Sample · anonymized

“A slow-cooled relationship that ended by attrition, not conflict.”

2,847 messages · 241 days · You & Them · Confidence 8.4/10

Messages

2,847

Span

241 days

Your latency

12 min

Their latency

3h 20m

Overview

Status: dormant

High-frequency reciprocal exchange (months 1–3) → asymmetric maintenance (4–6) → withdrawal (7–8). No single rupture — a gradual attrition of bids and repair.

Chart

Message volume, month by month

Stacked area of how many messages each side sent per month. Divergence between the two curves is the earliest sign of attrition — when one side keeps climbing while the other flattens, the conversation is being carried, not shared.

Reading & trend

Peak is "You total" at 2298; the weakest point is "Them total" at 1649. Mean sits at 1974. The gap (sharply uneven) is the story — the profile is being held up by "You total" while "Them total" is silently pulling it down.

Boost it

Prioritise "Them total" — it's the cheapest lift and the most visible gain. Anchor one concrete daily micro-move to it (a specific bid, question, or acknowledgment). Keep "You total" as-is; don't over-optimise your strength or it flips into a caricature.

Don't break it

Don't lecture the other side about "Them total" — pointing at the low bar creates defensiveness and drops it further. Also don't lean harder on "You total" to compensate; overuse of a strength is the fastest way to unbalance the whole profile.

Dimensions

Reciprocity & presence

Chart

Dimension spider — this module at a glance

Each spike is one Overview dimension scored 0–10. The red shape is this chat; the muted ring is the healthy benchmark. Dents pointing inward flag where the balance broke — usually initiation and presence before volume.

Reading & trend

Peak is "Responsiveness" at 5.1/10; the weakest point is "Initiation balance" at 2.9/10. Mean sits at 4.1/10. The gap (uneven) is the story — the profile is being held up by "Responsiveness" while "Initiation balance" is silently pulling it down.

Boost it

Prioritise "Initiation balance" — it's the cheapest lift and the most visible gain. Anchor one concrete daily micro-move to it (a specific bid, question, or acknowledgment). Keep "Responsiveness" as-is; don't over-optimise your strength or it flips into a caricature.

Don't break it

Don't lecture the other side about "Initiation balance" — pointing at the low bar creates defensiveness and drops it further. Also don't lean harder on "Responsiveness" to compensate; overuse of a strength is the fastest way to unbalance the whole profile.

4.2 /10

The give-and-take is meaningfully skewed toward you. In months 1–3 the exchange was near-symmetrical — both sides opened threads, asked follow-up questions, and matched effort. From month 4 onward, your bids started returning smaller replies or none at all, while your own investment held steady or grew.

The pattern isn't hostile — it's asymmetric maintenance: you keep the thread alive; they respond just enough to prevent an explicit rupture. That's the classic signature of a relationship cooling toward dormancy rather than ending in conflict.

Evidence from the chat

you· Sat 21:14
"still on for saturday? was thinking that place near yours, or wherever works — up to you"
Plan-confirmation bid two days before the date. No reply for 48h; the plan was quietly dropped.
them· Wed 09:02
"ya sounds good"
Reply to a 3-paragraph message from you about a work situation. Represents the modal reply length in months 5–8.
you· Mon 22:47
"hope your week's better than mine 🙃 also — did you end up going on sunday?"
Warm re-engagement with a specific callback question, five days after the last exchange. No reply until you sent a second nudge.

3.8 /10

You send roughly 2.1× the words they do across the full corpus, and the gap widens month over month. Their share of the wordcount drops from ~47% in month 1 to ~19% by month 8.

Volume asymmetry alone isn't a red flag — some people are just shorter texters. What matters here is that the drop tracks the drop in initiations and questions, which suggests reduced engagement rather than a style difference.

Evidence from the chat

you
"today was actually kind of a lot — the meeting I told you about went sideways in the last 10 minutes and I'm still processing it. tell me about yours?"
Multi-clause disclosure inviting reciprocation.
them
"oof. mine was fine"
Reply to the message above. No follow-up question, no elaboration.
you· Aug 12, 21:04
"okay full download — the trip got moved to the 22nd, my mom's asking if you're coming, I said I'd check. also the thing at work I told you about last week actually resolved (sort of). how was today?"
Three-topic update from you (68 words) that returned a 6-word reply the next morning. Representative of the H2 word-count gap: ~2.1× yours vs theirs across the corpus, ~4.9× in the last month alone.

2.9 /10

In February and March you and they opened new threads at almost identical rates (52% / 48%). By June, 71% of the first-message-after-6h-silence exchanges were coming from you. In July and August your share climbed to 84%; there are entire weeks in that window where every single new thread starts with a message from you.

The messages you're using to reopen the channel have also gotten softer and more self-deprecating over time — "hey stranger — you alive over there?", "no pressure but…", the sighed emojis. That drift in your own opener style is itself evidence you're feeling the imbalance and trying to hide it.

Evidence from the chat

you· Sun 10:11
"morning — how'd yesterday end up? tell me everything ☕"
Sunday-morning check-in initiated by you for the 4th consecutive week.
you· Thu 19:30
"hey stranger — you alive over there?"
First message after 3 days of silence. Note the softened, self-deprecating framing to lower the ask.
you· Aug 5, 22:58
"no pressure but if you're around this weekend I'd love to see you 🙃"
Sixth consecutive week where you opened every new thread. The "no pressure but" + sighed-emoji signature appears in 9 of your last 14 openers — it wasn't in a single opener before month 4.

5.1 /10

Your median reply is ~12 minutes; theirs is ~3h20m, with a long tail of 24h+ silences on emotionally charged threads. Response time is bimodal on their side: quick and warm on logistics, delayed and terse on anything relational.

That pattern — fast on low-stakes, slow on high-stakes — is a mild avoidant signature. It's not necessarily deliberate, but it consistently starves the vulnerable threads of momentum.

Evidence from the chat

them
"yes! 8pm works"
Reply within 4 minutes to a logistics question.
them
"sorry — busy day"
Reply 31 hours later to your message: "I've been thinking about what you said on friday. can we talk?"
them· Jun 18, 14:22
"which restaurant did you say again?"
Reply in 2 minutes to a logistics question, sent inside the same hour that a 19-hour silence was still open on your vulnerable message from the previous night. Latency bimodality captured in one afternoon.

4.4 /10

After month 4, their callbacks to things you'd told them nearly stopped. The family situation you shared in early April, the interview process you walked them through across three messages in mid-April, the sick friend you mentioned in May — none of these ever resurfaced from their side, not as a question, not as a check-in, not even as a passing reference.

Meanwhile you kept threading their details back into your own messages: their sister's move, the deadline they'd been dreading, the restaurant they said they wanted to try. That asymmetry — you holding both sides of the shared context — is what pulls this score below the benchmark. It's not that they're absent in the moment; it's that between messages, the relationship stops occupying space on their end.

Evidence from the chat

you
"btw — did you ever hear back about the thing your sister was dealing with? been thinking about you both"
You initiate a callback to a topic they mentioned 3 weeks earlier.
them
"lol what topic"
You had referenced a specific plan they'd described two weeks earlier. They no longer remembered mentioning it.
you· May 21, 08:15
"how did the interview end up going?? been thinking about you all morning ❤️"
Unprompted callback to their job interview from 4 days earlier — a topic they'd walked you through across three messages. No equivalent callback from their side to your April family news or May sick-friend disclosure.

Predictive trajectory

6-month projection

best18%

Direct DTR conversation → renewed cadence.

likely62%

Dormant → closed within 6 weeks.

worst20%

Brief revival followed by ghosting.

Chart

6-month health forecast with uncertainty band

The line is the model's most-likely trajectory of relationship health (0–10); the shaded band is the 80% confidence range. A narrowing band means the pattern is locked in; a widening band means small choices now still move the outcome.

Reading & trend

Peak is "M+1" at 3.8/10; the weakest point is "M+6" at 1.4/10. Mean sits at 2.5/10. The gap (uneven) is the story — the profile is being held up by "M+1" while "M+6" is silently pulling it down.

Boost it

Prioritise "M+6" — it's the cheapest lift and the most visible gain. Anchor one concrete daily micro-move to it (a specific bid, question, or acknowledgment). Keep "M+1" as-is; don't over-optimise your strength or it flips into a caricature.

Don't break it

Don't lecture the other side about "M+6" — pointing at the low bar creates defensiveness and drops it further. Also don't lean harder on "M+1" to compensate; overuse of a strength is the fastest way to unbalance the whole profile.

Dimensions

Momentum & risk

Chart

Dimension spider — this module at a glance

Spider of momentum, volatility, trajectory risk, and revival potential. Momentum + revival potential outward and volatility + risk inward is the resilient signature; the inverse is the drifting-apart signature.

Reading & trend

Peak is "Trajectory risk" at 7.8/10; the weakest point is "Momentum" at 2.4/10. Mean sits at 4.3/10. The gap (sharply uneven) is the story — the profile is being held up by "Trajectory risk" while "Momentum" is silently pulling it down.

Boost it

Prioritise "Momentum" — it's the cheapest lift and the most visible gain. Anchor one concrete daily micro-move to it (a specific bid, question, or acknowledgment). Keep "Trajectory risk" as-is; don't over-optimise your strength or it flips into a caricature.

Don't break it

Don't lecture the other side about "Momentum" — pointing at the low bar creates defensiveness and drops it further. Also don't lean harder on "Trajectory risk" to compensate; overuse of a strength is the fastest way to unbalance the whole profile.

2.4 /10

Every measurable channel — message volume, initiation, question rate, disclosure — is trending down on both sides, with steeper decline on theirs. There is no upward micro-trend in the last 6 weeks.

Without an intentional intervention, extrapolation is straightforward: the chat will thin further and become episodic within 4–8 weeks.

Evidence from the chat

them· Aug 25, 22:14
"yeah"
Modal reply shape in the last two weeks. Compare with the average length of their Feb replies (~34 words) — the volume decline made concrete in a single message.
you· Aug 28, 20:47
"how was today"
Your check-ins have shortened too — the H1 version would have been three sentences with a specific callback. Bilateral decline, steeper on their side.
them· Aug 30, 09:03
"fine"
One-word reply, 13 hours later. This exchange as a template is now the most common shape of a full "conversation" — the episodic pattern the score projects.

3.6 /10

Fit a line through the monthly message counts and the residuals are tiny — no month is more than ~15% off the trend line. There is no rupture week, no breakup-and-back-together spike, no month where things suddenly warmed back up. The 8-month curve is almost boringly smooth.

That smoothness is precisely what makes this configuration high-risk for a silent ending: there is no single week you can point at and say "that's where it went wrong", which means there is also no single week that has ever forced a real conversation.

Evidence from the chat

you· Jul 8, 22:41
"you never actually says the hard thing back — never does apparently"
The single most volatile line in 8 months. Aftermath: one flat reply the next morning, then everything returns to trend. The chat can't hold a spike long enough to force change.
them· Jul 9, 08:14
"i don't know what to say to that"
Response to the most volatile message of the corpus. Small, non-escalating, absorbs the spike back into the smooth curve within 12 hours.
you· Jul 10, 19:02
"anyway how was your day"
48 hours after the biggest spike of the year, you're already back on the trend line. The mechanical proof that volatility is too low to force a conversation.

7.8 /10

Extrapolating the current volume decline (them: −38% month over month in the last quarter; you: −28%), the crossover into <10 messages/week from their side lands in mid-October. Combined with the perception gap now sitting at 4.1 and repair success halving between H1 and H2 of the chat, the 6-month probability of the relationship ending without an explicit conversation is elevated.

This isn't fate. It's a description of the current slope: what the last 8 months of this specific chat, continued unchanged, add up to.

Evidence from the chat

them· Aug 31, 21:47
"sorry been a week. talk soon"
"Talk soon" appears 4 times in the last 30 days and never resolves into an actual thread. The exact linguistic marker of the trajectory the score projects.
you· Sep 2, 20:11
"no worries, whenever. here when you're around"
Your softeners are absorbing the decline instead of naming it. Continuing this exact response pattern is the input to the extrapolation.
them· Sep 6, 22:33
"this week was a lot. next week hopefully calmer"
"Next week" as an ongoing deferral. When "next week" becomes the default cadence, the mid-October crossover the analysis names is on schedule.

3.2 /10

The specific ingredients required for a revival are all documented in this chat: contempt score of 0.5 (nothing corrosive to un-say), the cat-contract joke still landing in month 6, four documented instances of accurate empathy from them, and your own investment still measurably intact. The blocker isn't capacity.

What's missing is a single documented instance across 8 months of them moving against their own deactivation — reopening a hard thread they'd let drop, calling back to a difficult conversation from their side, initiating a repair after their own silence. The chat contains the raw materials for revival; it does not yet contain the specific move that would trigger one.

Evidence from the chat

them· Aug 4, 21:11
"the cat contract still stands btw. non-negotiable"
Month 6, unprompted callback to the running joke from Feb. Proof the raw material for connection is intact.
them· Jul 12, 00:14
"hey — you don't have to hold it together for me right now. text whatever, even if it's not pretty. I'm here."
Accurate empathy landing in month 7. Capacity for connection isn't the blocker — the missing move is the one after this: reopening the thread the next day.
you· Aug 20, 19:47
"still would build a whole weekend around you if you asked"
Your investment measurably intact 8 months in. The materials for revival are on both sides; what's missing is the specific unilateral move from theirs.

Communication

Voice & register

Chart

Dimension spider — this module at a glance

Spider of the five communication dimensions. A lopsided shape (e.g. high verbosity but low question rate) means one side talks at the other; a balanced pentagon means real back-and-forth.

Reading & trend

Peak is "Humor" at 7.5/10; the weakest point is "Question rate" at 5.4/10. Mean sits at 6.6/10. The gap (uneven) is the story — the profile is being held up by "Humor" while "Question rate" is silently pulling it down.

Boost it

Prioritise "Question rate" — it's the cheapest lift and the most visible gain. Anchor one concrete daily micro-move to it (a specific bid, question, or acknowledgment). Keep "Humor" as-is; don't over-optimise your strength or it flips into a caricature.

Don't break it

Don't lecture the other side about "Question rate" — pointing at the low bar creates defensiveness and drops it further. Also don't lean harder on "Humor" to compensate; overuse of a strength is the fastest way to unbalance the whole profile.

6.8 /10

Your average message is ~34 words; theirs is ~11. Neither is inherently good — but the gap creates asymmetric context density. Your messages carry the emotional and logistical weight of the exchange; theirs largely react to it.

Over time, this pushes the relationship into a facilitator/responder dynamic that's hard to escape without an explicit reset.

Evidence from the chat

you
"okay so — three things: 1) I found that book you were talking about, 2) can't do thursday but friday still open?, 3) hope the presentation went okay ❤️"
them
"friday works. ty"
you· Jun 23, 20:41
"quick brain-dump because I can't sleep: the thing on friday is bothering me more than I thought. I think it's less about what actually happened and more about the fact that it was the third time this month I noticed it. anyway — how was your day, sorry for dumping"
58-word message got a 4-word reply the next afternoon. The verbosity gap isn't stylistic — it correlates with the same threads that get long latencies.

7.2 /10

Both sides show above-average vocabulary variety, but yours is meaningfully broader — you introduce ~2× the unique word stems per 100 messages.

Richness is a proxy for cognitive investment in the exchange. Their vocabulary tightens noticeably after month 4, drifting toward a small pool of acknowledgement tokens ("ya", "lol", "fair", "true").

Evidence from the chat

them
"true. fair enough tbh"
Representative of ~38% of their replies in months 6–8.
you· Apr 8, 19:12
"there's this specific brand of tired where you're not sleepy, you're just… hollowed out. today was that. not the meeting itself — the anticipation loop I ran for six hours before it."
You still deploy metaphor and precise register into month 4. Their reply: "ugh yeah". Same content-space, very different vocabulary investment.
them· Jul 30, 21:04
"ya. fair."
Modal 2-token reply in month 7. In months 1–3 their acknowledgements were still varied ("makes sense", "totally hear that", "you're not wrong") — that variety collapses to the "ya/fair/true/lol" pool by H2.

5.4 /10

You ask 3.1× more questions than they do. Questions are the primary tool for signalling curiosity about the other person; the imbalance means the conversation is largely one person interviewing another.

Worth noting: your questions are also higher-quality — open, specific, referencing prior details. Theirs, when present, are almost entirely logistical ("what time?", "where?").

Evidence from the chat

you
"how did you actually feel walking out of that meeting — like, in your body?"
them
"what time you thinking?"
you
"what's the part of this week you're least looking forward to?"

6.1 /10

Your disclosures reach layer 3–4 (values, fears, patterns). Theirs plateau at layer 2 (events, opinions). Neither side offers layer-5 vulnerability (identity-level shame, formative fear).

Disclosure is roughly proportional to felt safety. The one-way depth suggests you experienced this space as safer than they did — a common pattern in anxious/avoidant pairings.

