So... here we are three weeks later, and I'm not writing this nine days early to tell you the good news.

We haven't passed these predictions (which were by my model actually due June 6), but we'll pass the cases number by lunchtime tomorrow. Deaths soon. https://twitter.com/ConceptualJames/status/1256414194859155456

My model isn't meant to be robust or even strongly predictive. I actually made it to see what was going on and maybe to track a stat with some sensitivity to changes in circumstances. It operates on a very simple assumption, in the short term, local, viral spread is exponential.

I'm tracking three geographic regions only: USA as a whole, TN state, GA state, for reasons. I don't know the details of any other state or any county.

I can't tell you what will happen going forward from here, but I can tell you some things pretty certainly. Here they come.

1) The overall shape of the curve of the stat I'm tracking indicates that we're roughly following a Gompertz curve with the virus numbers, which is what one would expect. I don't know about the beginning phase of this because there's clearly the impact of testing taking off.

2) There's an unmistakable bump in the death graph right when NYC was being reported to be out of control. The healthcare system there might not have been overwhelmed completely, but it was partially. There's a bump in cases too at the same time, but that's probably testing.

3a) A Gompertz function is, in some sense, an exponential function with an exponentially decaying growth rate: f(t)=A*e^(a(e^(-bt))t). The graphs in all three regions I'm tracking roughly bear this shape out, but it's bigger than the one I predicted on May 1.

3b) That means the assumption stated in my May 1 tweet, *if we maintain equally effective suppression methods*, was not maintained. This is generally obvious to anyone looking outside or going anywhere, but we're a full 17 days ahead of that schedule due to softened measures.

4) It's still too soon to tell for sure, but it looks very much like all three graphs have varied by failing to exponentially decay to zero, as one would expect, and have instead decayed to a lower constant growth rate. If so, threat may remain significant for a long time.

4') Technically, it's almost certainly clear that Georgia's graph hasn't leveled off but has actually started rising again. Can't be totally positive, but.... It is not clear in the cases of the USA as a whole or TN that this is the case (and I'm tracking no other states).

5) All three of my graphs have converged to the relevant current rates for both case growth and death growth having very close to the same numbers. That means changes we're seeing in case numbers now are probably not "dirty" and reflect a proportion of true infections.

Verdict: We could be doing worse, but because we've left a strict, usual Gompertz fit, I'm not really sufficiently tooled or informed to make any cogent predictions for the weeks or month(s) ahead with regard to numbers.

Disclaimer: Just mathematician, toy model.

Oh, I also don't have any idea why the numbers have stayed so relatively low given the rather incredible amount of loosening of measures (I almost never see people except employees wearing masks here, for instance). Weather probably not. Vitamin D going up maybe?

I'll gladly admit I had given up on the tracking project until the end of April when states I thought rather cavalier (GEORGIA) in their decision started opening back up. I wanted to see if anything was happening. Answer: way less than I expected, but something.