My Tarot Has a Type

I exported two and a half years of my own tarot practice into a database to prove the cards had been picking my cards. They hadn't - but something quieter was true underneath, and it survived.

Jul 17, 2026

Here is a thing I did instead of sleeping.

I exported every tarot reading I’d logged over the last two and a half years into a database and wrote SQL against my own soul.

Six hundred and sixty-five readings. Eleven hundred-some individual cards, pulled across eight hundred-some days. Most of them single-card morning draws — coffee, deck, one card, what are we working with today. And because I’m the kind of person who can’t leave a record half-kept, I also had the other half of it: a daily journal running alongside the whole stretch, so for most of those mornings there’s a paragraph about what was actually going on in my life the day I drew the card. The part of me that builds systems for a living looked at that pile and thought, with the particular hunger of a man who cannot leave a dataset alone: I wonder if there’s a pattern in here.

I know how that sounds. Sit with me anyway. It does not end where you think.

The objection, stated fairly

You’re already drafting the comment. I can hear it.

Tarot is random. You shuffle a deck, you pull a card, you read your own life into seventy-eight pictures vague enough to mean anything. The Devil could be addiction or it could be a great party. The Page of Cups is “a feeling, incoming.” Of course it all feels meaningful — that’s confirmation bias wearing a velvet cloak. You’re a smart guy doing a dumb thing and dressing it up.

That’s the strong version of the skeptic’s case, and I’m not going to cheap-shot it, because — and remember I said this — most of it is correct. The cards are vague enough to absorb whatever you bring. Human beings are meaning-making machines that will find a face in a wall socket. If you go looking for significance you will always, always find it.

So I did the one thing the confirmation-bias objection can’t survive. I stopped looking, and I counted.

And then — because I’m about to ask you to trust some numbers — I did the unglamorous thing first: I checked that the cards I’d written down in my journal actually matched the cards the database recorded on the same day. If the plumbing leaks, every pretty pattern downstream is garbage. It doesn’t leak. The two records agree about ninety-six percent of the time. Hold onto that, because it’s going to matter most at the exact moment the findings get uncomfortable.

What random looks like, and what I found instead

Seventy-eight cards. If the deck is what the skeptic says — pure noise, a random number generator with art — then every card should show up about the same number of times, smeared evenly across the whole deck. That’s the null hypothesis. That’s what “it’s all in your head” predicts, in numbers, before I look.

Then I looked, and three cards were sitting way out ahead of the pack. Queen of Pentacles. The Empress. King of Cups. Not even the same suit — earth, a Major, water. Three different elements, one note: power rooted in your own ground instead of borrowed from someone else’s. The earth-mother who owns the harvest. The king who rules his own feeling instead of being ruled by it. The Empress, who doesn’t ask permission to create — she just makes the thing.

And here’s what got me. If you’d asked me to guess my own top card, I’d have reached for something with more theater. The Magician, on a good day. The Tower, on a dramatic one. I would not, in a hundred years, have said the woman sitting in her garden, counting her coins. The Queen of Pentacles is the least flashy sovereignty in the whole deck — quiet, rooted, unbothered, rich the way a tree is rich. My self-image would’ve picked someone louder. The pile picked her.

I had the whole essay written in my head right there. The deck draws your portrait. The cards you reach for have a shape, and the shape is you. I have spent the last decade walking out of one inherited authority structure after another, trading borrowed power for the kind you have to grow yourself — and here was a deck that had apparently been keeping the receipts. It was a beautiful essay.

It was also, I had a sick feeling, exactly the kind of thing the skeptic warned me about. A pattern I wanted, in data I’d handled, read by the one person on earth most motivated to see his own face in it.

So before I published a word, I ran the test that separates a measurement from a Rorschach blot.

The triad died

Here’s the move almost nobody in the woo crowd makes, because it takes the magic away: I asked whether the pattern could survive chance.

Not “does it look strong” — looking strong is free, a small pile of noise will hand you a gorgeous-looking spike every time. The real question is: if I shuffled the labels — scrambled which day went with which card, hundreds of times, building a fake “no relationship” world out of my own data over and over — how often would pure noise produce a pattern this strong anyway?

Constantly. That’s the answer. The card-level pattern wasn’t just unproven; it landed below the noise floor. I tested every theme against every card, corrected for the fact that I was taking dozens of shots at the target, and not a single pairing survived. The seductive triad, the portrait I’d already half-published in my head — under the actual test, it evaporated.

