Is understand that he estimated the number infected on 3/12 by using the number dead on 4/4 and working backwards based on the time from infection to death (23 days) and the assumed mortality rate (1%). Even though I am more optimistic than the 1% assumption, I’m ok with it. He is assuming that the number of infections doubled every 5 days from 3/12 until 4/4. IF this is the case, then the number of deaths will follow exactly linearly in his model, beginning with a doubling of deaths in 5 days after 4/4 on 4/9. This means that on 4/9 we will have 16,952 total deaths (doubled from 8,476 on 4/4 since the infections doubled from 84,760 to 1,695,200 from 3/12 to 3/17). Deaths would continue as follows:

4/4: 8,476
4/9: 16,952
4/14: 33,904
4/19: 67,808
4/24: 135,616
4/29: 271,232. (27,000 deaths/day at this point)

If you were to keep going his model would see the entire US population of around 350 million dead around June 15th.

My point is that even the most reputable models are projecting a peak deaths/day of 3000, whereas this guy is projecting 27,000 deaths/day at the end. Doubling time is a moving target that very well might have started at 5 days but is much greater now. Clearly he is wrong (which will be proven in the coming days)