Evidence from the chat

you
"I think part of why this scares me is that the last time I trusted someone with this exact thing, they used it later. so I'm trying to notice when I'm bracing for that with you."
them
"that sounds rough. sorry you dealt with that"
Empathic in tone, but doesn't reciprocate with an equivalent disclosure.
you· Mar 27, 23:52
"the thing I don't say out loud is that I'm always a little bit scoring the relationship in my head — points for callbacks, points for initiations. I know that's not how it's supposed to work. it's just what my system does when it doesn't feel safe."
Layer-4 disclosure (naming a pattern of your own). No comparable pattern-level self-disclosure from their side in the whole corpus — the deepest they go is opinion-shaped ("my family is complicated").

7.5 /10

The cat-contract bit from February is still running in August — they replied "the contract is under review 🐈" as recently as month 6, which is one of the very few callbacks they still initiate unprompted. The mock-formal register you two built around planning ("pursuant to your last message…") shows up in both sides' messages across all 8 months.

The one visible dip is the week of July 8–14: three consecutive playful openers from you land with straight-faced logistical replies from them. That week is the only time in the whole corpus where the humor register breaks on their side — and it lines up exactly with the withdrawal cluster in the Signals tab.

Evidence from the chat

them
"you're impossible 😂"
you
"reminder that you are legally obligated to send me a picture of your cat every tuesday. it's in the contract"
them
"the contract is under review 🐈"
Callback to a running joke from month 1 — one of very few callbacks they still initiate by month 6.

Tone mix

How each side sounds

Chart

Tone share per side

Percent of each side's messages that carry a given tone (warm, playful, logistical, complaint, withdrawn). Compare the two bars per tone — mismatched columns are where you're speaking past each other.

Reading & trend

Peak is "Warm · you" at 42%; the weakest point is "Cold · them" at 4%. Mean sits at 20%. The gap (uneven) is the story — the profile is being held up by "Warm · you" while "Cold · them" is silently pulling it down.

Boost it

Prioritise "Cold · them" — it's the cheapest lift and the most visible gain. Anchor one concrete daily micro-move to it (a specific bid, question, or acknowledgment). Keep "Warm · you" as-is; don't over-optimise your strength or it flips into a caricature.

Don't break it

Don't lecture the other side about "Cold · them" — pointing at the low bar creates defensiveness and drops it further. Also don't lean harder on "Warm · you" to compensate; overuse of a strength is the fastest way to unbalance the whole profile.

Emotional dynamics

Warmth, validation, labor

Chart

Dimension spider — this module at a glance

Spider of warmth, validation, empathy accuracy, affect range, and emotional labor. A shape pulled to one side means one person is doing most of the emotional work; a shrunken shape overall means the register has flattened.

Reading & trend

Peak is "Affect range" at 7.0/10; the weakest point is "Emotional labor" at 3.1/10. Mean sits at 5.7/10. The gap (uneven) is the story — the profile is being held up by "Affect range" while "Emotional labor" is silently pulling it down.

Boost it

Prioritise "Emotional labor" — it's the cheapest lift and the most visible gain. Anchor one concrete daily micro-move to it (a specific bid, question, or acknowledgment). Keep "Affect range" as-is; don't over-optimise your strength or it flips into a caricature.

Don't break it

Don't lecture the other side about "Emotional labor" — pointing at the low bar creates defensiveness and drops it further. Also don't lean harder on "Affect range" to compensate; overuse of a strength is the fastest way to unbalance the whole profile.

6.9 /10

Baseline affection stays above healthy median for most of the chat. Terms of endearment, soft closings, and check-ins occur regularly on both sides — though yours are more frequent and more elaborated.

Warmth alone is not enough to keep a relationship healthy; here it papers over a growing gap in effort, which is why the exchange feels pleasant even as it structurally decays.

Evidence from the chat

you
"go rest — you've earned it. text me tomorrow when you're up ❤️"
them
"safe travels ✈️ ttyl"
them· Aug 2, 18:33
"proud of you btw. today was a big one and you handled it 💛"
One of ~11 unprompted warm messages from their side in H2 — the warmth channel stays alive even as callbacks, questions, and initiations all decay. This is exactly why the drift feels invisible from inside the chat.

5.2 /10

When you name a feeling, they most often acknowledge but rarely reflect it back. Full validation — naming the emotion, endorsing its legitimacy, checking meaning — appears in ~11% of relevant openings from their side.

Under-validation isn't cruelty; it's usually skill or tolerance. But it teaches you to bring smaller feelings into the room, which shrinks the emotional bandwidth of the relationship.

Evidence from the chat

you
"I think I'm actually still angry about how last week ended and I don't want to pretend I'm not"
them
"ok"
Neither dismissive nor validating — a null response to a costly disclosure.
them· Apr 22, 21:07
"that makes sense. of course you'd feel that way after what she said — anyone would. what do you need right now?"
One of the ~11% of their replies that fully validate: names the feeling, endorses legitimacy, offers a next step. The capacity is documented; the frequency isn't.

6.4 /10

The four moments where they did land empathy in this chat — the late-April reply about your family ("that's not you being dramatic. that's a lot for anyone."), the June message the night before your presentation, the July "you don't have to hold it together for me right now", and the August response to the news about your friend — are all short, well-aimed, and non-explaining. When they engage, they read the feeling correctly.

The issue is the ratio. Across the 47 messages where you named a feeling explicitly, only those 4 got a matched-empathy reply; 22 got acknowledgement without engagement ("ok", "oof", "that sucks"), 21 got no response or a subject change. The capacity is documented in the chat; the delivery isn't.

Evidence from the chat

them
"that's not you being dramatic. that's a lot for anyone."
Late-night reply after a difficult family update from you. Warm and accurate — a moment worth preserving.
them· Jul 12, 00:14
"hey — you don't have to hold it together for me right now. text whatever, even if it's not pretty. I'm here."
Third of the four documented matched-empathy moments. Short, accurate, non-explaining — exactly the register that lands.
them· May 30, 22:41
"oof. that sucks"
Reply to your message about the sick friend. Falls in the 22-message bucket of acknowledgement-without-engagement — the modal shape, not the exceptional shape.

7.0 /10

In a single week in early May this chat holds: the tender back-and-forth after their grandmother's news (Tuesday), the sarcastic argument about the restaurant (Thursday), the giddy 2am voice-note transcript about the concert (Friday), and your quiet grief-adjacent message on Sunday. Neither of you is locked into one register — both sides can move between tenderness, teasing, frustration and joy inside 48 hours.

The range narrows on their side after month 5 — the giddy end and the grief end both drop out, leaving mostly a middle band of neutral-warm. Yours holds. The affective bandwidth you can still write into this chat is wider than the one it will receive back.

Evidence from the chat

them· May 3, 02:14
"OKAY the encore. the ENCORE. i can't. voice-note incoming when i stop shaking"
Top of their affective range — giddy, uncontained, mid-week. This exact register disappears from their side after month 5.
them· May 1, 21:02
"she went in her sleep. i'm not okay but i'm also weirdly okay. i don't know"
Bottom of their affective range — grief-adjacent, unedited. The other pole that stops appearing later in the chat.
them· Aug 22, 19:41
"long day. fine tho"
Modal shape of their H2 messages — flat middle-band. Neither high nor low; just the narrowed bandwidth the score describes.

3.1 /10

You initiate ~82% of the repair moves after rough patches — apologies, re-approaches, softening messages. You also do the majority of the emotional scaffolding (remembering hard dates, checking in after known stressors).

This is the single largest asymmetry in the relationship. Sustained one-sided emotional labor is the most reliable predictor of the burnout that produces silent exits like the one this chat is trending toward.

Evidence from the chat

you
"hey — I know today's the anniversary. no need to reply, just wanted you to know I remembered. thinking of you."
Unprompted repair move on a date you know is hard for them. No reciprocal move on a comparable date for you.
you
"can we start over — I don't like how last night ended and I don't want it to sit between us"
Third repair opening you've initiated in six weeks; no self-initiated repair from their side in the same window.
you· Jun 15, 07:58
"morning ❤️ know today's presentation day — you've got this. don't reply, just wanted you to know I remembered"
Pre-emptive scaffolding on a stressor of theirs. Your own comparable stressor two weeks earlier (the July 1 deadline they knew about) went unmentioned by them.

Sentiment trend

Affect over time

Chart

Weekly sentiment, side by side

Rolling weekly sentiment from −1 (negative) to +1 (positive). Watch when the two lines decouple: healthy relationships track together, distressed ones diverge before the tone becomes explicitly negative.

Reading & trend

The shape is flat — everything hovers around 7.1/10. Nothing dominates, nothing collapses; this is a stable but low-signal picture. Trend: incremental, no single lever will move the whole shape.

Boost it

Pick ONE dimension to invest in for two weeks (not all of them) — a small deliberate lift on one axis is what tips a flat shape into a distinctive one. Repeat the behaviour that produced today's average, but with 20% more intention.

Don't break it

Don't chase novelty on every axis at once — flat shapes crack when someone tries to fix everything and burns out. Also avoid treating "average" as "fine"; flatness is often the shape of quiet drift.

Two-Way Mirror

Two-way mirror

Perception gap: 4.1 / 10

Your view
"we're just going through a slow patch"

Their view
"we've been drifting for months"

Chart

Perception radar — how each of you sees the same axes

Same four axes rated by each side. Where the two shapes overlap, you agree; where they pull apart, that's the perception gap — the wider the wedge between the two lines on an axis, the more that dimension is a source of silent misalignment.

Reading & trend

Peak is "Effort · you" at 8.4/10; the weakest point is "Future · them" at 2.5/10. Mean sits at 5.8/10. The gap (sharply uneven) is the story — the profile is being held up by "Effort · you" while "Future · them" is silently pulling it down.

Boost it

Prioritise "Future · them" — it's the cheapest lift and the most visible gain. Anchor one concrete daily micro-move to it (a specific bid, question, or acknowledgment). Keep "Effort · you" as-is; don't over-optimise your strength or it flips into a caricature.

Don't break it

Don't lecture the other side about "Future · them" — pointing at the low bar creates defensiveness and drops it further. Also don't lean harder on "Effort · you" to compensate; overuse of a strength is the fastest way to unbalance the whole profile.

Dimensions

Perception breakdown

Chart

Dimension spider — this module at a glance

Zoom into the four perceived dimensions (investment, interest, intimacy, conflict weight). This is your side's read of the chat — compare each spike against the benchmark ring to see which perceptions are grounded and which are inflated.

Reading & trend

The shape is flat — everything hovers around 4.4/10. Nothing dominates, nothing collapses; this is a stable but low-signal picture. Trend: incremental, no single lever will move the whole shape.

Boost it

Don't break it

4.0 /10

You almost certainly perceive your own investment as roughly matched by theirs. The data disagrees: your investment (initiations, word count, questions, repair) runs ~2× to 3× higher across every measurable channel.

They likely perceive the relationship as "lower stakes" than you do — which explains why routine drift doesn't feel urgent to them and does to you.

Evidence from the chat

you· Jul 3, 22:41
"i keep thinking about you today btw. no reason, just wanted to say"
Unprompted investment bid. No comparable unprompted message from their side in the same 30-day window.
them· Jul 4, 09:12
"aw. same"
Reply arrives 10 hours later, two words. The exchange perfectly captures the asymmetry the score names.
you· Aug 9, 20:03
"planning to be free sunday if you want to do something — brunch, walk, movie, i'll build it around you"
You carry the planning entirely. Repeated eight times in H2 with no reciprocal plan-building from their side.

4.5 /10

Interest here is measured by callbacks, follow-up questions, and unsolicited references to shared context. Yours toward them is high and consistent. Theirs toward you is polite but shallow after month 4.

The gap between how interested they seem in the moment (warm, playful) and how interested they are in the pattern (rare callbacks, no volunteered updates) is the core of the perception gap.

Evidence from the chat

you· Jun 27, 18:22
"how did the thing with your mom on saturday actually go — you said you'd tell me after"
Callback to a detail from 6 days earlier. You track their life across the silences.
them· Jul 11, 21:14
"good, thanks for asking"
Reply to your update about the friend situation you'd been processing all week. No follow-up question, no callback to earlier context.
them· Aug 2, 22:47
"haha yeah tell me more sometime"
"Sometime" is the H2 tell — polite interest that never becomes an actual question. Warm in the moment, shallow across the pattern.

5.2 /10

Message-by-message this is the dimension where you and they still look most similar: both use the pet names from month 2, both close conversations warmly, both keep the private-jokes vocabulary alive. If someone read any single day of this chat in isolation they'd read the two of you as close.

Zoom out to the week and the picture changes. The intimacy is real inside each exchange and does not carry across the silences — there are no unprompted "thinking of you" messages from their side after month 4, no mid-week check-ins that aren't logistical. Closeness in the moment, absence between moments.

Evidence from the chat

them
"you're one of like three people who actually get this about me"
Month 3. This kind of statement never recurs after month 5, even though the surface warmth continues.
you· Aug 14, 23:11
"is it weird that we can be this warm in a single exchange and then not talk properly for a week? I keep noticing that."
Explicitly names the closeness-in-moment / absence-between-moments gap. Got no reply for 40 hours and then a subject change — a live demonstration of the exact pattern the message named.
them· Jul 15, 08:47
"still thinking about last night ❤️"
Warm in-moment intimacy. No follow-up message from their side for the next 5 days — the exact pattern the score reflects.

3.8 /10

You appear to remember the two rough patches (April, June) as significant and unresolved. Their subsequent messages treat them as fully closed — no callbacks, no soft repairs, no evidence they're still processing.

Asymmetric weight on the same conflict is one of the most reliable sources of quiet resentment. The heavier-carrying partner starts to feel unheld; the lighter-carrying partner reads any callback as "still on that?"

Evidence from the chat

you· Jun 14, 23:02
"i know we technically moved past april but i don't think we actually talked about it and it's still sitting in me a bit"
You explicitly re-open the April rough patch two months later. Signals live weight on your side.
them· Jun 15, 10:41
"oh — i thought we were good. what do you want to talk about specifically?"
Reply reads the past incident as closed and requires you to re-load it. Confirms the asymmetric-weight pattern.
you· Jul 22, 21:18
"like the june thing too — never really landed for me but i don't want to keep dragging things up"
Self-suppression of a second live thread. The heavier-carrying partner starts pre-emptively editing what they raise.

How They See Me · addon depth

Perception-gap addon. Every number is inferred from behavioural signal in the chat, not from a survey of them. Treat as a well-calibrated mirror, not a verdict.

How They See Me

The perception you don't get to see from inside

You over-estimate what you show them on three axes and under-estimate what you show them on two. The result is that they read you as more contained and less invested than you actually feel — and you calibrate every next message to that misread. This is the loop 'How They See Me' isolates.

Kenny SRM · meta-perception

Reflected Appraisal Accuracy: 58 / 100

Source: Kenny Social Relations Model — meta-perception & reflected appraisal (Kenny, 1994; Kenny & DePaulo, 1993)·Confidence 78%

Chart

Three overlaid views per axis

Blue = your view of you. Gold = their view of you (behavioural inference). Red = your guess at their view. Where blue and gold diverge = self-other agreement gap. Where red and gold diverge = meta-error.

Reading & trend

Peak is "Care · you" at 8.9/10; the weakest point is "Neediness · guess" at 2.9/10. Mean sits at 6.4/10. The gap (sharply uneven) is the story — the profile is being held up by "Care · you" while "Neediness · guess" is silently pulling it down.

Boost it

Prioritise "Neediness · guess" — it's the cheapest lift and the most visible gain. Anchor one concrete daily micro-move to it (a specific bid, question, or acknowledgment). Keep "Care · you" as-is; don't over-optimise your strength or it flips into a caricature.

Don't break it

Don't lecture the other side about "Neediness · guess" — pointing at the low bar creates defensiveness and drops it further. Also don't lean harder on "Care · you" to compensate; overuse of a strength is the fastest way to unbalance the whole profile.

Chart

Meta-error vs agreement gap per axis

Meta-error (primary) = how wrong your guess about their view is. Agreement gap (gold) = how far apart you and they actually are. Big meta-error + small gap = miscalibrated worry. Small meta-error + big gap = accurately seen misalignment.

Reading & trend

The shape is flat — everything hovers around 1.8. Nothing dominates, nothing collapses; this is a stable but low-signal picture. Trend: incremental, no single lever will move the whole shape.

Boost it

Don't break it

Warmth

Δ 1.5

you

8.4

them

6.9

guess

8.1

You think your warmth is fully visible. Their behaviour reads you as slightly reserved — the drafted register masks the warmth in your intent.

Investment

Δ 2.8

you

8.6

them

5.8

guess

7.2

The largest agreement gap. You feel deeply invested; they see effort but not the full weight because you compress it into logistics.

Vulnerability

Δ 2.6

you

6.5

them

3.9

guess

5.1

Your retract-and-resend hides the vulnerability you feel you're offering. They receive the final version, not the drafted one.

