What The AI Found I Fed 15 Years of My Private Conversations to an AI and Asked It Who I Am – Part 2

In Part 1, I explained why I decided to run a formal psychological analysis on 850,982 of my own chat messages and 691 blog posts, how I built the analytical framework, and what tools and methods I used. Now: what did it find?

I’ve been sitting with these results for a few days, and I want to be careful about how I present them. This isn’t a humble-brag disguised as self-awareness. This isn’t a redemption arc polished for social media consumption, complete with a three-act structure and a cathartic ending scored by explosions in the sky. Some of what follows is genuinely difficult. Some of it is unflattering. Some of it is the kind of thing that, if my boss or someone from HR stumbled across this blog, I’d want them to read in full context rather than as a bullet point on a Slack thread titled “did you see what Luit posted?”

So here’s the context: I asked an AI to be brutally honest about my behavioral patterns across sixteen years of data. It was. These are the findings I think are worth sharing – the ones that might be useful to someone other than me, the ones that tell a story about growth and pattern recognition and the weird, uncomfortable, occasionally hilarious experience of watching yourself get quantified by a machine that has no reason to be kind about it.

The Seven-Phase Arc

The analysis identified seven distinct developmental phases across the sixteen-year corpus. I’ll give you the compressed version here – each of these could be its own essay, and some of them already are, scattered across this blog’s archives like emotional shrapnel from a person who never learned to process things privately – quod erat demonstrandum. Mea culpa!

Phase 1: The Fabrication (2009–2010)

Let’s start with a statement: I wasn’t isolated. Not in the way that word usually implies. I had friends – a core group of about five, a wider circle of around twenty-five. We met almost weekly. We had fixed spots for hanging out, drinking alcohol, watching movies, playing games, messing around. On paper, my social life looked perfectly normal, maybe even healthy by German countryside standards (biased opinion!)

But:

I had been an outsider since late elementary school. Not the romantic, misunderstood-genius kind of outsider – the kind that gets bullied for years because he acts different, moves different, speaks different, dresses different, has different interests unusual for his age. I never had a strong relationship to any of my post-elementary school classmates. My friend group were fellow outsiders, slightly different themselves – nerds, geeks, metallers, foreigner – drawn together the way mismatched puzzle pieces cluster at the edge of the box. And I never felt connected to my hometown or its surroundings – the dialect, the culture, the traditions, the events, the deeply integrated family-and-relatives social fabric that everyone else seemed to navigate effortlessly and that felt to me like a costume that didn’t fit no matter how I adjusted it. It was alien – or was it me?

As we got older, the divergence sharpened. Some of my friends entered classic relationships and – I’m going to be honest here, even though it sounds harsh – stopped being interesting. There’s no judge. Responsibilities and focuses change; that’s just the way it goes. Others moved away to different parts of Europe. And the majority just… settled. They were normal people, happy with their lives, and there’s nothing wrong with that, except that I wanted something else entirely. I wanted to see the world. I wanted to experience things. I wanted to live, in a way that felt incompatible with the trajectory from private cabin party to private cabin party, from booze-up to booze-up, from conducting the same conversations with the same people in the same places until the heat death of the universe. Some of them even turned their backs on me, the troublesome revolutionary, for wanting more, which is the countryside’s quiet way of punishing ambition that doesn’t fit the well-established template.

So I wasn’t isolated from people. I was isolated from resonance. Surrounded by friends who cared about me but couldn’t meet me where I was, couldn’t talk about what actually mattered to me, and – as I would learn later, painfully – could see my struggles and chose silence because that’s what “traditional men” in rural areas do: they don’t talk about feelings, especially not someone else’s, especially not when those feelings might involve being different.

I had people. I didn’t have my people.

The internet became the escape hatch. Not from loneliness in the traditional sense, but from a specific kind of intellectual, emotional, and sexual starvation that you can absolutely experience while sitting in a room full of friends who love you and don’t understand you and have silently agreed to never mention either fact.

