Projections based on the previous few days reports don't necessarily predict where things stand

[onenote]

There is a lot of reporting based on signs that the epidemic is abating in some places based on declining numbers of cases and/or deaths. But suggests that such "trends" are reliable indicators that the virus is declining should be taken with a grain of salt. No place in the US has a longer timeline with respect to the virus than Washington State. Here are the reported number of deaths since March 20:

3/20 -15
3/21-5
3/22 - 1
3/23 - 15
3/24 - 13
3/25 - 9
3/26 - 15
3/27 - 28
3/28 - 20
3/29 -0
3/30 -0
3/31 - 52
4/1 - 15
4/2 - 22
4/3 - 26
4/4 - 28
4/5 - 34
4/6 - 22
4/7 -27

[Igel]

1. Their numbers are small.

Hence a small change in numbers can be a bit disconcerting.

NY's numbers are larger, but still have some built-in problems. Not the least is that the numbers we hear each day are the deaths reported, not the deaths that occurred. https://covid19.healthdata.org/united-states-of-america/new-york shows a dip, which people took as meaningful.

Smoothing the curve is, at a minimum, required. But the IMHE numbers could easily be skewed--that dip for the lasts few days followed by a sharp rise has to influence the predictions.

IMHE has also lowered their predictions. People'll complain if they're under, saying how unreliable they are, but had nothing much to say when the numbers were too high. It's a prediction with certain assumptions that are unchanging.

I need at least 5 days downward trending numbers, with most of the day-over-day numbers showing a decline, before I'll start thinking "that peak's past".