Frustration

Δ 1.6

you

5.8

them

7.4

guess

5.2

Biggest meta-error. You think your frustration is barely visible; theirs reads you as visibly frustrated more often than you realise (July cluster).

Care

Δ 1.2

you

8.9

them

7.7

guess

8.3

The most accurate meta-perception. Care lands because it shows up in tone regardless of drafting.

Neediness

Δ 2.2

you

3.2

them

5.4

guess

2.9

You think you look composed; the frequency of check-ins reads to them as need, even when each individual message reads calm.

Reflected-appraisal read

Reflected Appraisal Accuracy is 58/100 — moderate. You are calibrated well on Care and Warmth, and poorly on Frustration and Neediness (the two axes you most try to control). The pattern is classic: the more you edit an emotion before sending, the worse your read of how it lands.

Johari Window

What's open, blind, hidden, unknown

Source: Johari Window (Luft & Ingham, 1955)

Open

Known to you · known to them — the shared, functioning surface.

Reliability on logistics

You show up on time and follow through. They read this, you know it.

Apr 3 — "you're the one person I don't have to double-check on plans"

Playful register

The private-jokes vocabulary is co-authored and mutually acknowledged.

May 18 — "our stupid little code words are half the reason i miss you"

Curiosity about their world

You ask about their work, family, projects — this is visible and they name it back.

Jun 2 — "you're the only one who actually asks about the small stuff"

Blind spot

Unknown to you · known to them — how they experience you that you don't see.

You come across as more contained than you feel

Drafted messages arrive polished; the felt intensity behind them doesn't cross the wire.

Mar 22 — Visible edit down to "anyway. how are you." — they receive calm; you sent controlled.

Your frustration is louder to them than you think

You believe you kept July civil. They mapped July as a hard month before you did.

Jul 27 — "is this the thing we do now" — you experienced this as one hard message; they experienced it as the fourth in a pattern.

Your check-in frequency lands as need, not care

Each individual ping reads warm; the rhythm reads anxious.

Jun–Jul — 4.3 initiations/week from you vs 1.2 from them — cumulative asymmetry.

Hidden

Known to you · unknown to them — what you carry but don't show.

The weight you give April's rupture

You still track it monthly. They treated it as closed by May.

Jun 12 — "still think about April sometimes" — the only callback on your side, sent alone.

You are counting

Latency, initiation ratio, callback rate — you track these implicitly. They have no idea.

Aug 14 — "is it weird that we can be this warm in a single exchange and then not talk properly for a week" — one-time surfacing of the constant internal ledger.

The future you're imagining

Your language contains three concrete future scenarios; you have never shared any of them.

May 30 — "if we can't fix this weekend we can't fix anything" — the only near-visible reference; retracted by follow-up message.

Unknown

Unknown to you · unknown to them — the material that would need a conversation to surface.

Whether their withdrawal is protection or preference

The chat cannot distinguish between 'they can't do closeness right now' and 'they don't want this level of closeness'. This is the single unresolved question the report can't answer without them naming it.

How they'd behave if you stopped over-carrying

You have never held the line long enough to see what they do in the void. The information doesn't exist yet.

What repair looks like to them

Neither party has named what a successful repair would feel like. Both are guessing.

Impression profile · Fiske SCM + relational value

How you land on six impression axes

Source: Fiske Stereotype Content Model (Warmth × Competence) + Bakan agency + Leary relational value·Confidence 81%

Chart

Six-axis impression radar (0–10)

Warmth × Competence is the Fiske SCM backbone; Agency, Reliability, Desirability, and Depth complete the relational-value picture.

Reading & trend

Peak is "Reliability" at 8.5/10; the weakest point is "Depth" at 6.1/10. Mean sits at 7.2/10. The gap (uneven) is the story — the profile is being held up by "Reliability" while "Depth" is silently pulling it down.

Boost it

Prioritise "Depth" — it's the cheapest lift and the most visible gain. Anchor one concrete daily micro-move to it (a specific bid, question, or acknowledgment). Keep "Reliability" as-is; don't over-optimise your strength or it flips into a caricature.

Don't break it

Don't lecture the other side about "Depth" — pointing at the low bar creates defensiveness and drops it further. Also don't lean harder on "Reliability" to compensate; overuse of a strength is the fastest way to unbalance the whole profile.

What the shape means

The impression profile leads with Competence + Reliability (both >8) and lags on Depth and Desirability. Translation: you land as the trusted, capable partner they can rely on — and slightly less as the person they cannot imagine losing. That is the single most actionable perception gap in the report.

Warmth

6.9

High, but muted by drafted register. Warmth in intent > warmth as received.

Competence

8.2

Read as reliable, capable, articulate — the single strongest impression axis.

Agency

7.1

You initiate, plan, and steer. Reads as agency; sometimes reads as pressure.

Reliability

8.5

The highest axis. You are the person who follows through. This is the impression pillar.

Desirability

6.4

In-moment high, sustained-moderate. Desire language on their side is warm but not urgent.

Depth

6.1

Perceived as thoughtful but slightly opaque — the Hidden quadrant is what caps this.

Sociometer · relational-value signals

Aggregate index +2.1 / ±5

Source: Leary Sociometer Theory — relational-value signalling (Leary & Baumeister, 2000)·Confidence 83%

Chart

Six signal channels (−5 rejecting … +5 valuing)

Each channel is a distinct way a person signals relational value. The shape between channels is the story — steady inclusion with falling prioritisation is a very different picture from uniformly warm signal.

Reading & trend

Peak is "Inclusion" at 8.6/10; the weakest point is "Bid-response" at 5.8/10. Mean sits at 7.1/10. The gap (uneven) is the story — the profile is being held up by "Inclusion" while "Bid-response" is silently pulling it down.

Boost it

Prioritise "Bid-response" — it's the cheapest lift and the most visible gain. Anchor one concrete daily micro-move to it (a specific bid, question, or acknowledgment). Keep "Inclusion" as-is; don't over-optimise your strength or it flips into a caricature.

Don't break it

Don't lecture the other side about "Bid-response" — pointing at the low bar creates defensiveness and drops it further. Also don't lean harder on "Inclusion" to compensate; overuse of a strength is the fastest way to unbalance the whole profile.

Inclusion

flat · +3.6

Apr 3 — "you're the one person I don't have to double-check on plans" — inclusion signal, unchanged across corpus.

Prioritisation

falling · +1.2

H1 vs H2 — Priority-language ("come first", "before anything else") occurs 7× in H1 and 1× in H2.

Public claim

flat · +2.4

Feb 14 — "my person" — used 3× in H1, 0× in H2.

Investment of effort

falling · +1.8

Jun–Aug — Their word-count per exchange drops 41% between Q2 and Q3.

Repair after harm

flat · +2.9

Jul 29 — "That wasn't fair of me and I know it landed badly. I'm sorry." — repair capacity intact even as prioritisation slides.

Bid-response

falling · +0.8

corpus — Turn-toward rate 54% (below the 65–70% stable benchmark).

Read of the sociometer shape

The sociometer index sits at +2.1 (mild positive) but the shape matters more than the number. Inclusion and Repair are intact and stable — you are inside their circle and they can come back after harm. Prioritisation, Investment, and Bid-response are all trending down — you are inside the circle but sliding toward its edge. This is the mechanism behind the felt sense of 'we're fine but something is off'.

Confidence intervals

How stable are these perception estimates?

Estimate	Point	95% CI	Stability
Perception gap (overall)	4.1	3.6 – 4.6	87%
Reflected Appraisal Accuracy	58	52 – 64	83%
Sociometer index	2.1	1.6 – 2.5	86%
Meta-error: Frustration	2.2	1.7 – 2.7	81%
Impression: Desirability	6.4	5.9 – 6.9	84%

Attachment & bonding

Style pairing

You

Confidence 8.4/10

anxious

Anxiously-attached with high verbal insight and sharp self-monitoring

Shows up as

·Bid intensity rises within 20 minutes of perceived distance (avg latency between your first and third bid after a silence: 18 min)
·Uses meta-language about the relationship ("we", "us", "what's happening between us") 3.4× more than they do
·Softeners jump from 2/week in H1 to 14/week in H2 — asks are pre-emptively made smaller
·Self-blames early in disputes ("maybe I'm being too much") before the other side has pushed back
·High disclosure depth even under low reciprocity — vulnerability is not gated by their response

Strengths

·Names the pattern out loud — the chat has a language for what's happening
·Repairs quickly, without grudge accounting
·High capacity for warmth even under strain (Jun 12, Jul 3 bids after silences)

Tripwires

·Silences longer than 24h after emotional content
·Short, logistic-only replies to a vulnerable message
·Cancelled plans without a rescheduling offer

Anchored evidence

May 9, 22:47
"is everything actually okay? not fishing — just noticed you've been quieter this week and wanted to check in ❤️"
Textbook anxious bid: hyper-monitored, framed as care, disclaimed ("not fishing") to lower the threat of the ask.

Jun 12, 21:03
"cool cool cool. no rush lol 🙃"
Protest disguised as ease — the anxious 'sarcastic softener' pattern that reliably lengthens the other side's next latency.

Jul 20, 23:58
"honestly at this point I feel dumb for still trying"
Self-directed protest that still functions as pressure. Followed by 41h of silence — the deactivation response the message was trying to prevent.

Them

Confidence 7.9/10

avoidant

Deactivating avoidant with intermittent warmth — the 'composed retreater'

Shows up as

·Reply latency to emotional content is 4.2× their latency to logistical content (26h vs 6h median)
·Uses "busy", "underwater", "head's elsewhere" 11 times in H2 — the deactivation vocabulary
·Topic-pivots 7 documented times away from relational threads to neutral content
·Warm re-engagement almost always follows their own silence, not yours — regulation on their timeline
·Never initiates a bring-me-close bid; warmth comes as reward after distance, not as reach

Strengths

·Not cruel, not contemptuous — the withdrawal is defensive, not punitive
·Autonomy-respecting: no monitoring, no possessiveness, no guilt-trips
·When they do show up warmly, the quality is real (Mar 14 phone call offer, Apr 27 long voice note)

Tripwires

·Direct emotional asks that require a same-day response
·Meta-conversation about the relationship framed as a problem to solve
·Any message that reads as 'we need to talk'

Anchored evidence

May 10, 09:12
"ya all good, just busy"
The modal deflection to your check-in bids. Short, warm-ish, closes the thread without opening a new one — canonical avoidant containment.

Jun 3, 12:14
"anyway did you see that thing your friend posted lol"
Subject-change reply to your unresolved 'last night' message. Topic-pivot as regulation — moves the conversation off the emotional register without refusing it.

Jul 16, 19:47
"gonna reply properly later I promise"
The 'proper reply later' that appears 6 times in H2 and resolves once. Deactivation dressed as good intention.

Pairing

Anxious ↔ Avoidant

A textbook anxious/avoidant pairing running its full loop — with an unusually mild surface (no contempt, no cruelty) that makes the underlying attrition harder to see and easier to tolerate.

The dynamic in this chat

The loop runs on a 3–5 day cycle in H2. You register distance (a shorter reply, a skipped call), reach with a soft bid, receive a warm-but-thin acknowledgment, register that the warmth didn't close the distance, escalate the bid (longer message, more vulnerability, sometimes sarcasm), and hit deactivation ("busy", "tomorrow", topic-pivot). They then re-approach on their own timeline 2–4 days later with a light message that resets the surface without addressing what was raised. You accept the reset because the alternative is losing the thread entirely. The cycle repeats.

Why it happens

Your nervous system reads distance as danger and closes it with reach. Theirs reads reach-under-pressure as engulfment and closes it with retreat. Each side's regulation strategy is the exact stimulus that dysregulates the other. Neither is doing anything wrong in isolation — the incompatibility is in the pairing, not the people. The pattern intensifies as the stakes rise, which is why H1 (lower stakes) reads healthier than H2 (higher stakes) even though both people are the same.

What it costs

For you: chronic hypervigilance, a growing self-image as 'too much', and the erosion of the felt sense that reach works. For them: a growing association between this chat and low-grade obligation, which further reduces the appetite to open it. For the bond: the loss of its capacity to metabolise anything harder than logistics. The couple that could handle the March health scare together in real time can no longer, by July, handle a missed birthday cue.

What breaks the loop

Not more reach. Not less reach either. The loop breaks when the anxious side can hold a bid without escalating it, and the avoidant side can name their own retreat as a state ("I'm going quiet — it's not about you, I'll be back Thursday") instead of leaving it undefined. Concretely: (1) named windows for hard conversation instead of ambush timing, (2) a shared vocabulary for 'I need space' vs 'I need you', (3) explicit re-entry commitments with dates. Without one of these, this pairing does not self-correct.

One full cycle — traced from real messages

1 · Distance detected
you
May 8: their replies drop from 40 to 6 words over 3 days.
2 · Soft bid
you
May 9, 22:47: "is everything actually okay? … just wanted to check in ❤️"
3 · Thin acknowledgment
them
May 10, 09:12: "ya all good, just busy"
4 · Escalated bid
you
May 12: longer message naming the pattern, asking for a call.
5 · Deactivation
them
May 12–14: no reply, then subject-change to weekend logistics.
6 · Reset without resolution
them
May 15, 18:22: light meme + "how was your day" — surface restored, nothing addressed.
7 · Reluctant acceptance
you
May 15, 18:34: warm reply back, question dropped, loop primed to repeat.

Dimensions

Bonding markers

Chart

Dimension spider — this module at a glance

Radar of the five bonding behaviors. An anxious pattern pushes proximity-seeking and protest outward while autonomy-support collapses; an avoidant pattern inflates deactivation and shrinks safe-haven use. A balanced pentagon is secure.

Reading & trend

Peak is "Proximity seeking (you)" at 7.8/10; the weakest point is "Safe-haven use" at 5.0/10. Mean sits at 6.5/10. The gap (uneven) is the story — the profile is being held up by "Proximity seeking (you)" while "Safe-haven use" is silently pulling it down.

Boost it

Prioritise "Safe-haven use" — it's the cheapest lift and the most visible gain. Anchor one concrete daily micro-move to it (a specific bid, question, or acknowledgment). Keep "Proximity seeking (you)" as-is; don't over-optimise your strength or it flips into a caricature.

Don't break it

Don't lecture the other side about "Safe-haven use" — pointing at the low bar creates defensiveness and drops it further. Also don't lean harder on "Proximity seeking (you)" to compensate; overuse of a strength is the fastest way to unbalance the whole profile.

7.8 /10

You reach out under stress — for reassurance, co-regulation, or just presence. Bids intensify (more messages, more warmth, more questions) precisely when you sense distance.

Proximity seeking is a healthy anxious-attachment strategy when met. When systematically under-met, as here, it becomes exhausting for the seeker and overwhelming for the avoider, feeding the very distance it's trying to close.

Evidence from the chat

you
"can we hop on the phone for 5? not a big thing — just want to hear your voice for a sec"
you· May 9, 22:47
"is everything actually okay? not fishing — just noticed you've been quieter this week and wanted to check in ❤️"
Bid triggered by 4 days of shortened replies from their side. Landed with "ya all good, just busy" — the modal deflection that reliably follows your proximity bids in H2.
you· Jul 3, 08:11
"morning — thinking of you. no need to reply, just wanted you to feel it from over here"
Bid explicitly framed as no-ask, sent on a Monday after a silent weekend. Your "no need to reply" softeners jump from 2 in H1 to 14 in H2 — proximity seeking rerouted through pre-emptive lowering of the ask.

5.0 /10

You brought hard news to them in-real-time twice this year (the March health scare, the June work situation) and post-processed twice (the April family thing surfaced only after you'd already spoken to your sister; the July argument with your friend was recounted a full week later). The two you brought in real time both followed clusters of warm, responsive replies from them; the two you delayed both followed multi-day silences.

On their side there are no equivalent moments — no message that reads as "I'm bringing this to you first because I need you here." Their difficult-day mentions come in as summary after the fact ("today was rough btw") without an ask attached. The chat isn't currently functioning as regulation infrastructure for them.

Evidence from the chat

you· Mar 11, 15:22
"okay i'm scared. can you just be around this afternoon in case the scan comes back weird"
Real-time safe-haven use from you — arrives during a cluster of warm, responsive replies from them. When the chat feels safe, you bring the hard thing in live.
you· Jul 27, 21:33
"btw the thing with sarah last week — it got kind of bad. anyway. just telling you now"
Post-processed a full week later, arrives after a multi-day silence from them. Delay maps precisely to how safe the chat felt in the window before.
them· Aug 19, 23:41
"today was rough btw. anyway how was yours"
Their canonical difficult-day mention — summary, no ask, immediate subject change. No equivalent "i need you here" message from their side across 8 months.

6.2 /10

Both sides broadly respect each other's independence — no possessiveness, no jealousy markers, no monitoring behaviour. Where autonomy is at risk is on your side: your protest bids in months 5–7 occasionally read as pressure to their avoidant nervous system.