And this is where I did something I’m not proud of. Instead of showing up online as myself – which, given a lifetime of evidence that being myself around people resulted in bullying, nicknames, and social exile, felt like an insane risk – I constructed a comprehensive fake identity. Not a minor embellishment, not a catfish-lite “I’m two years younger and six centimeters taller” situation. A fully fabricated person with a different name, a different age, a different backstory, maintained consistently across dozens of conversation partners for months. The persona was designed, whether I understood it at the time or not, to be everything I believed the real me wasn’t: interesting, compelling, worthy of attention. Worthy of connection. And crucially – designed by someone who had spent his entire life receiving the message that the authentic version wasn’t acceptable. The fake self wasn’t born from absence of people. It was born from a lifetime of evidence that being yourself around people didn’t work.

I operated under this fabrication for months. The analysis classified it clinically as strategic deception – Machiavellianism, in the Dark Triad framework – and it’s hard to argue with that classification. I was systematically deceiving people for social connection. The mitigating context (outsider since childhood, intellectually starved, the fabrication served an attachment function rather than an exploitation function) doesn’t erase the behavior. It frames it. And that framing matters, as you’ll see in Pattern 4 below, but the behavior itself was what it was: I lied to people. Many people. Consistently and elaborately. I was damn good at it.

Phase 2: The Collapse (April 2010)

The fabrication didn’t end because I chose honesty. That would make a better story, but the data doesn’t support it. It ended because someone investigated, found my real identity, and confronted me. The constructed self died – not on my terms, not gracefully, not as part of some planned transition toward authenticity. It got ripped away.

My already depressed and messed-up self had reached its breaking point. And I tried to follow it. With a deliberate suicide attempt that was interrupted by friends who found me in time.

The crisis that followed was the most psychologically dense period in the entire corpus. I started therapy. I migrated to new platforms under my real name – my actual, boring, non-fabricated name. I began, for the first time in roughly a decade, building relationships on an authentic foundation, which turns out to be terrifying when you’ve spent years hiding behind a character you invented specifically because you thought the real version wasn’t good enough. I wrote about it, obliquely, in posts like Inneres Chaos and Normalität and Die Lehre von der Leere.

There’s one more thing about this period that I need to add, because it shaped everything that followed and because it says something about the environment I came from that I still haven’t fully forgiven. After my suicide attempt, I came out as gay to my local friends for the first time. Their reactions were – and this is both the good and the devastating part – universally positive. They told me they had always thought I was at least different, if not gay. They’d known. Or suspected. For years. They had watched my identity struggles, my pain, my inner fights, my increasingly visible crisis, and they had said nothing. Not because they didn’t care. Because that’s what some – not all – people do in rural alreas – from an early age, they are taught not to talk about emotions, and especially not someone else’s sexuality, and especially not when acknowledging it would require having a conversation that nobody had taught them how to have (and yes, this outdated social habit is still around in 2026). They loved me and they failed me, simultaneously, and the worst part is that they probably don’t even know the second half of that sentence. This – all of this, the silence, the unspoken knowledge, the structural inability of an entire social environment to say the obvious thing out loud – is why, until today, I carry a deep discomfort with German countryside culture that I know is subjective and biased and that I hold anyway, because some biases are earned.

Phase 3: Social Expansion (2010–2012)

The post-collapse period was a rapid social blossoming – almost aggressively so, like a plant that had been kept in a dark room for a decade and was now photosynthesizing with the desperate intensity of someone who knows the light could disappear again. My conversation partner count grew from 48 to 112, the peak social breadth in the entire corpus. I found a caregiving identity: writing people’s job applications, advising on mental health, guiding friends through sexual orientation questioning, becoming the person people came to when things got complicated. My outgoing ratio crossed 50% for the first time. I went from someone who mostly listened to someone who actively shaped conversations, held space, organized, supported. The helper. The fixer.

(The analysis has things to say about this role. They’re in Pattern 3. They’re not entirely flattering.)

