Furthermore, IHME has underestimated its own uncertainty. Although I am a mere policy analyst, I do know that when IHME offers 95 percent prediction intervals, then the actual values are supposed to fall outside those intervals only 5 percent of the time. Applying that standard, critics investigated how the IHME model has performed on what should be one of its easiest tasks — predicting the number of deaths that will occur the very next day. They found that over a four-day period, the actual number of next-day deaths in each state fell outside the model’s 95 percent interval about two-thirds of the time. The failure is self-evident. One need not have a “messianic complex” to reject this model’s predictions.
Finally, even if “all of the models say this is going to be one of the worst epidemics,” that is hardly the end of the policy debate. It can be simultaneously true that we face a terrible epidemic and that full lockdowns are an overreaction. The real issue is how far our mitigation attempts can go before they are no longer worth the economic and social costs. Even people who “understand science” might conclude that those attempts have already gone too far.