Better differentiation here would look like naming your need without requiring an immediate reply as proof of care.

Evidence from the chat

them· Apr 8, 19:22
"go, have fun with them. i'll be here whenever, no rush"
Clean autonomy support — no jealousy, no monitoring, no strings. This register is stable from their side across all 8 months.
you· Jun 5, 23:47
"do whatever you need. i just don't want to be an afterthought — is that fair to say"
Autonomy respected in the first clause; the second clause is where the score drops. Names the need but tethers it to their next move.
you· Jul 18, 21:03
"you haven't texted since tuesday — everything alright?"
Reads as care from you, reads as monitoring on their avoidant end. The exact pressure-register the analysis names.

6.4 /10

Under sustained distance, you protest — pointed messages, sighed emojis, sarcastic softening ("no rush lol"). Protest is a normal anxious response to perceived withdrawal, but it reliably lands as pressure on their end and accelerates the very deactivation you're protesting.

Protest → deactivation → escalated protest is the classic anxious/avoidant cycle. This chat shows it clearly in May–July.

Evidence from the chat

you
"why does this always happen with us — every time it gets real you just… evaporate"
Sent after a 3-day silence that followed a vulnerable disclosure. The framing is accurate but arrives as an attack.
you· Jun 12, 21:03
"cool cool cool. no rush lol 🙃"
Sarcastic-softened protest 26 hours after your "can we talk?". The 🙃 + "lol" wrapper is protest disguised as ease — it reliably increased their next-reply latency rather than shortening it.
you· Jul 20, 23:58
"honestly at this point I feel dumb for still trying"
Self-directed protest ("I feel dumb") that still functions as pressure on the other side. Followed by 41h of silence — the exact deactivation cycle the analysis names.

7.1 /10

Deactivation shows as: increased latency after emotional content, shorter replies to vulnerable messages, subject-changes away from relational threads, and the "I'll reply later" that never resolves.

Deactivation is not a moral failure — it's a nervous-system response that under-serves both people. It also tends to be the behaviour least visible to the person doing it, which is why the perception gap on this dimension is often the widest.

Evidence from the chat

them
"tomorrow — head's underwater today"
Reply to your "can we talk about last weekend?". Tomorrow's message, when it came, was about weekend plans.
them· Jun 3, 12:14
"anyway did you see that thing your friend posted lol"
Subject-change reply to your "I don't want to leave what happened last night unresolved." One of 7 documented topic-pivots away from relational content in H2.
them· Jul 16, 19:47
"gonna reply properly later I promise"
The "I'll reply properly later" that appears 6 times in H2 and resolves into an actual proper reply exactly once. Canonical deactivation signature.

Partner psychological profile

The Composed Retreater

A composed, autonomy-first personality who metabolises intimacy in private and returns already-processed — warm when regulated, unreachable when not.

Confidence 7.6/10·Read from 4,127 messages

Integrity note. This is a read of how they show up across 4,127 messages over 8 months — patterns of language, timing, and response. It is not a clinical diagnosis, and it says nothing about who they are with other people, in other contexts, or in periods this chat did not capture.

Trait profile

Big Five, evidenced

Chart

Big Five trait profile

Scored 0–10 from patterns of language, timing, and response across the whole corpus. Traits are chat-specific reads, not clinical measurements.

Openness

7.2

Wide referential vocabulary, comfortable with abstract framing, initiates 4 of the 6 book/film references in the corpus. Openness is high in intellect; noticeably lower in the emotional-experience facet, where they prefer to arrive with the processing done.

Conscientiousness

6.8

Reliable on plans they've committed to (0 no-shows across 41 documented plans), keeps their word on logistics. The reliability collapses specifically in the emotional register — 'I'll reply properly later' resolves once in six occurrences.

Extraversion

4.4

Low-to-mid. Prefers voice notes over calls (11 vs 2 initiated), depth over breadth (repeatedly redirects group-chat talk toward 1:1 threads), and recharges by disappearing — the H2 latency pattern reads as introvert regulation as much as avoidance.

Agreeableness

6.5

Warm baseline register, no contempt, no name-calling across the entire corpus, and disproportionately gentle apologies when they arrive. Agreeableness is high on the interpersonal-warmth facet; lower on the compliance facet, where they will simply not do what they don't want to do.

Emotional stability

5.2

Surface-stable, under-the-surface reactive. Almost never expresses distress in real time; the tell is behavioural — latency spikes, topic pivots, and the reappearance-with-warmth pattern after silence. Stability is real for daily-life stressors and thin for relational ones.

Interpersonal comparison

You vs them, side by side

Chart

Relational axes — you vs them

Six axes that most shape how this specific pairing lands. Gaps aren't good or bad in isolation — they predict where friction will recur.

Motivations

What drives them, what scares them

Core drivers

Autonomy — being un-owned

The single most consistent thread across 8 months: language that preserves optionality ("if I can", "we'll see", "maybe next week"), resistance to standing arrangements, and warmth that spikes after independent time. Autonomy isn't a preference here — it's an identity anchor.

Being seen as competent and composed

Self-corrections when a message reads too raw (3 documented cases of retracted-and-resent messages in H1). Careful punctuation. Rare use of caps or multi-punctuation even under friction. Self-image as calm and capable is actively maintained in the writing.

Depth on their own terms

The quality of their disclosure when they choose to disclose (Feb 17 voice note, Mar 22 message about their father, Apr 27 long reply) is genuinely high. The pattern is: depth exists, but is gated by their sense of readiness — not by yours.

Underlying fears

Being pulled off their own timing

The latency after 'can we talk?' messages is 3.6× their baseline latency. The fear is not the content of the conversation — it's the loss of control over when it happens.

Being seen as the problem

The defensive shape appears specifically when they're named. Neutral observations receive engagement; observations that identify them as the source of a pattern receive 'busy' or a topic-pivot. Ego-protection is fast and largely unconscious.

Loss of the freedom to disappear

Every explicit ask for more regular cadence (yours on Mar 30, Jun 8, Jul 8) triggered a measurable next-week dip. The retreat is the response to the ask, not the content.

Defenses

How they self-protect under pressure

Intellectualisation

Converts felt states into concepts. Feelings get re-framed as analyses of feelings, which lowers the temperature at the cost of the connection.

Jun 2, 14:08: "I think what's happening is I'm just in a period where my bandwidth is compressed." — a genuine observation, but delivered as diagnosis rather than experience.

Deactivation

Down-regulates attachment by suppressing awareness of the bond when the bond feels demanding. Not conscious withholding — it's the nervous system muting the signal.

The 6 'I'll reply properly later' messages in H2, and the pattern of forgetting to circle back to threads they explicitly said they would. The forgetting is the defense.

Distancing humour

Uses lightness to metabolise moments that would otherwise require repair. Effective in small doses, corrosive when it becomes the modal response to your vulnerable messages.

Jul 1, 22:19: your "I've been feeling really alone in this" → their "same but for different reasons lol 😂" — humour as an off-ramp from the register you opened.

Topic-pivot

Redirects an emotional thread to a neutral one without refusing it explicitly, which preserves the surface of the relationship while draining the content.

Jun 3, 12:14: reply to your unresolved 'last night' message opens with "anyway did you see that thing your friend posted lol" — 7 documented occurrences in H2.

Language & register

How they use words with you

Emotional register

Their default register is warm-neutral: friendly, mildly playful, low-affect. They rarely go up into overt joy or down into named distress inside the chat. When they do reach for higher intensity it's almost always in one direction — care ("thinking of you", "proud of you") — while the harder emotions (fear, hurt, need) surface only in retrospective summary, if at all. This gives the chat an emotional bandwidth of maybe 40% of what a securely-attached partner would provide, and it explains why the relationship reads calmer than it actually is — the volatility is being absorbed off-screen.

Communication schema

They use language to maintain shape, not to co-regulate. Messages are drafted, softened, and often shortened before send (visible in the 4 documented cases where you can see a longer message being cut before a shorter one arrives). They favour statements over questions (2.1 statements per question, vs your 1.1) — which is a mild avoidant tell: questions invite reach, statements don't require one. Their asks are structural ("want to grab dinner Thursday?") rather than relational ("how are we?"), and they land follow-up questions on your topics ~30% of the time in H2, down from ~65% in H1.

Values

What they treat as non-negotiable

Independence

Never negotiates plans that would compress their solo time. Explicitly named on Feb 4 and Jul 12.

Loyalty within chosen commitments

No third-party flirt language, no comparative framing, no monitoring of your other relationships — the loyalty is real and quiet.

Fairness and even-handedness

Apologies, when they come, own the specific behaviour and don't blame you (Mar 14, Apr 27, Jul 29). Fairness is a value, even when execution is inconsistent.

Being useful over being close

The 3 moments they showed up strongly (health scare in March, your work situation in June, family thing in April) were all practical: rides, logistics, resource-finding. Presence-as-help comes easier than presence-as-witnessing.

Patterns

Cognition, regulation, strengths, blind spots

Cognitive patterns

·Reframes relational conflict as logistics ("we should just plan better") — solves the wrong problem elegantly.
·Reasons from principles rather than precedent — will re-decide the same question each time rather than commit to a pattern.
·Discounts the present intensity of feelings by projecting them as temporary ("this week is just weird") — sometimes accurate, structurally under-serves urgency.
·Preference for private processing → arrives at conclusions the other person has no visibility into until they land.

Regulation strategies

·Distance and time — the primary regulator. 24–72h of low-contact almost always precedes the warm re-approach.
·Physical activity — 4 documented mentions of running/gym as the reset (Feb 22, Apr 9, Jun 5, Aug 1).
·Focused work — 'head's underwater' is not always deflection; it's often literal.
·One trusted external outlet — references to "talked to M. about it" 3 times. They do process, just not in this chat.

Strengths

·Genuine warmth when regulated — not performed, not owed
·Zero-cruelty baseline — no contempt, no threats, no punitive silences with intent to hurt
·High-quality follow-through on practical commitments
·Real emotional depth when it does surface — the quality is not the constraint, the frequency is
·Self-aware about their own pattern in the abstract ("I know I go quiet") even when it doesn't yet change behaviour

Blind spots

·Under-estimates the cost of latency — treats a 3-day silence as neutral, doesn't register it as content
·Assumes their internal certainty about their care is transmitted without needing to be said
·Reads your escalation as the problem rather than the signal that the loop is running
·Believes 'I got busy' is a full account of what happened — doesn't see the pattern the busyness selects for
·Confuses conflict-avoidance with harmony; the absence of fights reads as 'we're fine' rather than 'we're not metabolising'

Triggers

What reliably tips them

Direct emotional asks with an implied timeline

"can we talk tonight?" reliably produces 24h+ silence before response.

Meta-conversation about 'us'

The three DTR-shaped bids (Mar 30, Jun 8, Jul 8) each produced a next-week reply-latency spike of 2.4×.

Being named as the source of a pattern

Jul 20: your "you always evaporate when it gets real" met with a full 41h silence before a lightweight, unrelated reply.

Formation

Likely history — hypothesis, not diagnosis

Read this carefully. The chat cannot see their childhood, therapy, or inner life. This section names the developmental shape that would most plausibly produce the adult patterns observed above. Treat it as a working hypothesis for reflection, never as a story about who they are.

Hypothesis, not diagnosis. The pattern most consistent with the observed behaviour is early experience of an emotional environment where autonomy had to be actively protected — either because closeness came with a cost (over-involvement, guilt, or engulfment) or because reaching didn't reliably work. Nothing in the chat proves this; it's the shape that would produce the adult behaviour.

↳Formative caregiver was inconsistent or emotionally demanding, teaching the nervous system that regulation is safest done alone.
↳Or: formative environment was under-attuned in a way that taught reach doesn't get met, so the reach circuit was turned down rather than kept firing.
↳Either way, the adult adaptation is the same: autonomy = safety, and intimacy is safe only when the exit remains available.

Working with them

What actually lands — and what doesn't

What works

·Name windows, not ambushes — "can we talk Thursday evening about X?" performs dramatically better than "can we talk?"
·Distinguish reach from pressure explicitly — a message that ends with "no reply needed, just wanted you to know" gets 3× the eventual engagement of an open-ended ask.
·Ask for what you want in behavioural terms ("a call every Sunday") rather than in relational terms ("more connection"). The former they can hear; the latter reads as an evaluation.
·Give them the language for their own state — "are you in a heads-down week?" gets a truer answer than "are you okay?"
·When they re-approach after silence, don't punish it and don't skip past it — 'good to hear from you, I noticed you were quiet' names the pattern gently and slowly rewires it.
·Accept that depth here will be episodic rather than continuous. The question is whether the episodic version is enough — not whether you can convert it into the continuous one.

Compatibility risks

·Your need for regular relational check-ins vs their need for undefined solo time — neither is negotiable long-term without cost.
·Your instinct to close distance vs their instinct to open it — the loop in this chat is the trailer for the movie of the relationship if nothing changes.
·Your appetite for meta-conversation vs their appetite to let things settle without naming them — likely to become the recurring fight if this becomes serious.
·Different theories of what 'showing up' means — presence-as-witnessing (you) vs presence-as-usefulness (them). Both real; not interchangeable.

Scientific instruments — text-based estimates

Six validated psychological instruments estimated from this chat: HEXACO-60, ECR-R, SD3 Dark Triad, Gottman Four Horsemen, LIWC-22 linguistic markers, and a Coercive Control screen. Each score is a text-based proxy, not a clinical measurement. Sources cited under every card.

HEXACO-60 · Honesty-Humility

Honesty-Humility: 7.4 / 10

Source: HEXACO-60 Honesty-Humility factor (Ashton & Lee, 2009)·Confidence 74%·Estimated from language and behaviour in this chat. HEXACO is normally self-report; this is a text-based proxy and not a clinical measurement.

Chart

Four facets

Ashton & Lee's four Honesty-Humility facets, scored 0–10 from language and behaviour in this chat.

Reading & trend

The shape is flat — everything hovers around 7.4/10. Nothing dominates, nothing collapses; this is a stable but low-signal picture. Trend: incremental, no single lever will move the whole shape.

Boost it

Don't break it

What this means here

Honesty-Humility reads high. No manipulative framing, no entitlement language, no exploitative asks across 4,127 messages. What limits this from higher isn't dishonesty — it's the pattern of soft-declining rather than saying no outright, which reads as fairness-preserving but sincerity-adjacent.

Absence of manipulation frames

Mar 14, 09:22 — "You're right that I dropped that. I did."

Owns the specific behaviour without guilt-shifting or projection — a Sincerity marker.

No entitlement or status framing

Apr 27 — "I don't think I get to ask for that after this week."

Explicit awareness of relational debt; low on the Greed-Avoidance-adjacent 'I'm owed' register.

Modest self-presentation

Jun 2 — "I'm not handling this as well as I want to be."

Modesty facet: self-image maintained without inflation, even under friction.

ECR-R · Attachment (continuous)

Quadrant: dismissing

Source: Experiences in Close Relationships – Revised (Fraley, Waller & Brennan, 2000)·Confidence 81%·The two ECR-R axes estimated from behavioural markers (proximity-seeking, deactivation, protest, safe-haven use). Not a substitute for the 36-item self-report.

Anxiety axis

2.4/ 7 (Fraley/Waller)

Fear-of-abandonment, reassurance-seeking, protest behaviour.

Avoidance axis

4.9/ 7

Discomfort with closeness, deactivation, preference for independence.

Quadrant read

Low anxiety, moderately high avoidance places them in the Dismissing-Avoidant quadrant (Bartholomew/Fraley two-axis model). They are not distressed by the bond; they are made uncomfortable by demands from the bond. Distinct from Fearful-Avoidant — no push-pull, just steady down-regulation of intimacy signal.

Deactivation under closeness demand

Jul 8 — "Let me get through this week and I'll come back to it." (returned after 6 days)

Classic ECR-R avoidance item: 'I try to avoid getting too close.' Behavioural analogue is latency after intimacy asks.

Low anxiety about the bond

Apr 12 — "I know we're fine, I don't need to keep checking."

Anxiety-axis item reversed: low reassurance-seeking, low fear-of-abandonment language across the corpus.

Discomfort with dependence

Feb 4 — "I just need my week to be my week."

ECR-R avoidance item: 'I prefer not to depend on others' — expressed in the language of scheduling.

SD3 · Short Dark Triad

Dark Triad tendencies (1–5)

Source: SD3 – Short Dark Triad, 27-item (Jones & Paulhus, 2014)·Confidence 72%·Trait tendencies observed in this conversation, not a clinical diagnosis. Elevated SD3 scores in a text sample are not equivalent to clinical antisocial pathology.

Chart

Machiavellianism · Narcissism · Psychopathy

Scored on the SD3 1–5 scale from behavioural markers in the corpus. Elevated scores in text are not equivalent to clinical antisocial pathology.

Reading & trend

Peak is "Narcissism" at 5.2/10; the weakest point is "Psychopathy" at 2.8/10. Mean sits at 4.1/10. The gap (uneven) is the story — the profile is being held up by "Narcissism" while "Psychopathy" is silently pulling it down.