Phase 4: First “real” Love (2012–2013)

A person I’ll call Q from Frankfurt entered my life in April 2012, and everything reorganized around him. He was 17, I was 23. I know how that reads, and I’m not going to pretend the age gap doesn’t warrant a pause – it does, and it did at the time too. In Germany, where this happened, the age of consent is 14 (which is its own conversation), so there was no legal issue, but legality and emotional maturity are different things, and I was acutely aware of that asymmetry even then. The relationship was real, mutual, and formative for both of us. It was also the kind of intense first real love that, in retrospect, carried emotional weight that neither of us was fully equipped to handle (“real”, because there was another pre-alpha-version-relationship-approach before). Context matters here, and I’d rather give it than let the reader fill in blanks with assumptions.

What the data shows: 226 messages per day sustained over 85 days. Two hundred and twenty-six. Per day. That’s a message every 6.4 minutes during waking hours, which is either deeply romantic or clinically concerning, depending on which framework you apply. (The analysis applied both).

What the analysis initially interpreted as pure romantic intensity was recontextualized during the Q&A phase: a shared close friend died by suicide during this exact period, and Q and I were processing that trauma together while simultaneously falling in love. Grief and new love, layered on top of each other, producing communication volumes that made the data spike in ways that told a story the numbers alone couldn’t accurately narrate.

The relationship drove my first successful follow-through on years of stalled plans – I actually moved cities, got a new job, built a life outside my usual orbit. Not for myself, initially. For him. Which is both beautiful and slightly pathological, and the analysis noted both of those things without prioritizing either.

Phase 5: Collapse Redux (2014)

Q ended the relationship via chat message on the day we were supposed to move into our shared apartment. I was already there, setting it up. Arranging furniture for two people, one of whom had just decided to be one.

What followed was a psychiatric admission, a diagnosis that would reframe much of the preceding analysis (more on this in “The Diagnosis Question” below), and a slow, non-linear recovery documented in posts like the one I wrote from a hospital bed. Non-linear is doing heavy lifting in that sentence, by the way. Recovery wasn’t a U-shape. It was more like one of those stock charts from 2008 – technically trending upward if you zoom out far enough, but absolutely brutal at any given resolution.

Phase 6: Dissolution and Reconstruction (2014–2018)

Years of recovery, new relationships, career rebuilding, and eventually a radical act: dissolving my entire Frankfurt existence. Giving away possessions, ending the lease, symbolically burning down the identity I’d built there. It looked like self-sabotage, and the analysis flagged it as such. But it was also preparation for something I couldn’t yet name – a clearing of space that only makes sense in retrospect, the way controlled demolition only looks intentional if you know about the blueprint for what comes next.

My sexual identity, which had been on a long and winding exploratory journey – gay in 2010, bisexual around 2014, asexual around 2016 (a period my friends found confusing and I found exhausting to explain) – arrived at a formulation that finally fit: panromantic-demisexual. Attracted to people of all genders, but requiring deep emotional connection before sexual desire activates, at least within the context of a romantic relationship. Like a PC with a really, really long boot-up phase (exceptions prove the rule). This single identification retrospectively explained patterns that had puzzled both me and the analysis: the emotionally intimate but non-sexual bonds with women (a facet that underwent an intense shift in subsequent years), the delayed sexual activation in relationships, the intensity of emotional connection that preceded any physical dimension. Turns out I wasn’t confused. I was just running on a different operating system than everyone – and I – assumed.

Phase 7: Vietnam (2018–Present)

A work trip during Tết 2018 changed everything. The “omnipräsente Leere” – the omnipresent emptiness that had been my inner weather for as long as I could remember, the low-level emotional static that was so constant I’d stopped noticing it the way you stop noticing the hum of a refrigerator – started to fill. Not immediately. Not dramatically. But persistently, in a way that felt different from every previous attempt at feeling okay.

I emigrated in late 2019. I’m still here, in Phú Mỹ Hưng, Sài Gòn, working as a senior engineer & AI orchestrator, living with Jayden, supporting his fashion brands on the side, and writing blog posts about feeding my old chat logs to AI. Life is absurd and good. Those two things aren’t contradictions. They might even be prerequisites for each other.

