Like many people, I find the current projections of the @IHME_UW model absurdly optimistic, especially the further out uncertainty bounds. And, like many, I had similar thoughts in April/May 2020. So, I looked at performance of the models over time… @IHME_UW These plots show actual reported deaths with mean reference model projections (each up to ~100 days ahead), with and without a log scale. Performance has been relatively better since Summer 2020, after all those projections to 0, but there are still notable misses…