A friend tells me of his college-age daughter’s Covid experience at the University of Wisconsin, where 20% of her dorm (1800 kids) this week have the virus. Thankfully, very few have needed hospitalisation.

What she and millions of others, cannot be sure about is the “second order” effects on their health and the future rate of morbidity. The very same issue applies to the impact on health insurers, health systems and policymakers. The “tail risk” is, and will be huge. Epidemiological models lie in tatters.

There is hope too. More statistical information resides in the extremes rather than the “bulk” data. We can with tools such as extreme value theory (EVT) move with prudence in policy making. Most of all we need we need to understand model error before we use models. We go from reality to models, and not from models to reality.