FWIW, social distancing is a part and has been a part of these projections.The numbers keep improving because of the social distancing. Had we not aggressively pursued that the numbers would be different.
From the FAQ for the IMHE model:
How are social distancing measures used in the model?
The model uses the time from implementation of social distancing measures to the peak of deaths in locations where this peak has already been reached or passed in order to model this relationship for locations where daily deaths have not yet reached their maximum.
Initially, only Wuhan city had progressed far enough through its outbreak for this to be used in our model. An additional 7 locations have since reached or passed the peak of daily deaths (see updates April 5th) – many in less time than was observed for Wuhan city – and this broader evidence base is now used to forecast the date of the peak in daily deaths for each US state and countries in the EEA. The model includes the effects of social distancing measures implemented at the “first administrative level” (in the US this generally means the state level). We classified social distancing measures using the New Zealand Government alert system Level 4 and then assumed that locations that have instituted fewer than three of these measures will enact the remaining measures within seven days. With each model update, the assumption of full implementation of social distancing measures is reset; any delay will be reflected in the number of deaths, the timing of the peak of daily deaths, and thus the burden on hospital systems that the model estimates. The model does not yet explicitly address when or whether social distancing measures could be lifted.