I Didn't Know What I Was Writing About

My coffee group meets the first Friday of every month, and it has one rule: no work talk.

I was halfway through a latte, making the case for the latest Miley Cyrus record to a table that clearly hadn't pegged me as a fan, when someone broke the rule.

"Hey man, Karl isn't here yet. I noticed you're writing more about data on your Substack."

It wasn't a criticism. It wasn't really a compliment either. It was an observation, made in passing, by a guy who works in data. Then the conversation went back to Miley.

Around the same time, someone else said almost the opposite. "Dude, you build some amazing stuff. You should be writing about it. I don't see enough about how you actually do it." Tips, tricks, workflows. The building was showing up everywhere except the writing.

I honestly can't remember which comment came from the coffee table and which came from an IAA event. It doesn't matter. Two people, reading the same body of work from opposite directions. One noticed what had arrived. The other noticed what was missing.

I'd been writing for years and had never once looked at my work that way. I knew what each essay said, because I wrote it. I had no idea what all of them said together.

I'm constantly curious about my own transformation, and I assumed that meant I'd notice it. I didn't. Other people did. They saw the drift because they only checked in every few months, the way a relative tells you your kid got taller. I was changing right in front of myself.¹

The Line Between the Lines

Reading between the lines has been most of my career. The brief says one thing. The client means another. The culture is saying a third. I've called myself a cultural anthropologist without the degree, and the work is mostly that: noticing what a piece of content carries beyond what it says.

Every content object carries a second payload. An essay, a LinkedIn post, a reel, a three-minute film. On the surface, it's a message. Underneath, it's a record of what someone was paying attention to, when they were paying attention to it, and what they thought was worth saying. Read one at a time, that's content. Read in sequence, it's a signal.

The person making the content knows what they meant to say. The body of work records what they actually said.² Those aren't always the same thing.

We all carry a fairly tidy story about ourselves: what interests us, what we believe, what we're working toward. Our attention doesn't follow the story. We get interested in something without deciding to. An idea that once sat at the edges starts showing up more often. Something that used to take up a lot of our thinking quietly disappears.

We remember when we change our minds. We rarely notice when we change what we're paying attention to. Nobody wakes up and announces a new interest in the relationship between technology and human behavior. You find yourself asking different questions, then asking them again, then writing about them. Eventually someone across a coffee table points out you've been doing it for months.

My coffee group had read me in sequence. I never had.

Why the Feed Couldn't Show Me

Substack was good to me, and if you're reading this in your inbox, there's a decent chance that's where you found me. Thank you for coming along.

But a newsletter platform is built around the newest post. The archive is a stack: newest on top, everything else sinking. That's the right design for a feed. It's the wrong design for understanding how ideas move.

I could go back and reread something from a year earlier, but I couldn't see how it related to what I'd written last week. I couldn't see which themes kept resurfacing, or an idea that started as an aside and became the spine of another essay eight months later. I couldn't see what in the culture had bent the line, whether it was a film, a product launch, a client problem that wouldn't leave me alone, or a conversation over coffee.

And I couldn't see what was disappearing. What had I stopped writing about? Which questions had I abandoned? The absences might say as much as the subjects that kept coming back.

The information wasn't missing. It was all there. The format just presented every piece as a separate event instead of part of a pattern. So I built somewhere I could see the pattern.

The Experiment

The journal now lives at christophersmith.sg, and it works differently. Every essay is tagged by theme. Margin notes sit beside the text, holding the asides, references, and second thoughts that used to live in my head or a notes app. When a new essay picks up a thread from an older one, the older essay gets a forward note pointing to where that thought went next. The archive stops being a stack and becomes a web.

Then there's the Ledger. It reads the whole body of work as a moving system, looking for relationships you can't see when you're writing or reading one essay at a time. It shows which subjects are emerging, which are gaining ground, and which are quietly losing it. Where I can trace it, it also records what moved them: a cultural moment, a conversation, something I read or made, a question I couldn't stop turning over. It isn't grading the essays. It's showing me how my thinking moved.

Once it was built, the patterns were hard to miss.

The Epilogue Economy carried 65% of my first 40 essays.³ In the last 25, it's 16%. I hadn't noticed. How you build a life and a career after fifty matters enormously to me, and I hadn't put the name in a headline for more than a year without realizing it. But the idea didn't vanish. It changed clothes.⁴ Aging stopped showing up as a consumer market and started showing up as an argument about work: experience shortages, judgment, what juniors lose when AI skips the struggle. The idea kept going. It just stopped wearing its label, so I stopped recognizing it.

Friction, agents, and data climbed fastest. That part I could have predicted, because the work drove it: the systems I was building and the questions they raised. The surprise was AI filmmaking.⁵ Film is roughly half of what I do for clients, and it's the theme that has faded fastest in my writing. The person who told me to write about how I build was right. The building had gone quiet on the page.

None of it was hidden. The evidence had been sitting in plain sight. I just hadn't been reading it that way.

