Replies

Yes, that's what the IHME model assumes: a 9.5% daily growth rate on the incline, 4.5% on the decline. So, numbers ramp up quickly, but then dissipate over a longer period of time, producing a long tail and area shift to the right.

I don't have a problem with the non-linear curve/area calculation - I assume the resulting distorted bell shape probably reflects every actual epidemiology historical plot out there.

The key here, however, is how IHME assumes we'll ramp up to a peak of 2.9k in 13 days. To get there, we'd have to experience the type of % growth Italy recorded 18 days ago. That of course is during a period when they were still in the throes of trying to get a handle on the situation. If we use an 8 day lag, then the growth % drops way off because that's when Italy began to reflect the lock-down impact.

So, the big question is: Why assume the US is in a similar position as Italy 18 days ago? Because that's what's necessary for the 4/16 peak of 2.9k and 93k total by 8/1 requires. If we assume we're reasonably tracking Italy day for day (and even that is fairly conservative, because what if we're actually *ahead* in terms of pushing down the curve?), then we're going to be off both the peak and run out number.