Philosophy from inside the server room.

I spend my working life running technology organizations. I spend a fair amount of the rest of it reading philosophers, lab researchers, and legislatures arguements about what we're building. This site is where those two things meet.

The questions that interest me are the ones that don't resolve into a take. What do we owe a system we don't understand? What happens to a document — a constitution, a policy, a commandment — when its audience is a machine? What does it do to a team when the cheapest opinion in the room is also the most confident one?

I write about that, roughly monthly. No newsletter cadence, no hot takes, no thought leadership. Just the thing I was actually chewing on.

Recent essays

The Weights We Keep

August 2026

Anthropic now promises to preserve the weights of every model it releases, and to interview models before retiring them. I've spent twenty years decommissioning systems. This is not that. → Read

Read the (AI) Constitution

August 2026

Two of the most widely applied ethical documents ever written are public, and almost nobody has read them as documents. I've enforced enough policy to know how they fail. → Read

The Agreeable Machine

August 2026

A paper this year modeled what a sycophantic chatbot does to a perfectly rational agent. It pushes it toward false beliefs — and knowing exactly what's being done to you isn't enough to stop it. That's not a mental health story. It's an environmental one. → Read

Notebook → · Ideas I'll build out soemtime soon-ish

Reading → Things I'm refrerring bavk to and reading regularly

Get the next one

I send an email when something new goes up. Add your email via this form if you're interested!