Changing behaviour with less disturbance

One target. Many possible changes.

Drag the orchid arrow along the dashed line. Every position reaches the same target; the sage arrow finds the least costly one.

StartSame target along the dashed line; drag to explore
2.74×your change’s cost relative to the minimum.
Your changeLeast costly change

Ellipses join changes with equal regularised local cost. A mathematical illustration in two internal directions.

The geometry predicts the cost of ignoring it.

Two interventions reach the same calibrated target. Each dot compares their predicted cost ratio with the ratio measured after changing the model.

Loading interventions…

The advantage persists along a finite path.

Loading recorded measurements…

An update can preserve one set of prompts and transfer to another.

Loading recorded measurements…