The Patterns That Matter

The seven phases are the narrative. But narrative is what I’ve always been good at – shaping experience into story, finding the arc, making it mean something, packaging messy reality into clean prose with a beginning, a middle, and a closing paragraph that lands on an emotional note (you’re reading the apparatus right now). The analysis was designed to cut beneath the narrative and find the structural patterns – the recurring behavioral mechanics that persist regardless of which chapter of the story they show up in. Here are the ones that hit hardest:

Pattern 1: Preoccupied Attachment with Serial Exclusivity

This is the strongest finding in the entire analysis, and the one with the most precise quantitative evidence. It’s also the one that made me stare at a spreadsheet for twenty minutes without blinking, so buckle up.

I form intense bonds fast. Really fast. Embarrassingly, diagnostically fast. Declarations of love within days of meeting someone, message volumes of 150 to 230 per day, the complete reorganization of my social life around a single person – schedule, priorities, emotional bandwidth, all of it collapsed into one relationship like a gravitational singularity with a chat window. And when a new primary attachment forms, the old one doesn’t gradually fade. It drops off a cliff. The analysis tracked one friendship going from 11,629 messages in a single quarter to 125 in the next. A 98.9% decline. Not because of a fight. Not because of a betrayal. Because my attention had moved to someone new, and apparently my attention is a zero-sum resource that can only be pointed in one direction at a time.

That number – 11,629 to 125 – is the one I mentioned in Part 1. The one I said would be hard to forget. It represents a real person. Someone who went from being the center of my social universe to being a footnote in ninety days, through no fault of their own. The data doesn’t tell me how they experienced that transition. I suspect it wasn’t great.

In attachment theory (Bartholomew & Horowitz), this maps to a preoccupied/anxious pattern: hyperactivation of the attachment system, fear of abandonment expressed through verbal flooding, relationship as the organizing principle of life. I relocated to an entirely different city because a partner lived there. I dissolved years of stalled plans overnight because someone gave me a reason to. The pattern is consistent, it’s quantified, and it’s uncomfortable to see laid out in a table with quarterly breakdowns.

The good news: this pattern has evolved. The ratio data within the Q relationship showed self-correction – from 1.67 (pursuit phase, I’m talking nearly twice as much as he is, which is the numerical equivalent of being that person) to 1.00 (equilibrium, after we actually met in person). I modulated. I learned. The current relationship with Jayden, by my own account and by every observable indicator, is the most balanced and mutually self-sufficient I’ve ever had. The serial intensity transfer pattern appears to be gone – or at least fundamentally different in character. The analysis calls this “movement toward earned secure attachment.” I’ll take it. Gladly. With both hands and a slightly embarrassing amount of relief.

Pattern 2: The Insight–Action Gap

I consistently recognize my patterns while they’re happening. The chat logs are full of me articulating exactly what I’m doing wrong, in real time, to the people I’m doing it to:

• “Ich verliebe mich oft viel zu schnell” (I fall in love way too fast).
• “Es ist fast wie eine Sucht” (it’s almost like an addiction).
• “Wenn ich was aufgebaut hab, was mir wirklich wertvoll ist, dann mach ich es zunichte” (when I’ve built something that’s really valuable to me, I destroy it).
• “Ich weiß dass ich das tue, und ich tu’s trotzdem” (I know I’m doing this, and I do it anyway).

High metacognitive insight. Excellent self-analysis. Articulate, precise, self-aware. And then I do the thing anyway. Every. Single. Time.

The analysis flagged this as a core feature rather than a bug – which is a sentence that sounds optimistic until you understand what it means. The gap isn’t between knowing and not-knowing. It’s between the analytical system (which understands the pattern perfectly, can name it, can even predict its consequences in advance) and the attachment system (which overrides analytical conclusions the moment it gets activated, like a fire alarm that doesn’t care how calmly you’ve assessed the situation). Thinking my way out of an emotional pattern has never worked for me – not because I lack the thinking, but because the thinking isn’t the rate-limiting factor. The thinking is excellent.