Boost it

Prioritise "Psychopathy" — it's the cheapest lift and the most visible gain. Anchor one concrete daily micro-move to it (a specific bid, question, or acknowledgment). Keep "Narcissism" as-is; don't over-optimise your strength or it flips into a caricature.

Don't break it

Don't lecture the other side about "Psychopathy" — pointing at the low bar creates defensiveness and drops it further. Also don't lean harder on "Narcissism" to compensate; overuse of a strength is the fastest way to unbalance the whole profile.

What this means here

Broadly low. No strategic manipulation (Mach), mild agentic self-focus that reads more as self-protective than grandiose (Narc), and essentially no callousness or impulsive-antisociality signal (Psych). The residual Narcissism score is driven entirely by image-management, not by entitlement or exploitation.

Machiavellianism

No cost-benefit framing of the relationship, no triangulation, no strategic disclosures.

Absence of items like 'It's wise to keep track of information you can use against people later.'

Narcissism

Retraction-and-resend pattern on messages that read 'too raw' — a mild image-management tell, not grandiosity.

Mar 22 — visible edit of a longer message down to "anyway. how are you."

Psychopathy

Zero cruelty, zero contempt, zero threats across corpus. Remorse present after harm (Mar 14, Jul 29).

Jul 29 — "That wasn't fair of me and I know it landed badly. I'm sorry."

Gottman · Four Horsemen + bids

Positive : negative ratio 3.2:1

Source: Gottman Four Horsemen + Sound Relationship House model (Gottman & Silver, 1999; Gottman, 2014)·Confidence 86%·Marker counts extracted deterministically from the chat. Predictive validity of the Four Horsemen is established in observed conflict, not text — treat trends over absolute numbers.

Chart

Four Horsemen — occurrences

Raw counts of each horseman across the corpus. Contempt is the strongest predictor of dissolution — its absence is a positive prognostic sign in its own right.

Reading & trend

Peak is "Defensiveness" at 22; the weakest point is "Contempt" at 0. Mean sits at 11. The gap (sharply uneven) is the story — the profile is being held up by "Defensiveness" while "Contempt" is silently pulling it down.

Boost it

Prioritise "Contempt" — it's the cheapest lift and the most visible gain. Anchor one concrete daily micro-move to it (a specific bid, question, or acknowledgment). Keep "Defensiveness" as-is; don't over-optimise your strength or it flips into a caricature.

Don't break it

Don't lecture the other side about "Contempt" — pointing at the low bar creates defensiveness and drops it further. Also don't lean harder on "Defensiveness" to compensate; overuse of a strength is the fastest way to unbalance the whole profile.

Chart

Bids for connection — response pattern

Of every bid for connection observed, how it was met. Gottman's stable-couple benchmark is 65–70% turn-toward.

Reading & trend

Peak is "Turn toward" at 54%; the weakest point is "Turn against" at 4%. Mean sits at 33%. The gap (sharply uneven) is the story — the profile is being held up by "Turn toward" while "Turn against" is silently pulling it down.

Boost it

Prioritise "Turn against" — it's the cheapest lift and the most visible gain. Anchor one concrete daily micro-move to it (a specific bid, question, or acknowledgment). Keep "Turn toward" as-is; don't over-optimise your strength or it flips into a caricature.

Don't break it

Don't lecture the other side about "Turn against" — pointing at the low bar creates defensiveness and drops it further. Also don't lean harder on "Turn toward" to compensate; overuse of a strength is the fastest way to unbalance the whole profile.

criticism

you · 3.4/1k msgs

Jul 20, 21:04 — "You always evaporate when it gets real."

contempt

both · 0.0/1k msgs

No contempt markers detected across 4,127 messages — the strongest positive signal in this profile.

defensiveness

them · 5.3/1k msgs

Jun 3, 12:14 — "I've been busy, you know that."

stonewalling

them · 2.2/1k msgs

Jul 20 → Jul 22 — 41h silence after being named, followed by an unrelated lightweight reply.

Repair attempts

11 accepted / 18 made(61%)

Repair-attempt acceptance is Gottman's second-strongest stability predictor after low contempt.

Interpretation

Contempt is absent — the single strongest positive prognostic signal Gottman identifies. The concerning pair is Defensiveness (them) responding to Criticism (you), which is exactly the loop the Attachment tab traces. Bids for connection sit slightly below the healthy benchmark of 65–70% turn-toward; the 42% turn-away is not hostile, it's the avoidant register. Positive:negative ratio at 3.2:1 is under Gottman's 5:1 stability threshold — not crisis, but not resilient.

LIWC-22 · Linguistic markers

How they use words

Source: LIWC-22 linguistic markers (Pennebaker, Boyd, Jordan & Blackburn, 2022)·Confidence 83%·Descriptive linguistic summary from the corpus. LIWC categories describe how the person uses language, not who they are.

I / we ratio

4.6

Healthy 1.5–2.5

Analytical thinking

0–100

Authenticity

0–100

Tone (valence)

+0.18

-1 to +1

I-pronoun %

4.1%

Self-focus

We-pronoun %

0.9%

Couple-frame

You-pronoun %

3.6%

Other-directed

Read of the language

High Analytical Thinking (71) with low Authenticity (42) is the classic 'composed presenter' signature — language is drafted and structured rather than spontaneous. I/we ratio of 4.6 is elevated (healthy couples cluster 1.5–2.5), consistent with an individual-frame rather than a couple-frame narrator. Cognitive complexity is high (insight and causation words are frequent), which the Defense-Mechanisms section reads as Intellectualisation. Tone is faintly positive; certainty is moderate — no domineering assertion patterns.

IPIP-NEO 120 · Big Five with 30 facets

Personality profile — domain + facet

Source: IPIP-NEO 120 (Johnson, 2014) — Big Five with 30 facets·Confidence 78%·Text-based Big-Five estimation. Facet-level scores are less stable than domain scores; treat gaps of <10 points as noise.

Chart

Five domains (0–100)

Percentile-style domain scores. Neuroticism low here means low, not inverted — bars are read left-to-right on the same scale.

Reading & trend

Peak is "Conscient." at 74%; the weakest point is "Neuroticism" at 38%. Mean sits at 57%. The gap (uneven) is the story — the profile is being held up by "Conscient." while "Neuroticism" is silently pulling it down.

Boost it

Prioritise "Neuroticism" — it's the cheapest lift and the most visible gain. Anchor one concrete daily micro-move to it (a specific bid, question, or acknowledgment). Keep "Conscient." as-is; don't over-optimise your strength or it flips into a caricature.

Don't break it

Don't lecture the other side about "Neuroticism" — pointing at the low bar creates defensiveness and drops it further. Also don't lean harder on "Conscient." to compensate; overuse of a strength is the fastest way to unbalance the whole profile.

What the shape says

The 30-facet profile confirms the shape the other instruments hint at: Intellect (81) and Achievement (82) are the top drivers; Gregariousness (34), Excitement-seeking (27) and Immoderation (24) are the floor. The gap between Intellect (81) and Sympathy (41) is the most diagnostic split — they will out-analyse feelings faster than they can sit inside them, which is why every repair conversation ends with a plan rather than a landing. Neuroticism domain is genuinely low (38), so the withdrawal pattern is not driven by anxiety — it is a preference structure.

Openness

Imagination72

Metaphor-rich framing, hypothetical scenarios in disclosure asks.

Artistic interests63

References to music/film as emotional shorthand.

Emotionality (open)44

Names emotions in the abstract, rarely in the first person.

Adventurousness58

Comfortable renegotiating plans; low novelty aversion.

Intellect81

High insight & causation vocabulary — matches LIWC analytical=71.

Liberalism66

Non-traditional framing of relational norms.

Conscientiousness

Self-efficacy79

Confident execution on task-level asks.

Orderliness71

Structured message drafts, low typo/edit rate on first send.

Dutifulness76

Follows through on stated commitments (missed 2 of 34 tracked).

Achievement striving82

Work references dominate late-evening windows.

Self-discipline74

Consistent latency floor even during friction weeks.

Cautiousness61

Prefers to draft-and-hold on charged messages.

Extraversion

Friendliness55

Warm register, low first-move rate on greetings.

Gregariousness34

Rarely initiates group-frame topics; individual-frame narrator.

Assertiveness62

Names limits clearly when named; low unsolicited assertion.

Activity level44

Message bursts short; low sustained-conversation rate.

Excitement-seeking27

Low novelty/thrill vocabulary; predictability-preferring.

Cheerfulness46

Positive tone (LIWC +0.18) but restrained expression.

Agreeableness

Trust58

Gives benefit of the doubt on first offense; tightens after 2nd.

Morality (straightforwardness)74

Matches HEXACO Sincerity — no manipulation frames.

Altruism63

Offers logistical support during your hard weeks (3 documented).

Cooperation71

Repair-accepting under low arousal; defensive under high arousal.

Modesty68

Consistent with HEXACO modesty=7.2.

Sympathy41

Empathic acknowledgement is brief, often followed by task-pivot.

Neuroticism

Anxiety32

Low reassurance-seeking, low catastrophizing vocabulary.

Anger29

Zero contempt, minimal irritation markers even under criticism.

Depression44

Occasional flat-affect clusters in Mar & late-Jun windows.

Self-consciousness51

Retract-and-resend pattern; image-management under charged asks.

Immoderation24

No impulsive send patterns; drafted register.

Vulnerability47

Overwhelm-language during workload spikes, not relational friction.

Diagnostic facet highlights

Intellect (Openness)

May 9 — "I think the pattern is that I default to structure when I feel unarmoured."

Meta-cognitive precision on their own mechanism — high Intellect + moderate Self-consciousness.

Sympathy (Agreeableness)

Jul 21 — "That sounds hard. What's the actual bottleneck?"

Empathic acknowledgement, then instant task-pivot — the Sympathy-floor / Intellect-ceiling gap in one message.

Excitement-seeking (Extraversion)

Jun 12 — "I'd rather do the same thing well than a new thing badly."

Predictability-preferring — matches low Gregariousness and high Cautiousness.

Cognitive distortions · Beck / Burns 10-pattern

Thinking-pattern inventory

Source: Cognitive Distortions inventory — Beck (1976) via Burns (1980) 10-pattern·Confidence 70%·Distortions are counted from surface lexical/rhetorical patterns; low-frequency patterns can be noisy. Attribution to 'you' vs 'them' uses the chat's speaker labels.

Chart

Pattern frequency across the corpus

Raw counts per distortion pattern; colour shows attribution (primary = you, gold = them, muted = both).

Reading & trend

Peak is "Jumping to conclusions (mind-reading)" at 14; the weakest point is "Jumping to conclusions (fortune-telling)" at 2. Mean sits at 7. The gap (sharply uneven) is the story — the profile is being held up by "Jumping to conclusions (mind-reading)" while "Jumping to conclusions (fortune-telling)" is silently pulling it down.

Boost it

Prioritise "Jumping to conclusions (fortune-telling)" — it's the cheapest lift and the most visible gain. Anchor one concrete daily micro-move to it (a specific bid, question, or acknowledgment). Keep "Jumping to conclusions (mind-reading)" as-is; don't over-optimise your strength or it flips into a caricature.

Don't break it

Don't lecture the other side about "Jumping to conclusions (fortune-telling)" — pointing at the low bar creates defensiveness and drops it further. Also don't lean harder on "Jumping to conclusions (mind-reading)" to compensate; overuse of a strength is the fastest way to unbalance the whole profile.

Signature of the loop

The dominant pattern is Mind-reading (14 occurrences, both parties), followed by your All-or-nothing and Should statements. This trio is the linguistic signature of an anxious/avoidant loop mid-escalation: you predict their retreat, they hear the prediction as accusation, both parties defend against a future that has not happened yet. Catastrophizing is low (4) — this is not a doom-narrated relationship, it is a mid-conflict cognition problem.

All-or-nothing

you · 2.7/1k

Jul 20, 21:04 — "You always evaporate when it gets real."

Overgeneralization

you · 1.9/1k

Apr 27 — "Every time I need you, you're somewhere else."

Jumping to conclusions (mind-reading)

both · 3.4/1k

Jun 3 — "I know what you're going to say and I already don't want to hear it."

Catastrophizing

you · 1.0/1k

May 30 — "If we can't fix this weekend we can't fix anything."

Emotional reasoning

you · 1.5/1k

Jul 21 — "I feel invisible so I am invisible to you."

Should statements

both · 2.2/1k

Feb 14 — "You should have known that was important."

TAS-20 · Alexithymia (text proxy)

Total 58 / 100 — band: possible

Source: TAS-20 Toronto Alexithymia Scale (Bagby, Parker & Taylor, 1994) — text proxy·Confidence 68%·TAS-20 is a self-report; this is a linguistic proxy from the corpus. Not a clinical diagnosis of alexithymia.

Chart

Three subscales (0–100)

Higher = more alexithymic. The diagnostic pattern is the shape between subscales, not the raw total.

Reading & trend

Peak is "Describing feelings (DDF)" at 71%; the weakest point is "Externally-oriented (EOT)" at 41%. Mean sits at 58%. The gap (uneven) is the story — the profile is being held up by "Describing feelings (DDF)" while "Externally-oriented (EOT)" is silently pulling it down.

Boost it

Prioritise "Externally-oriented (EOT)" — it's the cheapest lift and the most visible gain. Anchor one concrete daily micro-move to it (a specific bid, question, or acknowledgment). Keep "Describing feelings (DDF)" as-is; don't over-optimise your strength or it flips into a caricature.

Don't break it

Don't lecture the other side about "Externally-oriented (EOT)" — pointing at the low bar creates defensiveness and drops it further. Also don't lean harder on "Describing feelings (DDF)" to compensate; overuse of a strength is the fastest way to unbalance the whole profile.

Read of the profile

The tell is Difficulty Describing Feelings (71) sitting well above Difficulty Identifying Feelings (62). They know something is happening internally — the retract-and-resend pattern makes that clear — but the verbal channel for it is narrow. Externally-Oriented Thinking is only mid-range (41), so this is not the 'concrete, no inner world' profile; it is more the 'rich inner life, poor export layer' profile. Reads as sub-clinical rather than clinical alexithymia, but explains why every emotional disclosure lands as a task summary.

DDF: emotion → task translation

Jul 8 — "I feel weird about this. Anyway, what time on Thursday?"

Emotion is named, then discharged into logistics without elaboration — classic DDF marker.

DIF: named-then-retracted feelings

Mar 22 — visible edit of a longer message down to "anyway. how are you."

The retract-and-resend suggests awareness of the affect but low confidence in how to describe it.

Low EOT: internal orientation preserved

May 9 — "I think the pattern is that I default to structure when I feel unarmoured."

When they do describe internal state, precision is high — inconsistent with globally externally-oriented cognition.

Demand-Withdraw · CPQ + Reis PPR

Interaction pattern & perceived responsiveness

Source: Christensen & Sullaway CPQ demand-withdraw (1984) + Reis & Shaver Perceived Partner Responsiveness (1988)·Confidence 79%·Interaction-pattern share estimated from turn-taking around identified conflict topics; PPR is scored from receiver-side language.

Chart

Conflict pattern share (%)

Christensen's demand-withdraw quadrants across tracked conflict episodes. Mutual constructive is the healthy target; asymmetric quadrants drive dissatisfaction.

Reading & trend

Peak is "You demand / them withdraw" at 46%; the weakest point is "Them demand / you withdraw" at 7%. Mean sits at 25%. The gap (uneven) is the story — the profile is being held up by "You demand / them withdraw" while "Them demand / you withdraw" is silently pulling it down.

Boost it

Prioritise "Them demand / you withdraw" — it's the cheapest lift and the most visible gain. Anchor one concrete daily micro-move to it (a specific bid, question, or acknowledgment). Keep "You demand / them withdraw" as-is; don't over-optimise your strength or it flips into a caricature.

Don't break it

Don't lecture the other side about "Them demand / you withdraw" — pointing at the low bar creates defensiveness and drops it further. Also don't lean harder on "You demand / them withdraw" to compensate; overuse of a strength is the fastest way to unbalance the whole profile.

Chart

Perceived Partner Responsiveness (1–7)

Reis three-component PPR from receiver-side language. A tall Caring bar with short Understanding/Validation bars is the classic 'cared-for-but-not-seen' asymmetry.

Reading & trend

Peak is "Caring" at 7.4/10; the weakest point is "Validation" at 4.9/10. Mean sits at 6.0/10. The gap (uneven) is the story — the profile is being held up by "Caring" while "Validation" is silently pulling it down.

Boost it

Prioritise "Validation" — it's the cheapest lift and the most visible gain. Anchor one concrete daily micro-move to it (a specific bid, question, or acknowledgment). Keep "Caring" as-is; don't over-optimise your strength or it flips into a caricature.

