I did not set out to be a journalist. I set out to understand how records get joined together, and the understanding turned into an obligation.
It started as a technical curiosity. Record linkage. The quiet math of deciding that this row in one database and that row in another are the same human being. There's an open-source tool for it. There are government programs that run it across the datasets of entire populations.
I read the documentation the way I used to read query strings — closely, from the demand side, looking for the shape underneath. And the shape was this: the state can assemble a single view of you from fragments you handed over for unrelated reasons, and it can do this at national scale, and almost no one being assembled has any idea it's happening.
That's not a hack. There's no break-in. No consent, no opt-out, no notification. Just statistics quietly becoming operations.
Once you see it you can't return to not seeing it. That's the whole trap of this work.
So I followed it. Where the tooling came from. Who deploys it. How the same infrastructure that sorts census records also helps sort people at a border. How a firm's software ends up threaded through a country's ministries with no single moment anyone can point to and say that's where we decided this.
And separately, a different thread with the same texture — influence operations run as registered foreign work, campaigns built in their own filings to bend what the AI models tell you when you ask them a question. Documented from the sworn paperwork. Verified against the primary record. With the honest negatives stated up front, because a claim you won't caveat is a claim you don't trust.
I work evidence-first. That phrase does real work, so let me define it. I do not ask how to prove what I already believe. I ask what I would need to see to be wrong, and then I go looking for that.
It's the same discipline as closing a security hole. You don't get to feel that the port is shut. You connect from the outside and watch it refuse you, or it isn't shut. Belief is not evidence. The external test is.
I keep the stories close to me, and I'm deliberate about that. I'm the source and I'm the story. I route the work to outlets by who'll give it the credit and the reach it needs, but I don't hand over the thread itself. I've been erased from my own work before. I don't build that way anymore.
People ask why a former affiliate marketer with a rhetoric degree is the one publishing surveillance dossiers. Fair question. Here's the answer.
Because I spent my career doing the demand-side version of exactly this. Linking identities across sites. Reading the encoded trails people leave. Manipulating what surfaces when someone searches. I know how this machinery works because I built the small commercial version of it, over and over, for money.
The surveillance state and the marketing stack are the same tool pointed at different targets. I recognize it the way you recognize a face. I just spent a long time on the wrong side of the lens.
The journalism is not a pivot away from the building. It's the same act. Take a system that's being run on people without their understanding, and make it legible. Sometimes that means shipping a cipher. Sometimes it means shipping a dossier.
I'm not a neutral observer and I've never claimed to be. I'm a defector with receipts, and the receipts check out. That's the only credential I'm offering, and it's the only one that matters.