I actually functionally developed the data science and ML processes that derive the datapoints (keyword/dictionary analysis, sentiment analysis, time modeling) I will VERY honestly say that a very trained llm on a lot of aggregate data from my relationship wrote most of the rest of this site. ill eventually rewrite it all. I havent exactly told my therapist about this yet. I work in tech and wanted to leverage locally derived ML techniques to distract myself because, well, to be blunt while i have a good sense of some secure attachment, my upbring very deeply solidified a fearful avoidant nature that im only recently realizing. SO.. while a majority of the data derived here was developed by me, i apologize that i had no energy to draft this site. I did feel i wanted to share this all though, maybe to help myself work through it all, and since FA's are so negatively framed maybe a data driven journey could bring some things to light for folks. So,, I had a large dataset of messages from a relationship that started long-distance. It was beautiful but tough, and ended in near misery for both of us, without true closure or healing as is probably expected. With a little bit of time to understand attachment styles and to give myself a distraction while i plan out new coping mechanisms, I wanted to put some skills to work at home. Getting value out of my last relationship's chats felt like something.... practically more closure than anything I received in reality after things fell apart. We were regular Telegram users, so the data was easy to export and I wanted a positive outcome. So: storytelling through data. De-identified, of course.

This is the story of a relationship between two people — referred to here as A and S — told through the actual record of what they said to each other. Over 5.5 years they produced 8,297 instances of verbal affirmation, 3,450 repair attempts after conflict, and 5,033 expressions of joy. They also produced repeating cycles of hurt that neither person fully understood while they were living through them.

This project is for people who have lived through difficult relational cycles and want something steadier than a single narrative. It aims to be unbiased and peaceful: not a prosecution, not a defense — just a measured view of what happened over time, with context and de-identified evidence.

If you recognize yourself here — the fearful-avoidant who shields until they shatter, the anxious pursuer who can't tolerate silence, the couple locked in a loop where each person's self-protection triggers the other's deepest alarm — you're not alone. If You Recognize This is written for you. In Their Words shows the actual messages.

At a glance

—

Weeks of messages

—

Time span

—

Messages analyzed

—

Difficult episodes

—

Episodes where someone tried to repair

Connection and strain over time

Two composite scores are tracked week by week. Connection combines engagement with warmth. Strain combines conflict signals, withdrawal signals, stress signals, and repair attempts.

Green = Connection Brown = Strain

Context note: the early high message volume reflects a long-distance period (logistics + access), and is presented as context rather than “intensity.”

Explore

In Their Words — the actual messages: what they wrote when falling in love, when hurt, when trying to repair.
Our Story — the arc from first message to last.
The Patterns — the three feedback loops that defined the relationship.
What Got Bruised — the injuries that accumulated and the self-protection that deepened them.
If You Recognize This — for anyone who's lived through similar dynamics.

Connection vs. Strain

Tip: The Data page supports zooming, normalization, and “parallelism” scaling.