Don't break it

Don't lecture the other side about "Validation" — pointing at the low bar creates defensiveness and drops it further. Also don't lean harder on "Caring" to compensate; overuse of a strength is the fastest way to unbalance the whole profile.

What the two together say

The single dominant conflict pattern is you-demand / them-withdraw (46% of tracked conflict episodes) — the exact profile the Attachment tab flags. Mutual constructive is only 28%, which is why conflicts feel unfinished. On PPR, Caring (5.2/7) sits well above Understanding (4.1) and Validation (3.4) — you feel cared about, but not understood or validated in the specific way you are asking to be. That gap is what the whole Two-Way Mirror tab is measuring, from a different angle.

You-demand / them-withdraw

Jul 20, 21:04 → Jul 22 — direct emotional ask followed by 41h silence and a lightweight pivot.

Christensen's textbook DW sequence — high-arousal ask meets deactivation.

PPR: Caring high, Validation low

Jun 3 — "I know you care. I don't feel like you get it."

Explicit user statement of the exact PPR asymmetry the scores show.

Mutual constructive when arousal is low

May 9 exchange — 26-message thread, zero horsemen, 4 accepted repairs.

Capacity for constructive engagement is present — it degrades under arousal, not baseline.

Coercive control screen

Screen score 0.4 / 10 — below threshold

Source: Coercive Control Screen — Stark (2007) markers + Biderman's Chart of Coercion (1957)·Confidence 90%·Screening tool, not a diagnostic instrument. A low score does not mean a relationship is healthy; it means the specific coercive-control markers this screen looks for are not present in this chat.

This chat does not carry the markers of a coercive-controlling dynamic. What it does carry is a repeating anxious/avoidant loop that is costly, not dangerous — different problem, different response.

Monitoring / surveillance

low · 0 obs.

No location-checks, no 'why didn't you reply', no monitoring of other relationships detected.

Absent across corpus.

Degradation

low · 0 obs.

No name-calling, no belittling, no contempt language — reinforces the Gottman contempt=0 finding.

Absent across corpus.

Intermittent reinforcement

low · 3 obs.

Latency dips followed by warm re-approach after silence.

Present but attributable to avoidant regulation rather than strategic reinforcement — treat as attachment pattern, not coercive tactic.

Confidence intervals · methodology

How stable are these scores?

Bootstrap 95% CI over 500 resamples of 120-message rolling windows. Stability = 1 − (CI width / instrument range).

Instrument	Scale	Point	95% CI	Stability
HEXACO Honesty-Humility	0–10	7.4	6.9 – 7.8	91%
ECR-R Anxiety	0–7	2.4	1.9 – 3	84%
ECR-R Avoidance	0–7	4.9	4.5 – 5.3	88%
SD3 Machiavellianism	1–5	2.1	1.8 – 2.4	85%
SD3 Narcissism	1–5	2.6	2.2 – 3	80%
SD3 Psychopathy	1–5	1.4	1.2 – 1.7	87%
IPIP-NEO Neuroticism	0–100	38	32 – 44	88%
IPIP-NEO Conscientiousness	0–100	74	69 – 79	90%
TAS-20 total	0–100	58	51 – 65	86%
PPR Validation	1–7	3.4	2.9 – 3.9	83%
Gottman positive:negative	ratio	3.2	2.7 – 3.7	82%
Coercive Control screen	0–10	0.4	0.2 – 0.9	93%

Stability under 75% (red) means the score moved meaningfully across resampled windows — read it as a range, not a number. Above 85% (green) means the pattern is present throughout the chat, not concentrated in one episode.

Signals

Magic ratio 3.2:1 · healthy ≥ 5:1

Gottman magic ratio

You sit at 3.2:1 — livable but leaking

Target ≥ 5.0 : 1 · you are at 3.2 (64% of stability threshold)

76% positive turns — warmth, humor, validation, curiosity, repair 24% negative turns — criticism, contempt, defensiveness, stonewalling

Flourishing≥ 5 : 1

Livable3 – 5 : 1

Warning1 – 3 : 1

Distress< 1 : 1

Above the neutral line but under the 5:1 stability threshold. The relationship works day-to-day; repair capacity thins under stress.

What it means

At 3.2:1 you are producing about 3 positive turns for every negative one. Gottman's longitudinal work puts the stability threshold at 5:1 in daily life and ~20:1 during conflict. Below 5:1, negatives cost more attention than positives can repay, so small ruptures accumulate faster than they heal.

Boost it

• Send one unprompted appreciation per day — specific, behaviour-named ("thanks for calling to check, that landed").
• Turn toward small bids: reply to the throwaway meme, the "look at this" photo. Bids ignored are the fastest drain on the ratio.
• Use "I noticed / I liked" openers before any complaint — pairs a positive with the negative in the same message.

Don't break it

• No trait-language complaints ("you always / you never") — those count double on the negative side.
• Don't stack negatives: one grievance per message, and never in the opening line after silence.
• Avoid contempt tells (sarcasm, eye-roll emojis, mocking repetition) — a single contempt hit erases ~10 positives.

Chart

Gottman markers — bars vs. clinical threshold (6.0)

Each bar is one Gottman marker scored 0–10. On the four horsemen (harsh startup, criticism, contempt, defensiveness, stonewalling), lower is better; on repair attempts, higher is better. Bars past the dashed 6.0 line on the horsemen are the ones clinicians flag as predictive of dissolution.

Reading & trend

Peak is "Repair attempts" at 6.2/10; the weakest point is "Contempt" at 0.5/10. Mean sits at 3.6/10. The gap (sharply uneven) is the story — the profile is being held up by "Repair attempts" while "Contempt" is silently pulling it down.

Boost it

Prioritise "Contempt" — it's the cheapest lift and the most visible gain. Anchor one concrete daily micro-move to it (a specific bid, question, or acknowledgment). Keep "Repair attempts" as-is; don't over-optimise your strength or it flips into a caricature.

Don't break it

Don't lecture the other side about "Contempt" — pointing at the low bar creates defensiveness and drops it further. Also don't lean harder on "Repair attempts" to compensate; overuse of a strength is the fastest way to unbalance the whole profile.

Chart

Dimension spider — this module at a glance

Spider view of the same six markers. A shape that bulges toward criticism/contempt/defensiveness while repair attempts sinks is the classic distress signature; a shape leaning toward repair with the four horsemen kept low is the resilient signature.

Reading & trend

Boost it

Don't break it

3.0 /10

Harsh startups are rare but present, and they cluster in July when your patience frays. Most of your openings remain soft ("hey — bad time?", "can we talk when you have a sec?").

The few harsh openings that do land tend to shut down the exchange for 24h+, which reinforces avoidance on both sides.

Evidence from the chat

you· Jul 8, 22:19
"forget it. doesn't matter. never does apparently"
After a third unanswered follow-up in a week. Not characteristic of you — but readable as harshness on their end.
you· Jul 27, 19:44
"so is this the thing we do now — you go quiet, I wait, we pretend nothing happened"
Second harsh startup in the same month. Accurate diagnosis, but the trait-language framing ("the thing we do now") lands as attack and gets 34h of silence.
you· Feb 19, 20:12
"hey — bad time? no rush"
Modal soft opener from H1 for contrast. ~92% of your openers across the corpus look like this; the harsh cluster is a July-only local exception, not a baseline.

2.4 /10

The 6 complaint-shaped messages in this chat (yours on Apr 19, May 2, Jun 12, Jul 8; theirs on May 24 and Jun 30) all stay behaviour-specific: "when you cancel last-minute I feel deprioritised", "you didn't reply for two days and I don't know how to read that". Nothing gets escalated to trait language — no "you always", no "you're just like…".

The closest either of you comes to trait-attack is your Jul 8 line ("never does apparently"), and even that generalises the situation rather than their character. In 2,847 messages there is no name-calling, no globalising insult, no diagnosis of the other person as a type.

Evidence from the chat

you· May 2, 20:11
"when you cancel last-minute like that i just feel really deprioritised, that's all"
Textbook behaviour-specific complaint — names the action, names the feeling, doesn't attack character. This is the shape criticism keeps in this chat.
them· May 24, 19:33
"i wish you'd tell me when something's off instead of going quiet — it makes me guess"
Their only real complaint of the year. Behaviour-specific, uses "i" language, no trait attack.
you· Jul 8, 22:41
"you never actually says the hard thing back — never does apparently"
The closest either of you comes to trait language. Even here it generalises the situation, not their character. The line the score is calibrated on.

0.5 /10

Zero instances in the 8-month corpus of mockery, name-calling, disgust-flavoured sarcasm, or the eye-roll emoji used against the other person. The only sarcasm in the chat is self-directed (yours) or affectionate mock-teasing (both sides on shared topics like their taste in movies).

When your Jul 8 message came in as a harsh startup, their reply the next morning did not counter-attack — it was flat and short ("I don't know what to say to that"), not contemptuous. Even the most strained moments of this relationship stayed clean of the specific poison this dimension tracks.

Evidence from the chat

you· Mar 18, 22:04
"your taste in movies is a personal tragedy and i love you anyway"
Affectionate mock-teasing, not contempt — closed with love, target is preference not person. The only "insult"-shaped exchanges in the chat all take this shape.
them· Jul 9, 08:14
"i don't know what to say to that"
Morning-after reply to your harsh Jul 8 startup. Flat, small, non-contemptuous — no counter-attack, no eye-roll register.
you· May 15, 23:47
"i am, as usual, the problem 🫠"
Sarcasm across the whole chat is self-directed. Zero instances of eye-roll or disgust register aimed at the other person in 2,847 messages.

4.1 /10

Their defensive phrases cluster around three topics: replies to your "you seem distant" messages (five "I've just been slammed with work" or "I told you I had that thing" replies between May and July), and reactions to your bids about the future ("we don't need to figure everything out right now", said three separate times). Your own defensive tell is the "I wasn't trying to make you feel…" opener, which appears 4 times, all in June.

The pattern that shows up: yours flares when they read pressure into your bids; theirs flares when you name the pattern of distance. Both defenses are pointed at the same live wire — the ambiguity neither of you has named directly.

Evidence from the chat

them· Jun 30, 21:47
"i've just been slammed with work, you know that"
Their canonical defensive shape — a fact-based deflection that answers the surface question and skips the emotional one underneath.
them· Jul 11, 22:03
"we don't need to figure everything out right now"
Said three times across June–July, always in response to future-shaped bids from you. Defensive against the frame of the question, not its content.
you· Jun 22, 20:11
"i wasn't trying to make you feel bad — that's not what i meant at all"
Your own defensive tell. Fires when their reply reads pressure back into your bid; four instances, all in June.

5.6 /10

The single most concerning signal in this chat. Extended silences (24h+) reliably follow high-affect messages from you — not fights, just intensity or vulnerability.

Stonewalling as a pattern trains the other partner to under-share, which then makes the relationship progressively shallower and easier to walk away from. It's likely unintentional on their side; that doesn't reduce the effect.

Evidence from the chat

them
(31 hours of silence)
Following your message: "you know I'm here right? whatever's going on, I'm not going anywhere."
them
(48 hours of silence)
Following your message naming the June cooling. When they reply, the topic is not addressed.
them· Jul 28
"hey — random question, did you end up finishing that show?"
First message after 34 hours of silence following your Jul 27 harsh startup. Bypasses the actual topic entirely — the classic "pretend the last message didn't happen" re-entry.

6.2 /10

Repair attempts do occur from both sides, though yours outnumber theirs 4:1. Quality of repair is decent — apologies name the behaviour, not just the feeling.

Repair success rate drops sharply after month 5. The moves are still made; they just increasingly bounce off.

Evidence from the chat

them
"sorry — long day. can we talk tonight?"
One of the rare repair openings they initiate. Notably concrete (offers a time), which is why it worked.
you
"I hear you. I was tired too and I let it come out sideways. I'm sorry — that's on me, not you."
you· Aug 1, 09:22
"okay — can we actually talk about what happened last week? I don't want the version where we skip it and it comes back later"
Late-corpus repair opener from you. Landed with "ya let's do that soon" — a canonical H2 repair failure: acknowledged without producing a shift, and "soon" never arrived.

Events

Marker log

2024-05-14Stonewalling · them"you know I'm here right?" — no reply 31h
2024-06-22Withdrawal · them"maybe next week then, if you're free"
2024-07-08Harsh startup · you"forget it, doesn't matter"

Conflict & repair

How disputes moved

Chart

Dimension spider — this module at a glance

Spider of the five conflict metrics. Healthy shapes stay low on escalation speed and grudge carry while pushing repair success and apology quality outward — the ratio of those two halves is the real diagnostic.

Reading & trend

Peak is "Apology quality" at 6.6/10; the weakest point is "Grudge carry" at 3.0/10. Mean sits at 4.6/10. The gap (uneven) is the story — the profile is being held up by "Apology quality" while "Grudge carry" is silently pulling it down.

Boost it

Prioritise "Grudge carry" — it's the cheapest lift and the most visible gain. Anchor one concrete daily micro-move to it (a specific bid, question, or acknowledgment). Keep "Apology quality" as-is; don't over-optimise your strength or it flips into a caricature.

Don't break it

Don't lecture the other side about "Grudge carry" — pointing at the low bar creates defensiveness and drops it further. Also don't lean harder on "Apology quality" to compensate; overuse of a strength is the fastest way to unbalance the whole profile.

3.5 /10

Explicit conflict is infrequent — maybe 3 named incidents across 8 months. That's low, but not necessarily good: relationships this conflict-avoidant often route unresolved issues through cooling and silence instead.

Here, the absence of conflict is doing exactly that work.

Evidence from the chat

you· Apr 19, 22:41
"I DID say that already, on wednesday, you were on your phone"
Incident 1 of 3. Textual escalation, resolved same-night. The only real fight of Q2.
them· Jun 2, 19:33
"seriously?? that's what you're going with??"
Incident 2. Their rare escalation. Absorbed by your softening reply and never re-raised.
you· Jul 27, 23:14
"forget it, doesn't matter"
Incident 3 — and the shape of everything the chat does instead of conflict. Named issue withdrawn, silence follows, cooling continues.

4.2 /10

The three friction moments (Apr 19, Jun 2, Jul 27) each escalate for 3–5 messages then flatten. Peak intensity is textual, not violent: caps in your Apr 19 line "I DID say that already", their multi-punctuation "seriously??" on Jun 2, your Jul 27 "forget it, doesn't matter". No name-calling, no threats, no ultimatums.

In all three cases the de-escalation was carried by you — the Apr 19 thread ended with your "okay let me start over", the Jun 2 thread with your softening emoji reply, the Jul 27 thread never actually de-escalated (they went quiet, you sent a repair opener 4 days later). The escalations don't get big; they also don't get resolved by mutual work.

Evidence from the chat

you· Apr 19, 22:53
"okay let me start over — i'm not trying to fight, i'm frustrated and doing it badly"
Same-night de-escalation, carried by you. Twelve minutes after your caps message. Sets the pattern for the other two incidents.
you· Jun 2, 19:47
"fair. that came out worse than i meant 😅"
Absorbs their "seriously??" push-back with a softening emoji. Ends the escalation in one move — again from your side.
you· Jul 31, 20:11
"hey — sorry for the sunday thing, don't want it to just sit there"
Four-day-delayed repair opener after Jul 27. No de-escalation from their side during the silence; you carry it end-to-end.

5.8 /10

Named repair attempts land ~60% of the time in months 1–4, dropping to ~30% in months 5–8. Falling repair success is the mechanical marker of a relationship losing its immune system.

What's still working: repair language is not blocked or ignored — it's absorbed politely without producing a shift.

Evidence from the chat

you· Feb 27, 21:11
"okay let's actually talk about last night — i don't want it to just fade"
H1 repair opener that landed. Produced a 40-minute exchange that closed the thread. This shape stops working in H2.
you· Jul 11, 22:41
"can we start over — i don't like how last night ended and i don't want it to sit between us"
Identical repair opener 5 months later. Reply arrives 14 hours later, three lines, thread closes without a real conversation. The mechanical drop in success this dimension tracks.
them· Jul 12, 12:41
"yeah we're good. all fine"
Repair absorbed politely, not blocked — but the shift the opener asked for never arrives. Exactly the pattern the analysis names.

3.0 /10

Across the whole 8-month corpus there are zero references from either side to past incidents used as ammunition. The Apr 19 cancellation, the Jun 2 silent treatment, the missed birthday cue on Jul 27 — none of them are re-raised in later arguments, none are list-built into a case, none get the "and remember when you…" treatment.

The only time an old incident resurfaces at all is when you name it as a pattern in your Jul 8 message, and even that stays present-tense ("this thing that keeps happening") rather than tally-shaped. Whatever repair conversation happens next won't have to un-litigate a backlog first.

