Algorithmic Relationship Modelling

A simple explanation

There is a way you would relate to the people you love if you had never seen a thirty-second relationship video. And there is the way you have begun to relate to them after eighteen months of those videos — a thousand small templates for how a partner should apologise, how a friend should show up, how a milestone should be marked, what a green flag looks like, what a red flag looks like, what counts as a good fight. The templates were not chosen. They were absorbed.

This is algorithmic relationship modelling. The Reward System, asked for connection, accepts whichever relational script the feed has been rewarding this season. The person you are with no longer just appears as themselves; they appear partly as a candidate for or against the template. The relationship runs against an algorithm-curated benchmark that no one in the relationship wrote.

An everyday example

Your partner forgets the small thing — the dentist appointment, the second coffee, the way you took the news. They apologise. Six months ago you would have moved on within an hour. Tonight a feeling sits longer than it should. By the next morning you can almost picture the comment-section reading of the moment: bare minimum, one-sided emotional labour, gentle parenting your boyfriend. The captions arrive faster than your own reading does.

Or your friend texts you something supportive. You read it once, do not respond, scroll for nine minutes, and then read it again against the what a real friend would have said template you saw on the For You page on Tuesday. The text is fine. The template was firmer. By the time you reply, your reply is colder than you wanted it to be, and neither of you knows why.

Why does my relationship feel like a TikTok script?

Because you have been training on TikTok scripts for hundreds of hours, and the system does not have a separate compartment for real life. The same Reward System that catalogues fluent feed-content catalogues fluent relationship-content. When the feed-content is dense with relational scripts — beats, verdicts, comparison shots, dramatic closings — those scripts become the available repertoire when the system reaches for how should this moment go.

Real relationships do not move in thirty-second beats. They move slower, messier, with longer recovery and more ambiguity. Measured against a script that was edited for engagement, almost every real moment falls slightly short — not because the moment was bad, but because the comparison is structurally tilted.

The behavioral loop

A loop that runs across many small interactions:

1. Template ingestion — short videos deliver an endless catalogue of relational beats: how he texts back, what she did when he forgot, how they handled the in-laws.
2. Implicit benchmarking — the Reward System files the beats as patterns. They become part of the available script for what relationships should look like.
3. Trigger event — a small real-life moment occurs. A reply, a forgetting, a tone, a milestone.
4. Template recall — within seconds, one or more of the beats arrives as a comparison. The recall is fast and often unnamed.
5. Verdict — the moment is scored against the beat. The actual person is slightly displaced; the candidate-for-the-template takes their place.
6. Performance or withholding — the response is shaped to fit the script — a colder reply, a sharper boundary, a captioned reaction, a withheld warmth.
7. Local reward — the script produces a recognisable beat. The Reward System logs success. The relationship loses a small unit of slack.
8. Drift — over months, the relationship begins to be experienced primarily through the templates. The partner or friend stops appearing as themselves and begins appearing as the role they are being cast in.

Emotional drivers

Four feelings that quietly run the loop:

• A faint chronic dissatisfaction with relationships that have not done anything new wrong, mistakenly attributed to the people involved.
• An anticipatory wariness of small moments — how will this hold up against the template I saw yesterday — which thins presence.
• A subtle pride in naming the dynamic in feed-vocabulary, which the Reward System logs as insight and which usually delays actual contact.
• A grief, often unnamed, when a relationship that was once spacious begins to feel evaluated.

What your nervous system does

Each comparison runs as a small evaluative spike — a fraction of a second of analysis under threat-tinted attention. The body learns to enter relational moments slightly braced, slightly forecasting, slightly camera-ready for a moment that could be captioned later. Over time the somatic signature of being-with shifts: less open, more measured, more analytic.

The partner or friend often feels the change before the loop-runner can name it. They will sometimes report a faint sense of being evaluated without being able to point to evidence. The evaluation is real, but it is not coming from the person across the table. It is being run silently against a feed-trained template the other person never agreed to.

The DojoWell interpretation

The Reward System's original ask was connection — the meaning-bearing deposit of being-with another person without measurement. The substitute is a curated relational template that produces beats but not depth. They share a surface property: both look, from the outside, like attentive relating. From the inside, one is contact and the other is casting.

Density reads false_progress because the templates are convincing. The Reward System sees them work — the boundary is set, the green flag is honoured, the milestone is hit, the moment is named — and it logs a clean win. What it cannot see is what is no longer being deposited. The slow accretion of shared idiosyncrasy that only this relationship would have produced. The internal vocabulary that no feed could supply. The trust that comes from un-scripted presence. These cannot be templated. They can only be lived.

This is also why the cost is most visible in the long arc. A relationship run on templates can feel intense in short windows and remarkably hollow at three years. The deposits did not happen. The residue — the small wariness, the rehearsed responses, the persistent comparison — did. The relationship is not less alive than it could have been because of the people in it. It is less alive because a third party — the feed — has been quietly co-authoring it.

How do I tell my own values from the relationship advice on my feed?

By introducing duration, again. The values that survive a month without the feed are closer to yours. The verdicts that arrive within seconds of a real moment are usually the feed's. The vocabulary you reach for under stress is a reliable tell — if it is platform-native, it is borrowed.

This is not an argument against the feed's vocabulary entirely. Some of it is genuinely clarifying. The work is to know which is which. A value held without the feed for a month, that still holds, is yours. A verdict that requires fresh exposure to keep firing is not.

Practical steps

1. Notice when a feed-phrase arrives faster than your own reading. Bare minimum, love-bombing, bread-crumbing, avoidant. The phrase is not always wrong; arriving faster than your own reading is the tell. Pause before deploying it.
2. Run a seventy-two-hour rule on relationship verdicts. Anything decided about a partner or friend within seventy-two hours of relevant feed-exposure waits. Most of the heat does not survive the wait.
3. Pick one beat you've imported and retire it for a season. A specific milestone, a specific conflict-style, a specific kind of reaction-video script. The relationship will not collapse; the slack will return.
4. Re-introduce un-photographable time. Long walks, slow meals, shared boredom. The relationship needs hours that cannot be captioned. The capacity for those hours is the first thing the templates erode.
5. **When a comparison fires, ask: whose script is this?** If the answer is not yours and not your partner's, the moment is being co-authored by a feed. Naming it is enough to recover the moment.

Reflection questions

• Which feed-phrase do you reach for most often when describing a current relationship? Whose vocabulary is it?
• Which moments in the relationship are for the relationship, and which are quietly composed against a template?
• How do I know if my dissatisfaction is with my partner or with the gap between them and a script that doesn't apply?
• If you had not seen a single relationship video this year, which of your current grievances would survive?

Frequently Asked Questions