I’m currently working on decision-centric systems — software that operates under real constraints rather than ideal assumptions. Most of my focus is on how decisions are shaped by boundaries, accountability, and failure conditions, especially in environments where outcomes matter and reversibility is limited — questions closely related to work like Normal Accidents.

Lately, I’ve been thinking a lot about governance-first architectures: defining what is allowed, what is excluded, and where escalation should stop before any optimization or automation takes place. The goal isn’t efficiency or cleverness, but operational clarity — systems that behave predictably even when information is incomplete or politically constrained, a perspective explored in Seeing Like a State.

This work is less about building tools and more about designing structures that can act without constant human involvement. I’m exploring how responsibility, escalation paths, and fallback logic can be embedded directly into systems, so that they hold up under real-world pressure rather than ideal scenarios.

Notes and essays I’ve written over time.

• Design, Delegate, Disappear: The Architecture of One
• The Cost of Intelligence Is Approaching Zero.
• AI: Drive Yourself – Don’t Just Be a Passenger

→ full /archive

Working assumptions that have remained stable across projects.

• The World Is an Edge Case
• Build Useful Things
• Ease of Use
• Don't Make Bad Things Better

All Book Notes