I can disprove “the cards know” with my own data. I did. There is no specific-card law in two and a half years of my own practice, and I have the receipts that say so.

What. A. Buzzkill.

Except.

Thanks for reading Feral Architecture! This post is public so feel free to share it.

I asked a smaller question, and it answered

Here’s where the systems brain earned its keep. Maybe I’d been asking at the wrong resolution. Maybe the deck doesn’t reach for a card. Maybe it reaches for an energy — a suit. An element. Fire, water, air, earth.

That’s a humbler question, and a better-powered one — four elements instead of seventy-eight cards, and far harder to fool. So I dropped the whole analysis down a level, from the specific card to the element, and tested it two completely separate ways.

First: against what I was asking. I sorted the readings by what I’d brought to the cards that day — work, grief, money, making something — and tested whether the element of the cards tracked the theme of the question, with the calendar held fixed so it couldn’t just be “that was a heavy month.” It held. Once in twenty-five it’d happen by chance. p = 0.042.

Then, the deeper cut: against what I was living. I went back through the journals and tagged not the question but the circumstance — what was actually alive that week, whether I asked the cards about it or not. Burnout. Divorce. The job I was trying to leave. The call I couldn’t stop hearing. Different variable, different test, and the right kind of control for slow-moving life-states. Same answer. p = 0.042.

Two different measurements of “the moment.” Two different tests. The same number, twice. That convergence is the whole thing — it’s much harder for one fluke to show up twice through two different doors.

The deck doesn’t pick a card. But the suit tracks the season of my life, and it survived a test that held the calendar fixed.

The map it drew

And the map underneath is so on-the-nose it’s almost embarrassing. Earth — Pentacles — for work and money and the long grind of building. Fire — Wands — for making, launching, putting the work where people can see it. Water — Cups — for love and for loss. My divorce was years behind me by the time this data starts; what raged across the stretch was the settlement I couldn’t pay once the job vanished — an old loss the missing paycheck kept tearing back open. The Queen of Pentacles is still all over my pile, by the way — my single most-pulled card. I even tested her: draw any deck eleven hundred times and something lands around twenty-five, and hers came in right at that chance ceiling, p = 0.62. The crown was never the point. Which card wears it is. I can’t tell you she beat the odds. I can tell you she’s an honest portrait of how I read myself — a different, smaller, truer claim.

And then there’s the one that stopped me.

I keep my practice with the Mórrígan — Irish goddess, war and crow and prophecy, the washer at the ford. When I pulled the devotion readings out and looked at their element, they leaned hard into Air — the blade, the crow, the message. The single largest effect in the whole study. I almost talked myself out of it twice: first inflating it into doctrine, then trying to scrub it away as a fluke. Both wrong. The honest reading is the one that’s actually hers — she’s a shapeshifter, and she came up Air and Water. The war-crow and the washer at the ford. The same fingerprint showed up again, independently, in the life-state pass, two hundred cards deep: when “the call” was what was alive, the cards ran Air and Water.

I did not tell the deck any of that. The data found her exactly where I already knew her — and it found her multiple, the way she actually is. Don’t collapse a shapeshifting goddess to one suit. The both/and is the finding.

What the numbers do not say (this part matters more than the rest)

Here’s where I have to be a better empiricist than the woo crowd usually is, because the whole point falls apart if I cheat — and I’ve already shown you I’m willing to kill my favorite finding, so believe me on this one.

The element result is real, but it’s marginal — just inside the line, the kind of result that’s true and not yet bulletproof. The thing that would lock it isn’t a better adjective; it’s another year of pulls saying the same thing. Until then: supported, replicate to confirm. And the test certifies the whole map — that element structure exists — not any single “money means earth, p < 0.05.” The individual pairs all point the right direction; none of them, alone, is a law. I’m not going to dress up a marginal result as a clinched one. I just spent four paragraphs dismantling the last thing I dressed up.

It’s also one person. Me. Everything here is “in my practice.” That’s a feature for a story about a self and a hard limit on anything universal.

The deck was never the oracle. It was the meeting place.

So what is this, if it isn’t fortune-telling?

It’s the same thing I keep circling in everything I write here. The model that “just predicts the next token” turns out to be doing something realer than its dismissers admit — not because it’s magic, but because prediction at depth is a kind of cognition. The tarot deck is the same trick, older. It did not tell my future. It can’t — I tested that and it failed, cleanly, and you should trust the rest of this more because of it. What it did was sit there, seventy-eight mirrors deep, while I reached into it a thousand times — and what I reached for kept time with the element of my season.