What finally narrowed the gap wasn’t more insight. (I had plenty. I had so many insight, they were coming out of my ears. I could have written a dissertation on my own dysfunction while actively dysfunctioning). It was accumulated experience, processed trauma, and eventually – the creation of life conditions (Sài Gòn, stable career, Jayden) that support rather than trigger the attachment system. The Vietnam move was the first major life change I initiated proactively rather than in response to a crisis or a person. Not “I’m moving because someone lives there.” Just “I’m moving because this place makes me feel alive.” That’s the evidence of the gap closing. Not the insight. The action.

Pattern 3: Caregiving as Sublimation

I take care of people. It’s one of the most consistent behaviors across the entire sixteen-year span: writing job applications for friends, advising on mental health, guiding people through coming-out processes, holding space for grief, supporting a friend through reporting sexual violence in Vietnam’s legal system. If you’ve been in my life for more than a few months, there’s a reasonable chance I’ve helped you with something that wasn’t my problem to solve. You’re welcome. Also, we need to talk about why I do this.

The analysis identified caregiving as operating on two levels simultaneously, which is the polite academic way of saying “this behavior is both genuine and a coping mechanism, and good luck separating the two.” Level one: it’s genuine. I’m good at emotional support, I invest real time and energy, and the people I help generally benefit. Level two: it’s also a defense mechanism. By focusing on other people’s problems, I maintain a sense of competence and purpose while potentially deferring my own unresolved material indefinitely. The technical term is sublimation – channeling difficult emotions into socially valued behavior. It’s the most socially acceptable defense mechanism in the hierarchy, which makes it the hardest one to identify, because everyone around you is thanking you for it.

The most extreme expression of this pattern: after my close friend died by suicide in 2012, I assumed the caregiver role for approximately fifteen of his friends. Daily catch-ups. Collective mourning. Emotional triage for a group of grieving people, most of whom were falling apart in different ways and at different speeds. While simultaneously being in the honeymoon phase of the new relationship with Q. I was 23, processing my own trauma, and holding everyone else’s grief, and at no point did anyone – including me – think to ask whether the person running the support group might also need support. The subsequent burnout – and the psychiatric admission two years later – was, as the analysis puts it with characteristic clinical understatement, “overdetermined rather than surprising.”

I still do this. The difference is awareness and boundaries – both of which I’ve learned slowly, expensively, and mostly through the consequences of not having them. The caregiving pattern is part of who I am, and I don’t want to lose it. But there’s a difference between holding space for someone and hollowing yourself out to make room, and that difference took me the better part of a decade to learn.

Pattern 4: The Fabrication–Attachment Nexus

The analysis’s most structurally interesting finding, and the one that reframed how I think about the worst thing I’ve ever done: the fake identity I maintained at 21 wasn’t primarily about deception. It was an attachment strategy.

Think about it from the inside – and I’m asking you to do something uncomfortable here, which is to empathize with a person who was actively deceiving dozens of people, but stay with me for a (long) paragraph. As I already said, I’ve spent great parts of my childhood and adolescence as an outsider. Bullied for years. Different in ways I couldn’t hide and couldn’t explain. Surrounded by people who cared about me but couldn’t reach me, who saw my pain and chose the rural default of silence. I’m not isolated from humans – I’m isolated from understanding. And I’m desperate for connection, real connection, the kind where someone actually gets it. The fabricated persona solves this problem perfectly. I can be “open” and “vulnerable” about “my life” because nothing I’m sharing is actually mine. Maximum intimacy, zero genuine risk. And crucially – no chance of the rejection I’d learned to expect when presenting the real version. The persona was optimized not for exploitation but for connection: designed to be interesting enough to attract, vulnerable enough to bond with, compelling enough to hold attention. It was, in the Dark Triad framework, Machiavellian – but the goal wasn’t power. It was love. Or something close enough to love that a lonely 21-year-old who’d never been told his feelings were worth talking about couldn’t tell the difference.

