All models are wrong, but some are useful: mathematical models at the time of Covid-19

Abstract

Epidemiological models have been crucial tools throughout all stages of the 2020-21 Coronavirus pandemic: using promptly available or historical data, they have studied and tried to anticipate its progression, providing valuable guidelines for public health officials, policymakers, and other medical and non-medical audiences. While useful, models are not designed to be infallible, and for this reason, they have been frequently subject to criticism. There is a discrepancy between what models do and how they are presented and perceived. Several juxtaposing factors, including current beliefs about scientific reliability, the role of quantification, and the epistemic values grounding the field, are at the core of this discrepancy. While scientific literacy may play a role in addressing this discrepancy, analyzing and becoming better aware of these factors may suggest long-term strategies to address, acknowledge, and communicate the pandemic’s inherent complexity and stochastic qualities.