The Problem With Studying Yourself

This experiment has an obvious flaw.⁶ A writer's body of work isn't a clean dataset. I choose what to write and what to publish. The topics I avoid never appear, and the ideas I abandon halfway through a draft might be as revealing as the ones that make it onto the page.

What we publish isn't what we think. It's what we choose to make visible. And the person generating this dataset is also the person interpreting it. An anthropologist studying himself is a compromised observer. I know that.

Hindsight is the other trap. Once you see a pattern, it's easy to build a convincing story about why it happened. The Ledger can show me that a subject appeared more often. It can't tell me why I got interested in it. Correlation is not autobiography. That part still takes interpretation, context, and a willingness to question my own assumptions.

Then there's the observer problem. Now that I can see the patterns, will they change what I write? Will I chase themes because the Ledger says they're rising, or fill gaps on purpose because I can see them?

Probably. That's no reason to stop. I'd rather run the experiment in public, flaws visible, than pretend it's objective science. And if watching my own trajectory bends it, that's worth watching too.

The Ideas Before the Ideas

What if an archive could show an emerging interest before its creator recognized it? Not a prediction. An early signal.

We like to think ideas arrive: a flash, a sudden connection, an insight fully formed. In my experience they accumulate. A conversation leaves something behind. A film raises a question. A client problem introduces a contradiction. Most of it doesn't register at the time. Months later you're writing an essay about a subject you never meant to explore.

Earlier this month I published an essay arguing that we optimized the wrong thing: we made everything convenient, then started paying to get the effort back. It felt new. It came with a new phrase, friction worth having.⁷

It wasn't new. The trail goes back two and a half years, to a list of life lessons I wrote on the eve of turning sixty. Somewhere between money and marriage, I explained why I still play records: you are active, you are present⁸. I thought I was talking about vinyl.

Sixteen months later it surfaced in, of all places, an essay about Reddit as a research tool. Frictionless often means lifeless, I wrote, in passing.⁹ A sleepless night with Bradbury and Proust gave it a mechanism. A bookstore, a walk in Bucharest, and a dating app gave it evidence.

And in the middle of all that, I'd argued the opposite. In July 2025 I defined an AI-first culture as post-friction, and I meant it.¹⁰

The idea didn't just grow. It argued with me on the way.

That changes what an archive is. It isn't only a record of finished thoughts. It also holds the unfinished ones: questions that haven't become arguments yet, observations still looking for their context. An archive doesn't just show where you've been. It may show where your attention is heading before you know it yourself.

Beyond One Writer

If one writer's ideas have a visible trajectory, so does everything else that publishes. Every brand that has posted daily for a decade, every agency with a twelve-year-old blog, every creator sitting on four thousand reels, every organization with years of speeches, presentations, campaigns, and reports.

Most organizations treat all of that as spent inventory: published, measured for a week, archived.

We've become very good at measuring how individual pieces perform. Who saw it, how long they watched, whether they engaged, what they did next. None of that tells you how the thinking behind the content evolved.¹¹

There are three different questions here. How did this piece perform? That's performance measurement. What has the organization been talking about? That's content analysis. How has its attention changed over time? That's something closer to organizational self-awareness.

Imagine reading ten years of a company's published communications¹² and discovering that the strategic priority announced last quarter had been surfacing in its language for years. Or that a brand that sees itself as a champion of innovation has drifted, year by year, toward efficiency, predictability, and risk reduction. Or that an organization that talks constantly about customers has spent almost no time on how those customers' lives are changing.

None of that means anyone did something wrong. It shows the gap between what an organization believes it's paying attention to and what its archive says.

Corporate archives are curated. Everything passes through strategy, approvals, commercial priorities, and politics. A company's published record isn't a clear window into its thinking. It is a record of what it repeatedly chose to make visible, and that has value of its own. The subjects it returns to and the ideas it quietly drops say something nobody inside has examined.

We call these archives institutional memory. Memory preserves what happened. I'm more interested in interpreting it: the archive as evidence of how an organization thinks, or at least how its public thinking has moved.

What Happens When We Start Reading the Patterns?

I don't know yet what this looks like at the scale of a brand, a newsroom, or an organization with decades of published material. I know what it looks like for one writer, and that's a start.

For years, we've used data to understand how people respond to content. Now we have tools that can examine what the content itself reveals: how its themes develop, how its language shifts, which ideas connect, what patterns run across thousands of pieces.

That doesn't remove the need for human interpretation. It raises it. The patterns are only the beginning. Someone still has to ask what they mean, whether they mean anything at all, and what explains them. But seeing them changes the questions you can ask.

We experience our work one project, one conversation, one essay at a time. Our thinking doesn't develop in those neat units. It moves through connections, repetitions, contradictions, and the occasional change of direction that only shows up over time.

I started this because two people made passing observations about my writing. One noticed what I was talking about. The other noticed what I wasn't. Neither meant anything profound by it, but together they showed me the story I thought I was telling wasn't the only one there.

People keep telling me what I'm writing about. Now I can see it too.

Each essay is a snapshot. The trajectory is the film.¹³