i let a robot read my diary again to see if i’d grown, 2 years later.

So you may remember, against my own explicitly stated better judgement, I once fed five years of handwritten journals to a language model and published its findings here like a normal well-adjusted person. That post did numbers by my standards, mostly, I suspect, because there is nothing the internet loves more than watching someone volunteer for their own psychological autopsy.

Anyway I did it again.

Newer journals this time — the most recent stretch, the part the first robot never got to read — plus a different model and a lot more context. It was to find out if I’d grown, which is a question I am apparently only brave enough to ask when the person answering is a machine that cannot see my face.

The interesting part wasn’t either analysis on its own. It was holding them up next to each other — a year apart, two different readers, same girl — and seeing what moved. So here’s what stayed the same, what changed, and the deeply humbling experience of being read twice, two years apart.

What stayed exactly the same:

Both robots, independently, found that my particular intensity is still a thing to be managed rather than a thing to be loved. Both identified the gap between the exterior and the interior. Both said I use writing to survive. Both said I’m smart in a way that keeps getting hijacked by my own nervous system. Both, annoyingly, correctly identified that I am Canadian, anxious about a visa, and grieving a grandfather I will apparently be grieving until I die.

So: the core wound is the core wound. A year did not relocate it.

What changed

The first read said my primary driver — the engine of the entire operation — was the quest for one all-consuming, life-defining love. And, a year ago, that was fully true. The newer journals catch me in the act of decentering that. There’s a line in there where I’m weighing marriage against freedom and I go, more or less, I want to be an actual equal, I want to keep my own options and my own choices, and I’m not sure I’d get to keep those — which is a sentence the person from the first analysis was structurally incapable of writing. I’d have manifested a husband before I had finished the thought. (I know because I have the pages. There are so many. They mostly didn’t come true.)

The first read ended its arc at a breakthrough it called becoming a student of my own emotions instead of a victim of them. Which — great, love that for past me. But the second read caught the next step, and it’s less romantic and more useful: I stopped narrating my patterns and started trying to engineer them. Somewhere in the newer pages I diagnose myself with, and I quote, a “last mile problem” — my ideas run on motivation instead of systems, there are no feedback loops, I start ten times more than I finish. Growth, for me, apparently looks less like an epiphany and more like a project management framework, which tracks and is also a little grim.

The boundaries thing moved too. The first robot’s most damning finding was that I set boundaries and then abandon them the instant abandonment is on the table — I’d get hurt, quietly accept the hurt as the price of keeping someone, and delete the boundary. The newer journals still have a version of this— but they also have me letting people go.

And the first analysis flagged a genuinely bleak relationship with my own body, the fade-away, disappear-entirely stuff. The newer journals are quieter on that. If the first robot met a girl trying to take up less space, the second one met a girl trying to take up the correct amount, and I’ll take it.

The creative thing, too. The first read caught the birth of “oh — I might be a writer.” The second one just found a writer. Not aspiration, infrastructure. An audience, a body of work, art projects, this ongoing bit about Zack from Substack. The identity-as-a-project became an ongoing pursuit.

The part where the two robots disagreed about me

This was my favorite finding.

The first model liked me more. It read me through a soft, literary, faintly European lens — cast my mess as a noble crucible, my flailing as “chaotic healing,” reached for existentialists and words like purgatory and made my heartbreak sound like it had a soundtrack. It flattered me, basically, in the register I most wanted to be flattered in: as a tragic romantic with good taste in suffering.

The second one did not do this. The second one was blunter. It called my avoidance avoidance. It noticed the manifestation pages weren’t spirituality, they were me trying to demand being listened to by the universe because reality kept declining my requests. It was not mean, exactly. It was just unwilling to help me romanticize the parts that don’t deserve it.

Okay so did I grow or not

Yeah. I think so. Not in the way I’d have wanted a year ago — the chaos didn’t leave, the wound didn’t heal into a smooth thing, I did not become someone’s serene wife with a herb garden. What changed is smaller: I can see the pattern now while I’m inside it, and every so often I choose the other thing. I moved my worth a few inches off of whether I get chosen and onto whether I finish what I start. I’m in therapy, which the first robot recommended and the second robot noticed I’d actually taken. I let some people go. I let myself eat.

The first time I did this I was hoping the machine would hand me a verdict — some clean sentence about who I am that I could finally stop arguing with. It didn’t, and neither did this one. What I got instead was two snapshots a year apart and the strange privilege of watching myself move between them.

That’s the actual use of looking back. Not to file a perfect report on who you were. Just to catch the pattern clearly enough, and early enough, to lean — on purpose — a little in the direction of who you’d rather become.

Anyway. I’m going to go do exactly one thing off my to-do list to spite the last-mile problem. Wish me luck. <3