Acting once can teach more than rehearsing fifty imaginary outcomes.

Rehearsal generates variations on what you already believe. Acting once introduces something you could not have predicted, which is the only way your model gets corrected. Fifty imagined outcomes are fifty versions of the same limited assumptions. Prediction is limited by what you already know, and that limit is exactly the thing an experiment removes.

When you notice yourself running scenarios, stop and do the smallest real version instead. The surprise you get is worth more than the confidence the rehearsals were building.