I trained twice. First in applied physics, where I learned that a theoretical understanding of how things work is necessary but not sufficient. Then in industrial design, where I learned that making things work in the world requires a different kind of thinking — one that holds material, manufacturing, human use, and intent in the same frame.

The combination is unusual enough that I've sometimes called it design physicist, though the label matters less than the practice it describes: rigorous about what's true, equally rigorous about what's usable, suspicious of solutions that only satisfy one of those constraints.

Heading: How I work

Design, for me, is the work of making complex things appear simple. The simplest things take the most work to implement — felt-tip pens should be crisp, Wi-Fi should be fast, agent systems should do what was actually asked. I don't accept things that don't work, and I don't accept things that work but feel arbitrary.

The method is iterative and ruthless about its own bad ideas. The first several hundred concepts on any problem are usually wrong. The work is understanding why they're wrong precisely enough that the next iteration is informed by it. Watch, learn, do, fail, repeat — at increasing levels of resolution until something earns its existence.

Heading: What I'm working on now

The questions I'm most interested in right now sit upstream of any specific product: what does it mean to design systems whose implementer reasons rather than executes? How should architectural decisions be encoded so they survive being read by something other than the person who wrote them? What's the difference between architecture and prediction in agent systems, and why does that distinction keep collapsing in current practice?

I publish work on these questions at "Before It's Built." The name reflects the orientation — the most consequential design decisions happen before anything is built, and they deserve the same rigor as the things that get shipped.