The difference in wellbeing, B − A
How it gets there
Five more charts: the training run, investment, the labor share, growth and a displaced worker's consumption
Each assumption's rank correlation with the outcome across simulated futures: how closely the spread tracks it. The swing test pins one assumption at its 10th and then its 90th percentile, keeps every other one drawn, and reruns on the same seeds. Both measure the model's sensitivity, not the merit of any lever.
Rank correlation with each outcome
Swing test on the difference B − A
Every yearly variable is a node showing the formula it is computed with, grouped from compute down to wellbeing. Grey arrows read values from the same year. Violet arrows carry a value into next year: they end at the dashed boxes on the right of each group. The engine builds the diagram from the same formulas it runs, so the diagram always matches the model. Hover, focus or tap a node to pick out its arrows; select it to see its distribution over time.