When the fabrication collapsed, the attachment system was exposed to real risk for the first time. No persona to hide behind, no character to sacrifice instead of yourself. And here’s the detail the analysis found significant – structurally significant, not just narratively satisfying: the first person I trusted with my real self was the person who’d exposed the lie. The person who saw through the mask became the first person trusted with the face underneath. That’s not a coincidence the analysis was willing to dismiss. It suggests that the exposure, traumatic as it was, created a paradoxical foundation of trust: this person already knows the worst. There is nothing left to hide. This person became my previously mentioned pre-alpha-version-relationship.

This reframing doesn’t excuse the behavior. I want to be clear about that, because I’ve spent years being clear about it and I’m not stopping now: I deceived people. I maintained a systematic lie across dozens of digital relationships. I caused real emotional damage to real people who trusted someone who didn’t exist. The analysis calls it what it is: Machiavellian. But the function wasn’t predatory – it was protective. An elaborate, immature defense mechanism deployed by someone who didn’t have healthier tools yet. Understanding why I did it doesn’t undo that I did it. But it makes the “never again” more structurally sound, because I can address the underlying need (connection, attachment, being seen) without the destructive method (fabrication, deception, strategic performance of a false self).

Pattern 5: Private VS Public Divergence

Here’s a finding that’s awkward to share on a blog: the blog amplifies.

When chat logs and blog posts cover the same period, a consistent pattern emerges: the blog version is more dramatic, more literary, more crisis-shaped than what the private data shows. The April 2010 collapse appears on this blog as an existential turning point, a death and rebirth, a narrative inflection with the dramatic weight of a novel’s climax. In the chat logs, it’s a messy week of confrontation, confession, confusion, and gradual rebuilding – not a single transformative moment but a slow, unglamorous process that happened to contain some very dramatic hours. The 2014 psychiatric admission is depicted here with visceral literary intensity. The day-to-day experience, based on the surrounding data, was probably more mundane. More waiting rooms and bad coffee than existential reckoning.

The analysis doesn’t call this dishonesty. It calls it meaning-making through narrative construction – consistent with high openness, high verbal ability, and the need to transform raw experience into coherent story. This blog has always been a meaning-making apparatus. A machine for converting messy, contradictory life into prose that has an arc and a point. The risk – and the analysis flagged this explicitly – is that the constructed narrative becomes the remembered reality, and the actual developmental process (which was more gradual, more boring, more full of days where nothing happened and nothing changed) gets overwritten by the story I told about it. And, on a side note, the analysis also showed that I often withheld personal events/thoughts from people I knew but published them for people I didn’t.

Et voilá – I’m writing about this on the blog that does the amplifying. Using the meaning-making apparatus to describe the meaning-making apparatus’s limitations. The irony is not lost on me. It’s practically waving at me from across the digital realm.

The Diagnosis Question

In 2014, a psychiatric facility in Frankfurt produced a discharge report with findings that recontextualized the entire analysis. I’ll share what’s relevant and what I’m comfortable making public. (The boundary between those two categories was one of the harder editorial decisions in writing this post):

• Asperger syndrome.
  Diagnosed. And looking at the longitudinal data through that lens, it fits like a key in a lock you didn’t know existed: the intensity of monofocal relationship investment, the extreme contrast between professional communication (near-silence with supervisors, minimal small talk, the kind of workplace presence that reads as “heeeyyyy” when it’s actually “I don’t have the script for this interaction“) and personal communication (hundreds of messages daily with close connections, emotional depth that goes to eleven immediately), the special interests pursued with unusual depth and systematization, the ritualistic language patterns, the systemizing approach to everything including – and I realize the recursion here is almost too perfect – my own psychology. The social skills are there. I can code-switch, read rooms, deploy humor appropriately, navigate professional contexts without incident. But they were developed skills, not intuitive ones. Built through observation, copying, practice, trial & error, and – the analysis suggests this with a level of confidence that made me uncomfortable – initially through the fabrication era, where I essentially learned social performance by performing a character. “Masking,” in the current autism discourse. Except my mask had a different name and a different birthday.
• Depression, recurrent.
  Documented across the timeline: the decade of emotional isolation before the chat logs, the 2010 suicide attempt, the chronic emptiness I wrote about in post after post after post (the “omnipräsente Leere” that shows up in my German writing like a recurring character who won’t leave the stage), the 2014 psychiatric admission. The blog’s emigration-era writing describes fifteen-plus years of depression. Current status: what I’d describe as substantial remission. I feel stable and peaceful in a way that would have been unimaginable to the person in those chat logs. Not the absence of difficulty – but the absence of that background static, that hum that had been running so long I forgot it was there until it stopped.
• Clinical assessment also surfaced patterns I had to sit with.
  Things about how I’d related to power, attention, and recognition. Things that were difficult to read about myself, especially in the context of the fabrication era, where the desire for attention and the willingness to deceive to obtain it formed a pattern the analysis did not look away from. These patterns were assessed as sub-threshold – present but not at diagnostic levels – and they’ve diminished substantially with time and distance. I’m not going to detail the clinical specifics here, because clinical specifics without clinical context are weapons, and I’d rather talk about what I’ve done with the awareness than give anyone a checklist to hold against me. Or worse, against themselves.

