That site is from the U of Washington, the same folks I've posted models from for the past few weeks, as they've seemed to have more realistic modeling scenarios. They keep saying they're going to update their models, but they've delayed it all weekend.
My guess is they're digesting all the new data, especially the observation that we might already be into the peak number of cases in NY and possibly NJ, meaning the peak deaths will come sooner and likely be at a lower peak than modeled so far, and that they're going to come out with lower estimated deaths for the US/NY/NJ than the ones below, which were from 4/1. Maybe down 20-30%, to perhaps 70K for the US vs. the current 93K (I was guessing maybe 50K). Or maybe there will be minimal change, since they're seeing more than I am, I'm sure.
We'll see - this is not too different from when I try to guess where the NWS is going to go based on new data that we all have access to, but I can be nimbler with the analysis than they can. Of course a huge difference is this is predicting deaths, which matter far more than the weather, so if they need some more time, they should take it.
Edit: turns out the first three graphs, below, from last night, were a mix of projections. The NY and US graphs were from the model run on 4/1, while the NJ graph is from the model run on 4/5 and the 9600 deaths shown are much higher than the ~2200 deaths from the model run on 4/1 (not shown). I added the 4/5 model projections for US/NY after these three. You can tell when the model was run from where the orange solid line ends and becomes dotted, also.
Edit: adding in the 4/5 model runs for the US and NY, below. The US projected deaths dropped from 96K to 82K, which is good (not the 70K I thought it might drop to) and the NY deaths projected dropped from 16.3K to 15.6K.
