Welcome to Updated Priors

In the last eighteen months (writing in summer 2026) the way that I interact with code has shifted dramatically. Tools and techniques that were daily drivers for years are nearly obsolete, and methods of working that I rely on heavily didn’t exist mere months ago. What does this mean for how we work and live as a society? Are we ushering in a renaissance that will solve all of the world’s problems, or are we on the edge of a precipice about to plunge our society into disaster? The throughline is that radical change is coming. Yet in all this change, I’m seeing timeless patterns hold true—although refracted through a different lens. Anyone who says all the rules have changed is probably operating at a different level of abstraction.

But if you walk with either story about where this technology is headed far enough—either the optimists or the doomsayers—they end up at the same destination: people reduced to a set of sophisticated data points. In one telling, we’re functions that can be optimized; in the other, we’re deprecated components about to be rendered obsolete. Little more than brains in vats, whichever end you’re rooting for. I don’t buy it. We’re more alive than that and irreducible to a vector of any length. We found ourselves here because we’ve believed, for a long time, that with the right technique we can control anything—including each other. It’s how we learned to assemble cars faster, and it’s how we learned to turn a career’s worth of engineering judgment into story points and sprint velocity. A person shaped by enough of that doesn’t need a vat. The ports don’t need to be visible.

That same faith—that technology pushed far enough gets you total control—doesn’t stop at people. It shows up as a belief that with the right technique, we can know anything completely: extract it, verify it, eliminate the uncertainty. Once you stop believing you can control a person that way, it’s not much of a leap to stop believing you can control knowledge that way either. When our world is re-enchanted and re-personed, so much more comes alive. J.R.R. Tolkien said “It’s a dangerous business, Frodo, going out your door. You step onto the road, and if you don’t keep your feet, there’s no knowing where you might be swept off to.” If we’re truly open to knowing as a journey, we never know where we are going to end up or how we’ll be changed. This blog is a logbook of my journey.

I’m Matthew Cox. Minneapolis-based. And yep, took four paragraphs to get around to that. True to form—ideas first, name later.

The name “Updated Priors” comes from my training in Bayesian statistics. It’s one of the ways I navigate a world where our ability to control turns out to be more tenuous than we’d like. I see a pattern, I try to make sense of it. And then one day I encounter something surprising that upends my prior stance. And I move a little in that direction. Holding contradictory ideas in tension is a feature, not a bug. It helps guide me to where I’m landing next. The code we write is binary, we are not. Don’t be surprised if I change my mind as we continue to experiment and the world shifts around us. A lot of my thoughts here come from a long and varied professional journey—from non-profits, to small startups, to big tech. I’ve built data warehouses, architected search systems, and trained transformer models. Here are some of my quirky side projects. The throughline is that I’m constantly curious and integrating what I’m encountering.

I’m not writing this from up ahead, already knowing how the road goes. I’m on it too, getting swept along the same as anyone who steps out their door. We may end up back where we started, but we won’t be the same. That’s what it means to update a prior instead of defend one.