What I will say is this: sitting with an unflinching assessment of your own shadow material is one of the most uncomfortable and useful things I’ve ever done. The analysis didn’t tell me anything I hadn’t, on some level, already suspected. But suspecting and seeing it documented with evidence – timestamped, quantified, cross-referenced across dimensions – are very different experiences. Suspicion lets you maintain plausible deniability with yourself. Evidence doesn’t.

What Changed (The Growth Part)

The analysis tracks change using Prochaska’s Transtheoretical Model – the “stages of change” framework – and maps my trajectory with what it calls “unusual clarity,” which I’m choosing to interpret as a compliment rather than a comment on how textbook-predictable my psychological development has been:

• Precontemplation (before 2010):
  No awareness that the intellectual isolation, the fabrication, the patterns are problems. The fabricated identity feels like a solution, not a symptom. (This is the stage where, if someone had told me I was doing something destructive, I would have argued with them. Convincingly. Because I was very good at arguing and very bad at self-awareness, which is a dangerous combination).
• Contemplation (early 2010):
  First confession to someone. Entering therapy. But the fabrication continues with others – awareness of the problem coexists with continued engagement in the behavior. Knowing and doing remain separate systems. (See: Pattern 2).
• Preparation (2010–2012):
  Real name, real platforms, real relationships. Elaborate plans for life changes. But execution lags – years of planning without follow-through. The gap between insight and action at its widest.
• Action (2012–2013):
  A relationship catalyzes concrete behavioral change. I move cities. I get a new job. I follow through. For the first time, the plans actually become actions. (That it took another person to catalyze this is both romantic and, per the analysis, “consistent with the externally-referenced motivation pattern.” Translation: I did it for someone else before I learned to do it for myself).
• Relapse (2014):
  Breakup triggers psychiatric crisis. Return to destructive patterns. But – and the analysis flags this as significant – the relapse doesn’t regress to the fabrication level. I fall back to crisis, but not to deception. The floor is higher than it was. The worst version of 2014-me is still better than the average version of 2009-me. That’s progress, even if it didn’t feel like it at the time.
• Renewed action (2015–2018):
  Recovery. New relationships. Career stabilization. Identity consolidation. Preparation for emigration – this time, not running from something but toward something, which turns out to be a fundamentally different kind of movement even when the suitcase looks the same.
• Maintenance (2018–present):
  Sài Gòn. Sustained professional role. Integrated identity. Settled relationship. The longest period of psychological equilibrium in the documented history. Eight years and counting, which doesn’t sound like much unless you’ve read the preceding phases and understand what the baseline used to look like.

The post-traumatic growth indicators are visible across every dimension: increased emotional vocabulary (compare my 2009 chat messages to my 2023 blog prose – the difference isn’t just linguistic maturity, it’s the ability to name what I’m feeling with enough specificity that the naming itself becomes useful), shift from external validation-seeking to internal self-definition, development of repair behaviors in relationships (I can apologize now, genuinely, without it feeling like a structural collapse), increased perspective-taking, and – the one that makes me smile – evolution of humor from defensive to connective. My 2009 humor deflected. My 2026 humor invites. That might be the single best metric of growth in the entire analysis, and it’s one no clinical framework was designed to capture.