Evidence from the chat

you· Jul 8, 22:31
"this thing that keeps happening — the quiet after the real conversations — that's what i mean"
The only place an old pattern gets named. Stays present-tense ("keeps happening"), doesn't build a list, doesn't cite dates. Not grudge-shape.
them· Jun 2, 20:12
"okay, sorry — i genuinely didn't mean it that way"
Their apology after the June escalation. Never re-raised by you in any subsequent argument as "and that time you…". Consistent across the corpus.
you· May 30, 19:14
"btw we're totally past the april thing right? just checking, not re-opening"
You explicitly refuse the grudge move even when you check on old material. This is why the dimension scores low — either side could have made this a tally, neither did.

6.6 /10

When apologies happen they are, on average, well-formed — they name the behaviour, take ownership, and (sometimes) propose a small change. Yours are more frequent and slightly higher-quality than theirs.

Quality is not the problem. Availability under mild pressure is.

Evidence from the chat

them
"I hear you — that was unfair of me. I got defensive because it hit a nerve, but that's not on you."
A high-quality repair. Unfortunately one of few from their side.
you· Apr 20, 08:04
"I owe you a real apology for last night — not "sorry if you felt" — sorry that I raised my voice and made it about proving I was right. that wasn't fair."
Names the behaviour, takes ownership, refuses the passive-"if you felt" hedge. Sets the quality bar the rest of the corpus mostly clears.
them· Jul 29, 21:33
"sorry if that came off wrong"
The one clearly lower-quality apology from their side — passive framing, no ownership, no named behaviour. Rare, but shows the shape H2 apologies drift toward when they do appear.

Incidents

Repair log

Chart

Per-incident repair score

Each row is a real dispute in the chat, scored 0–10 on how well it was repaired. Green bars ended in a genuine repair; amber were patched but unresolved; red left residue. Look for a trend, not a single incident — one bad repair is life, a downward trend is a pattern.

Reading & trend

Peak is "Cancelled plans" at 7.2/10; the weakest point is "Missed birthday cue" at 2.0/10. Mean sits at 4.4/10. The gap (sharply uneven) is the story — the profile is being held up by "Cancelled plans" while "Missed birthday cue" is silently pulling it down.

Boost it

Prioritise "Missed birthday cue" — it's the cheapest lift and the most visible gain. Anchor one concrete daily micro-move to it (a specific bid, question, or acknowledgment). Keep "Cancelled plans" as-is; don't over-optimise your strength or it flips into a caricature.

Don't break it

Don't lecture the other side about "Missed birthday cue" — pointing at the low bar creates defensiveness and drops it further. Also don't lean harder on "Cancelled plans" to compensate; overuse of a strength is the fastest way to unbalance the whole profile.

Sexuality

Intimacy register

Flirt & explicit register over time

Chart

Monthly flirt vs. explicit register

Two curves per month: flirt (playful, charged) and explicit (overtly sexual). Healthy couples usually keep flirt elevated even when explicit dips; when both curves fall together and stay flat, the erotic thread is going quiet — that's the signal, not the score.

Reading & trend

Peak is "Mar" at 7.4/10; the weakest point is "Sep" at 1.7/10. Mean sits at 5.0/10. The gap (sharply uneven) is the story — the profile is being held up by "Mar" while "Sep" is silently pulling it down.

Boost it

Prioritise "Sep" — it's the cheapest lift and the most visible gain. Anchor one concrete daily micro-move to it (a specific bid, question, or acknowledgment). Keep "Mar" as-is; don't over-optimise your strength or it flips into a caricature.

Don't break it

Don't lecture the other side about "Sep" — pointing at the low bar creates defensiveness and drops it further. Also don't lean harder on "Mar" to compensate; overuse of a strength is the fastest way to unbalance the whole profile.

Dimensions

Desire, initiation, aftercare

Chart

Dimension spider — this module at a glance

Spider of the six sexuality dimensions. Watch the balance between desire expression / initiation on one side and consent language / aftercare warmth on the other — a shape strong on desire but thin on aftercare is where the register gets brittle.

Reading & trend

Peak is "Consent & comfort language" at 8.0/10; the weakest point is "Sexual initiation balance" at 4.1/10. Mean sits at 6.2/10. The gap (uneven) is the story — the profile is being held up by "Consent & comfort language" while "Sexual initiation balance" is silently pulling it down.

Boost it

Prioritise "Sexual initiation balance" — it's the cheapest lift and the most visible gain. Anchor one concrete daily micro-move to it (a specific bid, question, or acknowledgment). Keep "Consent & comfort language" as-is; don't over-optimise your strength or it flips into a caricature.

Don't break it

Don't lecture the other side about "Sexual initiation balance" — pointing at the low bar creates defensiveness and drops it further. Also don't lean harder on "Consent & comfort language" to compensate; overuse of a strength is the fastest way to unbalance the whole profile.

5.7 /10

Explicit desire language from your side is present and specific — the March "can't stop thinking about the way you…" thread, the two voice-note transcripts in April, the recurring "wish you were here right now" line across months 2–5. When you want them, you say it in language they can't misread.

From their side the register is softer and more oblique: "missing you" instead of "wanting you", "come over" instead of naming what they want to happen. The desire is documented in the chat; the vocabulary for it is asymmetric, which mirrors the disclosure-depth gap in the Communication tab.

Evidence from the chat

you· Apr 3, 23:41
"the way you looked at me on the couch tonight — I'm not going to sleep any time soon"
Sent an hour after they left. Landed with a matched-warmth reply within 4 minutes.
them· May 17, 20:12
"i miss you. come over?"
Typical of their register — invitation-shaped rather than desire-shaped.
you· Mar 11, 00:37
"wish you were here right now — specifically the you-at-2am version, the one that goes quiet and just looks at me"
Names both wanting and the specific quality being wanted. This precision-of-desire register appears 9 times from your side across the corpus and 0 times from theirs.

6.9 /10

Flirt registers stay alive longer than most other emotional textures in this chat. The double-entendre thread that started with the kitchen joke in February is still being callbacked in July; the mock-formal "pursuant to your last message" gets reused as a flirt frame at least monthly through August.

The cooling shows here too, though — flirt initiations drop from ~4 per week in months 1–3 to <1 per week by month 7. When it does happen in H2 it's almost always you opening the frame and them accepting rather than initiating.

Evidence from the chat

them· Jun 4, 22:03
"you're going to be trouble tonight, i can already tell 😏"
One of the last flirt initiations from their side. Callback to a running joke from month 2.
you· Jul 22, 21:19
"pursuant to your last message the court finds you 100% guilty and requests you present yourself for sentencing 😏"
You opening the mock-formal flirt frame in month 7. They accepted ("…what's the sentence 👀") but didn't build. Frame maintained by you; not co-created anymore.
them· Feb 24, 23:41
"if you keep sending stuff like that I'm not going to make it through this meeting tomorrow 🫠"
H1 flirt initiation from them for contrast — spontaneous, escalating, not a callback to a running frame. This shape stops appearing entirely after month 4.

4.1 /10

In months 1–3, sexual/erotic openers split roughly 55/45 (you slightly higher). Starting month 4, that ratio moves to 78/22 and by month 7 essentially every explicit opener is coming from you. There is no month in the second half of the chat where they initiated an erotic thread first.

The pattern maps almost perfectly onto the general initiation-balance curve. What's specific to this dimension: your erotic openers in H2 have started including pre-emptive de-risking ("only if you're up for it", "no pressure") that wasn't there in H1 — the same softening drift you show in your reopening messages generally.

Evidence from the chat

you· Jul 20, 23:14
"thinking about you tonight in a very specific way — only if you're up for it, absolutely no pressure"
Canonical H2 erotic opener from you — includes pre-emptive de-risking that never appeared in H1. The softening drift the score names.
them· Mar 2, 22:41
"come over. now. i'm not asking politely"
H1 initiation from their side — direct, undiluted, unambiguous. This register does not recur once after month 4.
you· Aug 12, 22:47
"if you want to, i want to — and if you don't, ignore this entirely, all good"
Peak de-risking. In H1 the same intent would have been one sentence long. The whole ratio-shift lives inside the difference between these two shapes.

5.4 /10

You have shared four fantasy-shaped disclosures in this chat (Feb 22, Mar 14, Apr 30, Jun 8), each progressively more specific. The first three got engaged replies that built on the frame; the June one got a warm but generic "mmm that sounds hot" that closed the thread rather than opened it further.

From their side there is one — the late-March "can I tell you something I've been thinking about" thread — and it lands as the most sexually intimate exchange in the whole corpus. There is no repeat of that from them across the following 5 months.

Evidence from the chat

you· Jun 8, 23:22
"there's this specific thing i keep coming back to when i think about you — can i tell you"
Fourth and most specific fantasy disclosure. Reply below closed rather than opened the thread.
them· Jun 9, 09:14
"mmm that sounds hot"
Warm but generic — no build, no callback, no question back. The H2 shape of engagement that reads present but doesn't develop the frame.
them· Mar 27, 22:41
"can i tell you something i've been thinking about — since like the second week"
The only fantasy-opener from their side in 8 months. Landed as the most sexually intimate exchange in the corpus. No repeat across the following 5 months.

7.2 /10

The morning-after messages in this chat are one of the most consistently warm patterns across all 8 months. Both sides send them, both sides use the specific pet names that live only in the post-intimacy register, both sides check in about how the other slept and felt.

This is one of the very few dimensions where the H1 → H2 curve doesn't dip. Even in July, when general responsiveness had cratered, the two morning-after threads that month were among the warmest exchanges of the entire cooling phase.

Evidence from the chat

them· Jul 15, 08:47
"still thinking about last night. slept better than i have in weeks. how's your morning ❤️"
One of the warmest replies from them in all of July — a month where the general register was terse and delayed.
you· Jul 15, 09:12
"you were so calm last night, in that specific way I love. slept in your hoodie, don't tell anyone 🫠"
Your matching-warmth reply. Names both the specific quality you loved and a tender detail. The aftercare register carries precision that other H2 registers have lost.
them· Apr 4, 08:22
"okay last night was actually a lot. in a good way. checking on you ❤️"
H1 morning-after from them. Note the identical shape 3 months later on Jul 15 — this specific register genuinely doesn't decay across the corpus.

8.0 /10

Explicit checking-in language is present throughout: the "is this okay?" thread from February that set the pattern, the "tell me if I'm reading this wrong" you use before escalating flirt, their "only if you actually want to" replies. Both sides name comfort and discomfort directly rather than performing.

The two moments where one of you said "not tonight" (yours in April, theirs in June) were both received without pressure and without cooling the register afterward. Whatever else is fragile in this relationship, the ability to name what you do and don't want in this specific dimension is genuinely intact.

Evidence from the chat

you· Feb 14, 22:47
"tell me if i'm reading this wrong — is this the register you want tonight"
The pattern-setter. Explicit check-in before escalating. Establishes a norm both sides use for the next 8 months.
them· Jun 18, 21:33
"honestly not tonight — head is elsewhere. can we tomorrow ❤️"
Direct "no" with a reason and a next-time offer. No cooling of the register in the following exchange. Textbook clean consent shape.
you· Apr 22, 23:11
"not tonight, i'm just tired-tired, not distant-tired — still yours in the morning"
Names the "no", distinguishes it from a pattern-message, and closes warm. This dimension is one of the healthiest things this chat does end-to-end.

Next steps

Recommendations

What to do with this

Name the pattern once

low effort

One direct, non-blaming message: what you observe, what you feel, what you'd like to know.

Stop over-functioning

low effort

Pause initiation for 10 days. Reciprocity data is the answer.

Decide the frame

medium effort

Set an internal deadline. Ambiguity is the cost you're paying.

Dimensions

Where change compounds

Chart

Dimension spider — this module at a glance

Spider of the three change levers. The lever furthest from the benchmark ring is your highest-ROI move — small effort there compounds faster than large effort on the ones already close to benchmark.

Reading & trend

Peak is "Clarity action" at 8.1/10; the weakest point is "Reinvestment ROI" at 3.0/10. Mean sits at 5.8/10. The gap (sharply uneven) is the story — the profile is being held up by "Clarity action" while "Reinvestment ROI" is silently pulling it down.

Boost it

Prioritise "Reinvestment ROI" — it's the cheapest lift and the most visible gain. Anchor one concrete daily micro-move to it (a specific bid, question, or acknowledgment). Keep "Clarity action" as-is; don't over-optimise your strength or it flips into a caricature.

Don't break it

Don't lecture the other side about "Reinvestment ROI" — pointing at the low bar creates defensiveness and drops it further. Also don't lean harder on "Clarity action" to compensate; overuse of a strength is the fastest way to unbalance the whole profile.

8.1 /10

Your own Jul 8 message, minus the frustrated ending, is already 80% of the intervention: "this thing that keeps happening — you go quiet after the real conversations, I chase, we recover surface, we don't recover the thread" is a working draft. Your Jun 12 opener ("can we talk about where this actually is?") is the other half. The language exists inside this chat; it hasn't been sent in one clean sending.

The reason this scores high is that you already have written the sentences — the intervention isn't finding words, it's assembling ones you've already produced without the sighed emojis and self-deprecating softeners that let the other side treat them as mood, not signal.

Evidence from the chat

you· Jul 8, 22:31
"this thing that keeps happening — you go quiet after the real conversations, i chase, we recover surface, we don't recover the thread"
80% of the intervention, already written. Ended in the sentence that turned it into mood ("never does apparently") — the softener is what let them absorb rather than answer.
you· Jun 12, 20:03
"can we talk about where this actually is?"
The other half of the clean opener. Sent alone, without the frustrated tail, this is the exact sentence a clarity conversation would start with.
you· Aug 14, 23:11
"is it weird that we can be this warm in a single exchange and then not talk properly for a week? i keep noticing that"
Third piece of the working draft. Names the pattern with zero attack. Combining these three, minus softeners, is the whole clarity move.

6.4 /10

The two documented moments where you did hold a line — declining the last-minute plan in early May ("can't tonight, need a quiet one"), and not chasing after their Jun 22 vague deferral for four days — both worked: response quality and speed improved for the following exchange. You have the muscle.

What's missing is applying it to your own over-functioning: 84% of July initiations from you, 4:1 repair ratio, remembering their hard dates that they don't remember for you. The boundary that would actually move the picture is not against them — it's against your own default to close every gap the moment it appears.

Evidence from the chat

you· May 6, 20:14
"can't tonight, need a quiet one"
Clean line-hold from you. Their reply the next day was warmer and faster than average — evidence the muscle exists.
them· Jun 26, 19:22
"okay yeah — was starting to worry. how are you"
Landing after four days of you not chasing their Jun 22 vague deferral. Not-closing-the-gap did more than closing it ever has.
you· Jul 4, 21:47
"morning — knew today was going to be heavy for you. thinking of you ❤️"
You remembering their hard date. No equivalent from them on your comparable date two weeks earlier. The boundary that would move the picture is the one against this default.

3.0 /10

The last four months of this chat are a controlled experiment on your part: your word count is up 12% while theirs is down 51%; you added an average of 2 extra initiations per week starting in May; your repair-opener count in Q3 is triple what it was in Q1. None of these increases produced a matching shift in their engagement — the reply-quality curve is flat while your effort curve rises.

This isn't a judgment of the relationship. It's the readout of a specific strategy — more effort, more warmth, more repair — running out of return inside this specific chat. Continuing it is the expensive option; changing it (one clear conversation, then observing what returns without your effort holding it up) is the cheap one.

Evidence from the chat

you· Jul 15, 20:03
"built the weekend around when you're free — friday walk, saturday movie at mine if you want, sunday completely yours"
Peak effort — three plans in one message. Their reply: "maybe sat, will see how the week goes". Effort up, return flat.
you· Aug 3, 21:41
"i've been trying to make it easier for you to reach for me — is that landing, or is it doing the opposite"
You explicitly name the experiment mid-run. Their reply the next morning was one short paragraph that didn't engage the question. The strategy is auditable inside the chat.
them· Aug 4, 09:47
"you're really good to me, honestly. things have just been a lot"
The canonical H2 response to your reinvestment — warm acknowledgement, no matching move. The ROI curve the score reads flat.

Playbook

Slow the cadence, hold the middle-ground, and stop over-functioning — their nervous system opens under space, not pressure.

Confidence 8.1/10Personalised to this dyad · 2026 relational landscape

Every script below is generalised advice grounded in this profile, not a promise about outcome. Adjust wording to your voice; keep the underlying move.

First seven days

Do these, in order

Day 1

Halve your reply cadence to match theirs, silently.

“(no message — mirror their latency for 24h.)”

Why: Breaks the pursuer/distancer loop by removing the top of the cycle.

How we know · Initiation ratio 84/16 (H2) · Attachment: anxious ↔ avoidant loop.

Day 3

Send one warm, specific, low-ask message.

“"Thinking about that thing you said on the walk in April — it landed better the second time."”

Why: Signals presence without demand — the register that historically earns their fullest replies.

How we know · Turn-toward rate 54% overall, but 78% on callback-style openers with no ask (LIWC · Bids).

Day 5-7

If they open a plan, agree once and stop planning.

“"Yes to Saturday. I'll leave the where up to you."”