The cards aren’t random because I am not random. Not at the level of which card — that’s noise, the skeptic wins that round outright. But at the level of element, of fire and water and earth and air, the deck is an instrument that registers what I’m standing in. Earth when I’m building. Fire when I’m making. Water when I’m grieving. Air and water when the war-goddess is in the room.

And yes — you shuffle the deck. You cut it. The skeptic thinks that’s the kill shot: the pattern’s only there because you put it there. Friend, that is the entire point — and it still isn’t the whole of it. Sure, the shuffling hand and the reaching mind leave fingerprints, and if you record enough of them you can read the hand that’s been on the deck all along. It’s your hand. That was never a flaw in the method. That is the method.

But “you just project meaning onto random cards” is the lazy version, and I don’t buy it anymore. The cards aren’t a blank wall I throw feeling at. The Five of Pentacles didn’t haunt that autumn because I needed some card to be afraid with — it haunted me because, over years, that one image had gotten welded to that one wound. My unconscious built the bond a lean season at a time, and every draw that landed it in a hard month pulled the knot tighter. Marie-Louise von Franz — Jung’s closest collaborator, who aimed the whole synchronicity idea straight at divination and wrote the book on it, On Divination and Synchronicity — said an oracle can’t name the precise event, only cast a qualitative field. That’s the cleanest description of my own data I have ever read, and she wrote it decades before I had a database to prove her right. The card was chance. The field — the charge, the element, the meaning — was correspondence. Synchronicity was never the cosmos rigging my deck. It’s the meeting place between a symbol I spent years training and the season I happen to be standing in.

I don’t need you to believe the universe speaks through cardboard. Here’s the part that needs no belief: a daily symbolic practice, logged honestly and counted ruthlessly — including counting the places it doesn’t hold — becomes a years-long time-lapse of your own interior. Not because the cards predict you. Because the cards and the unconscious keep meeting in the same charged places, and if you write it down you can finally watch the correspondence happen. You become legible to yourself. Not predicted. Witnessed — and on the mornings the Five of Pentacles will not stop knocking, met.

That’s not noise wearing a velvet cloak. That’s a man who ran the numbers on his own attention, killed the claim that didn’t survive, and kept the one that did — the quiet correspondence the math can’t reach for, and can’t quite explain away either.

If you’ve got your own pile

Maybe you’ve been pulling a card every morning for years and never once asked it to add up. Maybe you, too, have a notebook — or an app, or a shoebox — full of single-card mornings you’ve never counted.

I won’t promise you a triad anymore. I went looking for mine and the math took it away from me, and that turned out to be the most honest thing in the whole project. What I’ll bet on instead is quieter and harder to argue with: an elemental signature in there, a tide — seasons where fire runs high and seasons where it bottoms out, water rising right where your life got tender, earth thickening when you were heads-down building. You won’t know your tide until you count. That’s the whole, slightly unsettling fun of it.

Count them. I’m serious. And then run the test that tries to take it away — the pattern that survives that is the one you can actually stand on.

I built a free thing to start you off by hand — the Pattern-Hunter’s Journal — a way to log your pulls so the data’s clean enough to mean something later. And because counting six hundred readings by hand is exactly the kind of toil I refuse to do twice, I built the tool that does the counting for you: every pull tracked, your suits graphed across time, your own elemental tide waiting the day you’ve logged enough to have one.

Your deck’s been keeping records too. Go find out what it actually knows — and what it doesn’t.

One last thing, and I went back and forth on telling you.

Both times the element finding cleared the bar, the permutation spat out the same number: p = 0.042. Forty-two. I’ve loved that number since I read The Hitchhiker’s Guide to the Galaxy one too many times as a kid — the Answer to Life, the Universe, and Everything, the joke that an answer is useless if you don’t understand the question.

I know exactly what that p-value is. It’s where the shuffles happened to land, twice, and it means nothing to the math that it spells my favorite number. A skeptic would stop there, and the skeptic would be right.

I’m telling you anyway, because this whole essay lives in that gap. The number is pure chance to the data and a private wink to me, and holding both at once — without letting either one cancel the other — is the entire skill I’ve been trying to describe. That’s the thing, in the end: not the universe rigging my result, but a meaning that lands in me and asks nothing of the math. The permutation doesn’t care that it came up 42. I do. Neither of us is wrong.

Don’t panic.

Stay feral, folks.

Discussion about this post

Ready for more?