Most of my work right now is about governing behavior, not just composing interfaces. That pulls me toward material on systems, uncertainty, and socio-technical design — the disciplines that took the question of "what should this thing do" seriously long before AI agents made it fashionable again.
This is the list I'm working through in 2026.
- Thinking in Systems by Donella Meadows — the primer on stocks, flows, feedback, and leverage points. Still the clearest book on why interventions backfire.
- Human Compatible by Stuart Russell — the case for aligning machine behavior to human preference, and what it costs to get that wrong.
- Seeing Like a State by James C. Scott — how high-modernist attempts to make society legible collapse the local knowledge that actually runs things.
- Sources of Power by Gary Klein — how experienced people actually decide under time pressure, which is not how the requirements doc says they do.
- The Art of Uncertainty by David Spiegelhalter — how to reason about probabilities without pretending they are certainties.
- The Logic of Failure by Dietrich Dörner — why smart people make systems worse by acting on partial models of a complex situation.
- Normal Accidents by Christopher Perrow — why tightly coupled, interactively complex systems fail in ways no checklist can prevent.
- Notes on the Synthesis of Form by Christopher Alexander — the argument that good form is a fit between context and design pressures, not a matter of taste.
- Domain-Driven Design by Eric Evans — the vocabulary for modeling a system in terms the business and the code can both hold onto.
And adjacent work in software architecture, data systems, uncertainty, and socio-technical design.
I keep this list short on purpose. Reading more books than I can use is just procrastination with better branding. Each one earns its place only if it changes how I frame a real problem.