Why: Returns the executive load to them; interrupts your over-functioning pattern.

How we know · You initiated 41/50 last plans (Conflict · logistics asymmetry).

Communication protocol

Rules of engagement

Message length

≤ 3 sentences per message in the first 3 weeks; batch, don't drip.

Cadence

Match their median latency (~3h) — never faster in H2 reconciliation.

Tone

Warm-neutral. No stacked emojis, no exclamation points on serious content.

Escalate to voice when

Any exchange crosses 6 messages without resolving — voice within 24h or drop it.

Open with

A specific callback to something they said
A concrete observation from your day
A question that has one clear answer

Never open with

"So…"
"Are we okay?"
"Just checking in"
Multiple messages back-to-back

Conflict playbook

When it ignites

Feeling asked to account for their internal state on someone else's timeline.

Step 1
Name the moment
“"I notice this got tight — I want to slow down, not push."”
Step 2
Offer time, not distance
“"Take the evening. I'll be here tomorrow."”
Step 3
Return with one clean sentence
“"Here's what I actually needed to say — one line."”
Step 4
Ask one closed question
“"Is Sunday better for this conversation than tonight?"”
Step 5
Close the loop out loud
“"Okay. Parked. Talking Sunday."”

Do not say

"You always…"
"We need to talk"
Anything at 11pm+
Same point restated three ways

Boundary scripts

Exact words for the moments that recur

Situation

Vague plan drift

“"I'll hold Saturday til Thursday morning — after that I'll make other plans."”

If they push back: "Not a pressure — I just don't want to keep the day open forever."

Situation

Silence after vulnerability

“"I sent something real yesterday. I don't need a full answer — a signal is enough."”

If they push back: "Fair. When you have the words, I'll be here."

Situation

Repeated last-minute cancels

“"Two cancels in a row lands hard. Can we make the next one solid or leave it for a bit?"”

If they push back: "I hear that. I'm not asking why — I'm asking what's realistic."

Situation

Late-night reopens on hard topics

“"I care about this — and I don't do it well after 11. Tomorrow?"”

If they push back: "Yes I know I said that last time. That's the point."

Repair

Apologies, offered and received

When you owe

"I was too fast on that. You were right to push back. I'm sorry."

When they owe

"That one stung. I'm not stacking it — I just want you to know it landed."

Receiving theirs

"Thank you. Received. Moving on." (Do not re-open it.)

Escalation ladder

If the pattern doesn't shift

2 weeks

If you see

Cadence hasn't shifted; initiation still ≥ 75/25 yours.

Then do

One direct, calm sentence naming it. No paragraph.

4 weeks

If you see

Still no reciprocal reach; plans keep dying at their end.

Then do

Have the meta-conversation live (voice/in-person), 20 min max.

8 weeks

If you see

Nothing measurable has changed and you're still tracking metrics.

Then do

Decide whether this shape of relationship is the one you want. Quietly.

Do / Don't

Calibrated to their patterns

• Match their latency (Initiation asymmetry)
• Batch messages (Anxious pursuit spikes)
• End threads cleanly (Loop closure)
• Ask closed questions when tight (Their deactivation)
• Voice for anything > 6 turns (Text escalation)
• State one need at a time (Overload avoidance)
• Honour your own plans (Self-erasure)
• Name repairs out loud (Repair-attempt acceptance)
• Rest between vulnerable sends (Emotional pacing)
• Track what shifts, not what's said (Behaviour over promises)

Don't

• Send a follow-up before 12h (Bid stacking)
• Ask "are we okay" (Reassurance-seeking)
• Re-litigate April (Historical stacking)
• Apologise pre-emptively (Self-erasure)
• Emoji-soften a real ask (Signal muddying)
• Screenshot to a friend mid-thread (External regulation)
• Draft-then-delete more than twice (Image management)
• Reward silence with warmth (Intermittent reinforcement)
• Chase a cancel with a longer plan (Over-functioning)
• Text about it at midnight (Regulation window)

2026 relational landscape

Counter-moves for this era

Breadcrumbing

Let a warm-but-empty message sit 24h before replying. Respond to substance, ignore the sparkle.

Removes the reinforcement schedule that keeps low-effort contact rewarding for them.

How we know · Turn-toward drops to 34% on their sparkle-only pings; you currently reply within 8 min.

Avoidant deactivation

Name the space, don't fill it. "Take the week — I'm not going anywhere."

Their nervous system reads unclaimed space as safety; reads chasing as demand.

How we know · 3 longest re-engagements followed your longest quiets (48h+), not your longest asks.

Situationship drift

Every 6 weeks, one out-loud definition question. Nothing between.

Rhythm > frequency. A predictable cadence for the meta-conversation lowers the stakes each time.

How we know · Prioritisation signal -66% H1→H2 without any single defining conversation.

Phone-first intimacy

Move any emotional beat to voice within 24h. No emotional escalation over text after 10pm.

Text amplifies your intensity and their withdrawal — the wrong medium for this specific pairing.

How we know · 78% of the sharpest exchanges begin after 22:00; 0% of resolved ones do.

If they say

Try this instead

If they say

“"I've been busy, you know that."”

Try this

“"I do. I'm not counting — I just miss you."”

How we know · Defensiveness reliably softens after non-accounting acknowledgements in the corpus.

If they say

“"Maybe next week then, if you're free."”

Try this

“"Pick a day, I'll clear it." (then hold the boundary if they don't)”

How we know · Executive load returns to them; you stop over-functioning.

If they say

“"Let me get through this week."”

Try this

“"Okay. Ping me when you surface."”

How we know · Deactivation script that has historically produced their fullest re-approach (avg 5.4 days).

If they say

“"You're overthinking this."”

Try this

“"Possibly. Also possibly it's real. Both can be true."”

How we know · Blocks the gaslighting-adjacent invalidation without escalating.

If they say

“"I don't know what you want."”

Try this

“"One thing: [name one specific behaviour, once]."”

How we know · Their processing style prefers one concrete input over a general ask.

Your guardrails

Boundaries to keep yourself regulated

No metrics-checking after 10pm — you'll misread everything you see.
One trusted friend, not a group chat, for the hard moments. Rotate topics.
Two evenings a week that have nothing to do with them, on the calendar.
If you draft-and-delete twice, close the app and sleep on it.
Bodywork weekly (walk, swim, gym) — nervous-system regulation isn't optional here.

Cosmic Layer · Addon

“A fire-water pairing running an anxious/avoidant loop wrapped in Lovers-then-Tower symbolism.”

The Cosmic Layer is symbolic vocabulary, not prediction or diagnosis. It reframes patterns already found in your chat through archetype, myth and number — use it as a mirror, not a map.

Synastry

Charts, aspects, and how they echo the chat

You

Leo

element · fire

Them

Scorpio

element · water

Compatibility

6.2/10

symbolic index

Fire meeting deep water: your Leo warmth wants to be seen and answered in kind; their Scorpio water runs deep but rarely surfaces. The pairing is magnetic on contact and eroding at distance — exactly what the chat records.

Sun (Leo) square Moon (Scorpio)

Expressive-self clashes with private-emotional-self. One partner needs the feeling spoken; the other needs it held silently.

Chat echo · Maps onto the disclosure-depth gap: your 4.1 openness score vs. their 2.3. You name feelings out loud; they metabolise them behind a wall.

Venus (Virgo) trine Mars (Cancer)

Care and desire flow easily when both people feel safe — expressed through small acts and quiet tenderness rather than grand gestures.

Chat echo · The March–April flirt cadence (6.9 score) and the callback jokes still alive in July belong here. When the pairing is safe, it is genuinely warm.

Saturn conjunct 7th house cusp

A karmic-tone lesson around commitment: the relationship keeps testing whether both people will show up on time, on purpose.

Chat echo · The presence score (5.4) and the 22 defensiveness events cluster around exactly this question. The chat keeps asking it; neither of you has answered it yet.

Elemental balance

Fire, earth, air, water — who runs what

Your chart runs hot and airy — 76% fire+air — which is why your messages read as expressive and idea-forward. Theirs runs cool and deep — 74% earth+water — which is why replies arrive slower, more grounded, more felt-than-said. Neither is wrong; the chat is a fire-water pair trying to speak two native languages at once.

Chakra profile

Seven energy centres, side by side

The energetic map mirrors the psychometrics almost line by line: the widest chakra gap (Throat, 4.5 points) is the same axis that shows up as disclosure-depth in HEXACO and as avoidance in ECR-R. Symbolic language, same finding.

Root · Muladhara

They anchor logistics; you improvise around them.

Sacral · Svadhisthana

Flirt cadence and creative bids live on your side of the chat.

Solar Plexus · Manipura

Both under-assertive at the exact moments assertion was needed.

Heart · Anahata

You keep the heart channel open even after silence.

Throat · Vishuddha

The single widest gap — disclosure depth 4.1 vs 2.3.

Third Eye · Ajna

They see the pattern earlier; they just don't name it.

Crown · Sahasrara

Meaning-making is present on both sides, quietly.

Cosmic wheel

The 12 houses — where the chat's energy actually landed

The wheel lights up hardest in the 3rd (communication), 7th (partnership) and 12th (shadow) — a textbook shape for a relationship being lived mostly through words about the relationship, with a lot happening just below the surface. The dim 8th (depth) is the room the chat keeps not entering.

1st · Self

6.1

identity work in the chat

2nd · Values

4.2

quietly divergent

3rd · Communication

9.4

dominant house — where everything lives

4th · Home & roots

5.8

referenced, rarely opened

5th · Play & romance

7.2

callback jokes still alive

6th · Daily rhythm

6.7

logistics carry the bond

7th · Partnership

8.6

the karmic hot spot

8th · Depth & intimacy

4.1

under-lit — the missing floor

9th · Meaning

5.5

both curious, rarely together

10th · Public self

3.2

outside the chat's frame

11th · Community

4.8

friend group barely mentioned

12th · Shadow

7.9

the unspoken third participant

Compatibility radar

Six relational axes side by side

Read the shape, not the scores: your polygon leans up and left (expressive/emotional), theirs leans down and right (independent/steady). Complementary on paper, high-friction in practice unless the middle ground is named out loud.

Numerology

Life paths, relationship number, cycle year

Your Life Path

core number

Their Life Path

core number

Relationship #

of the bond

Cycle Year

personal year

A 3–7 pairing under a Relationship Number 1 in a personal year of 4: expression meeting introspection, being asked to lay foundations rather than improvise.

Your Life Path 3 — Expression

You lead with words, colour and warmth. Chat evidence: your verbosity is 1.8× theirs and you initiate 63% of exchanges. This is your gift and, in this pairing, your over-function.

Their Life Path 7 — The Analyst

They lead with observation, silence and inner processing. Chat evidence: their disclosure depth 2.3, their preference for oblique language, the 4.9 avoidance score. They are wired to withdraw before they answer.

Relationship Number 1 — Initiation

The bond keeps asking one of you to start something. The chat records you doing that 8 out of 10 times. A 1-relationship is uncomfortable when only one person carries the initiating current.

Personal Year 4 — Build or dissolve

This is a foundations year, not a fireworks year. The pattern the chat surfaces will either be named and worked on, or it will quietly close by the end of the cycle.

Vibration timeline · month by month

The vibration curve is not prophecy — it's a symbolic overlay on months you already lived. The 9 landing on August is worth noticing because it lines up with a real inflection in the deterministic metrics too (initiation ratio 5.1 → 8.2 in your favor).

Feb · 5Freedom / spark — the chat opens.

Mar · 6Harmony peak — the honeymoon month.

Apr · 6Still 6 — sustained warmth, voice notes.

May · 8Power / stress — first real conflict window.

Jun · 4Foundation test — silences start showing up.

Jul · 7Introspection — you write more, they answer less.

Aug · 9Closure vibration — a decision month by symbol.

Archetypes

The Lover, the Hermit, and the pairing they make

You

The Lover

Warm, expressive, staying in feeling longer than is comfortable.

Recurring 'I want to tell you something' openings, the two long voice-note transcripts in April, and the pattern of holding space for their return after every withdrawal.

Them

The Hermit

Composed, private, retreating into their own tower to think.

Multi-day silences after emotionally-loaded messages, the pattern of resurfacing with logistics instead of the missing reply, and a preference for oblique register over direct disclosure.

Lover ↔ Hermit

A classic asymmetry: one archetype seeks contact to feel real, the other seeks distance to feel real. Neither is wrong. The relationship works when the Hermit builds a bridge back on time and the Lover trusts the bridge will hold — the chat shows the bridge arriving late roughly 4 times out of 5.

Tarot spread · past / present / future

past

The Lovers

The opening months of the chat are pure Lovers energy — recognition, easy warmth, mutual sense-making. This is real; the report is not saying it wasn't.

present

The Tower

A structure is cracking — not from a betrayal but from unaddressed truth. The Tower doesn't punish; it removes what was never load-bearing. The disclosure-depth gap is the crack.

future

Two of Swords

A choice is about to become unavoidable: name the pattern or accept the drift. The blindfold in the card is the ambiguity you're both currently paying for.

Moon journey

Four phases, four prompts — the next lunar cycle

●new

Aug 4

Name the pattern in one sentence — no history, no diagnosis.

◐waxing

Aug 11

Notice which bids get answered and which don't. Data, not verdict.

○full

Aug 19

Release the over-functioning. Say less; let the silence land where it lands.

◑waning

Aug 26

Decide, quietly. The Tower card doesn't require a fight to fall.

Four prompts, one per phase. If the chat's pattern is going to shift, an intention-set → observation → release → decision arc is the shape that best matches what the report is already telling you.

Ritual & Reflection

Something to actually do with all of this

Affirmation

“I am allowed to want the answer in words, and I am allowed to leave rooms that never give it.”

✦ The next new moon in your chart falls in a house of communication. If you're going to name the pattern, that window is generous.

Evening 3-line letter

daily

Before sleep, write three lines to them you will not send: one thing you noticed today, one thing you wanted, one thing you're releasing. This moves the over-functioning current out of the chat and back into you.

Sunday water ritual

weekly

Once a week, run water over your hands for a full minute and picture the anxious loop leaving through your fingers. Symbolic, but the somatic anchor is the point — it teaches your body that release is a choice you can make.

The Tower conversation

one-off

One time, at the next new moon, say the pattern out loud to them in one sentence — no diagnosis, no history, one observation. Whatever comes back is the answer the chat has been asking for.

Get a report like this on your chat.

Start free preview

Share your card

One image. Zero private text.

Anonymized headline + four metrics, ready for IG/TikTok. Nothing from the chat is exposed.

Unlock “How they see me” — free

Invite 2 friends. Get the Mirror.

The $9.99 second-perspective module unlocks automatically when 2 people run their first report through your link.

Try free preview$14.99

Overview

Snapshot

Status: dormant

Message volume, month by month

Reciprocity & presence

Dimension spider — this module at a glance

Trajectory

6-month projection

6-month health forecast with uncertainty band

Momentum & risk

Dimension spider — this module at a glance

Communication

Style

Voice & register

Dimension spider — this module at a glance

How each side sounds

Tone share per side

Emotional

Warmth, validation, labor

Dimension spider — this module at a glance

Affect over time

Weekly sentiment, side by side

Two-Way Mirror

Two-Way Mirror

Perception gap: 4.1 / 10

Perception radar — how each of you sees the same axes

Perception breakdown

Dimension spider — this module at a glance

The perception you don't get to see from inside

Reflected Appraisal Accuracy: 58 / 100

Three overlaid views per axis

Meta-error vs agreement gap per axis

What's open, blind, hidden, unknown

How you land on six impression axes

Six-axis impression radar (0–10)

Aggregate index +2.1 / ±5

Six signal channels (−5 rejecting … +5 valuing)

How stable are these perception estimates?

Attachment

Style pairing

Bonding markers

Dimension spider — this module at a glance

Their Profile

The Composed Retreater

Big Five, evidenced

Big Five trait profile

You vs them, side by side

Relational axes — you vs them

What drives them, what scares them

How they self-protect under pressure

How they use words with you

What they treat as non-negotiable

Cognition, regulation, strengths, blind spots

What reliably tips them

Likely history — hypothesis, not diagnosis

What actually lands — and what doesn't

Honesty-Humility: 7.4 / 10

Four facets

Quadrant: dismissing

Dark Triad tendencies (1–5)

Machiavellianism · Narcissism · Psychopathy

Positive : negative ratio 3.2:1

Four Horsemen — occurrences

Bids for connection — response pattern

How they use words

Personality profile — domain + facet

Five domains (0–100)

Thinking-pattern inventory

Pattern frequency across the corpus

Total 58 / 100 — band: possible

Three subscales (0–100)

Interaction pattern & perceived responsiveness

Conflict pattern share (%)

Perceived Partner Responsiveness (1–7)

Screen score 0.4 / 10 — below threshold

How stable are these scores?

Signals

Gottman signals

Magic ratio 3.2:1 · healthy ≥ 5:1

You sit at 3.2:1 — livable but leaking