What I’m Still Sitting With

The analysis produced a formal recommendations section, and some of its suggestions landed hard enough that I’m still feeling the impact:

• Watch the caregiving pattern.
  It’s durable, it’s valuable, and it can destroy me if I let it run unchecked. The 2012 version – managing fifteen people’s grief while processing my own trauma – nearly killed me, on a two-year delay. The lesson isn’t “stop caring.” The lesson is “sustainable support, not self-depleting over-extension.” I’m better at this than I was. I’m not as good at it as I’d like to be. (See: gap between insight and action. See: everything).
• Honor the integration.
  My identity development was unusually long and exploratory – the analysis’s polite way of saying “you spent sixteen years trying on identities like someone at a fitting room with no mirror.” The analysis calls the current result – panromantic-demisexual, bicultural German-Vietnamese, software engineer, blog writer, all of it – a “stable integration that accommodates complexity.” The recommendation isn’t to keep searching for the next reinvention. It’s to build on what’s been built. (This feels right. For the first time in my life, it feels right. And the fact that I can write that sentence without immediately qualifying it with doubt or irony is itself a data point).
• And one more thing that the analysis didn’t explicitly recommend, but that the process itself demonstrated:

   
  The stories we tell about ourselves matter, but they’re not the whole truth.
  This blog has been my meaning-making apparatus for almost twenty years. It’s valuable. It’s how I’ve processed grief, joy, confusion, growth, boredom, love, sex, Sài Gòn, and the general chaos of being a person who thinks too much about everything. But the patterns underneath the narratives – the ones you can only see with enough data and enough distance and a machine that doesn’t care about your literary framing – are a different kind of truth. Slower, less dramatic, less shaped. Messier, more contradictory, more human.

I’m glad I looked.

A Final Note on Method

If you’ve read this far and thought “I want to try this” – you can. Everyone can. The chat log archives are the hard part; most people don’t have years of Trillian logs lying around (and if you do, congratulations on your data hoarding, we should start a support group). But you’d be surprised what you do have: WhatsApp exports, email archives, old Facebook data downloads, years of Instagram DMs, Twitter threads, Tinder flirts, years of social media posts. The data exists. The frameworks are published. The AI tools are available.

What you might not be ready for is what they find. Not the big dramatic revelations – those you probably already know. The small quantitative truths: how your communication patterns shift when you’re anxious, who you stop talking to when someone new arrives, the gap between what you say you want and what your behavior actually pursues. The spreadsheet version of your personality, stripped of narrative and laid out in numbers that don’t care about your self-image.

That’s the uncomfortable part. And the useful one. What a ride!

Disclosure

This analysis was conducted entirely by Claude Opus 4.6, an AI model by Anthropic. All psychological frameworks were applied by the AI based on a purpose-built analytical rubric – not by a licensed therapist, psychologist, or psychiatrist.

AI-generated analysis is not a substitute for professional mental health assessment. The findings described in this series are observational patterns produced by a language model processing text. They should be understood as one perspective – systematic and thorough, but limited by the inherent constraints of AI-driven text analysis. Claude, by default, can’t read body language, hear tone of voice, observe microexpressions, or pick up on the thousand nonverbal signals a human clinician uses. It reads words. Very many words, very systematically, but words only.

AI can make mistakes, misinterpret context, and miss nuances that a human clinician would catch. If any of this resonates with your own experience – the attachment patterns, the insight–action gap, the caregiving, the fabrication, the feeling of watching yourself do something you know isn’t working and being unable to stop – please talk to an actual human professional. They’re better at this than any AI, including the very impressive one that helped me here.

The data told me a story. But the story isn’t finished, and the next chapters should probably be co-authored with someone who has a license and an office with a comfortable couch. And coffee, black!

Hero image: Selfie taken by me on Jan 1, 2026. It was visually extended on March 16, 2026, at 12:10 PM with the help of google/nano-banana-pro.