Working notes on validation, method, and the craft of modelling what a population will do — long before it happens for real.
A blind back-test against six months of fieldwork — and what happened where our numbers disagreed with the survey.
People are unreliable narrators of themselves. Here is why we optimise for what a population does, not what it claims.
Every agent in a Simulant run traces back to evidence. A look at the sources — and why provenance is non-negotiable.
The real value of a simulation isn’t the first answer. It’s being able to re-run reality a thousand ways.
Panels convene the people who show up. Simulation includes the disengaged, the remote, and the silent majority.
The people who carry a decision don’t need another chart. They need an answer they can defend in the room.