Accurate predictions are not required to design an effective policy to control the pandemic.

Monte Carlo simulation results: The median is indicated with a thick line, the 25–75th percentiles and the min-max are indicated with shaded regions. Left: Under an open-loop policy, 40.25% of the 400 realizations reach a maximal number of infections over 13500 after June 30th, 2 months after the level of activity u(t) is increased. Right: With the same dynamic feedback policy in place starting May first for all realizations, and bi-weekly intervention updates, 100% of realizations maintain infections below 550 after June 30th. The variability in model parameters leads to adjustments in the interventions, which compensate for the differences between the models.