As with weather, there is a limit to how far into the future we can accurately forecast the future course of epidemics. Yet, we often need to make plans further into the future than we can reasonably forecast, and in many cases our actions themselves will change the course of an epidemic. One solution to this conundrum is the construction of “planning scenarios” where we project the likely course of an epidemic well beyond the limits of proper forecasts under sets of assumptions that constrain future epidemic drivers, thereby allowing longer term conditional forecasts. The US COVID-19 Scenario Modeling Hub (SMH) is a multi-team effort aimed at creating such planning scenarios while leveraging the power of ensembled predictions from multiple independent teams. The SMH has been impactful in a variety of domains, but confronts a problem shared by all planning scenario based approaches; how do we evaluate projections that by their very nature make assumptions about the future that are unlikely to be realized? In this talk I will explore the nature of planning scenarios, the challenges of their evaluation and their practical performance through the lens of the 16 rounds of projections produced by the hub over the past two and a half years.

Zoom Link