Today the U.S. reached 38.9 new cases per 100,000 population per day.
That’s just 0.5 cases higher than yesterday, and reinforces the idea that growth in new cases is slowing down. The current 26%-per-week growth rate for daily new cases is half of what it was two weeks ago.
That slowdown in case growth is concentrated among states with high case counts. And that, in turn, brings down the U.S. average growth, because most of the cases in the U.S. are in that handful of states with high case counts.
By eye, it looks as if the U.S. peak ought to be in early September, which would be reasonably consistent with with the “9 weeks rule” outlined yesterday. At which point, we can start planning for the winter wave.
Details follow.
Data source for this and other graphs of new case counts: Calculated from The New York Times. (2021). Coronavirus (Covid-19) Data in the United States. Retrieved 8/13/2021, from https://github.com/nytimes/covid-19-data.” The NY Times U.S. tracking page may be found at https://www.nytimes.com/interactive/2020/us/coronavirus-us-cases.html.
The growth is obviously slowing, as show by the downward curve of the lines when graphed on a log scale. The same is obvious when I simply graph the growth rate. Weekly percent change in new cases is in blue, weekly absolute change (count of cases / 100K / day) is in orange.
I might as well do the crude-but-obvious thing, and literally sketch in the future. Below, I’ve tacked four more weeks onto the national curve, and drawn in a plausible-looking continuation of the U.S. curve from the first graph above. By eye, assuming a nice, smooth curve, we’re maybe two or three weeks from the top. September-first-ish. (Ignore the thin blue line – I just needed that to get Excel to give me those four additional weeks.)
Here’s an oddity, though. The slowdown in growth is concentrated among states with high current new case counts. I think that’s pretty obvious when I contrast the curves (in logs) for ten high-case-count states, and ten low-case-count states. The lines curve in the top graph, but not in the bottom one.
These lines curve at the end:
But these lines don’t:
Maybe that’s not so odd after all. That suggests that growth is slowing because they’re running out of people to infect. At some point, you’d think that you’d start to exhaust the population of people who lack immunity but who are out-and-about, unprotected.
You have to be a cautious with that armchair epidemiology, however, because the high-rate states are so geographically clustered. Basically, the very highest rates are clustered in the South and South Central regions.
If I plot the states by region, it looks like there’s a fairly strong geographic component. Note that all the states in the South Atlantic region are showing signs of slowing case growth. That, even though some of them still have relatively modest numbers of new cases / 100K/ day.
By contrast, if I look at, say, the Northeast, it’s still mostly straight lines, no slowing of new case growth. So the slowdowns are very geographically clustered. Is that due to the level of cases, or the weather?
A third possible explanation of the clustering of the slowdown in growth in certain states is the clustering of the start of the Delta wave. For example, in the two graphs just above, note that the Delta wave started around 1.5 weeks later in the Northeast than it did in the South Atlantic region. If this is all driven by the “9 week rule” — if each state’s Delta wave lasts more-or-less 9 weeks — then we’d expect to see the Northeast region start their slowdown in new case growth in another week and a half.
Unrelated footnote: A shout-out on technical assistance.
Used canning lids, straight off the jar (left), after 20 minutes of boiling (right).
This website does not use Secure Sockets Layer (SSL) encryption. It’s a plain-old HTTP website, not HTTPS. There’s no reason to use SSL. It’s not like I’m asking for your credit card information.
That said, two readers alerted me that their browsers were throwing “expired security certificate” messages when they tried to access this website. I couldn’t replicate the error, and I couldn’t imagine how I could get an SSL error when I didn’t use SSL.
As it turns out, I did in fact have an expired SSL certificate. It came “free” (that is, prepaid) with the blog host that I use. Mine had finally expired, and that tripped the warning even through the website itself does not use SSL. I asked my website host to get rid of the expired certificate, and everything should now work correctly.
I’d like to thank those readers for alerting me to the issue. I’d never have known otherwise.
There’s an odd little addendum to the story.
A lot of people read this blog for the running commentary on COVID-19.
But in fact, the overwhelming majority of people who visit this blog come to read about jar lids used in home canning. By far, my most-read blog entry concerns the U.S. shortage of canning lids (Post #G21-003).
I’m pretty sure that’s because almost nobody in the U.S. treats that shortage as something serious. News coverage tends to be along the lines of “isn’t that quaint”. And I imagine that if your produce is rotting for lack of canning lids, that sort of treatment is irritating and unhelpful. In any case, that blog post shows up among the first few entries when you Google “canning lid shortage”.
After months of getting hundreds of readers a day for that one post, volume had finally started to decline. I figured that either the lid shortage was easing (not!), or canning season was underway and people had found their workarounds.
But now I realize that readers might have been turned away by the SSL certificate error. These days, with hacked this and spyware that, I think long and hard before I’ll click through that warning to visit a site. I imagine that most people would feel the same about clicking through that to reach this site.
In any case, if you’ve gotten that warning in the past couple of weeks, that’s the reason. I’ve never configured SSL on this website, but I did in fact have an expired SSL certificate. With any luck, that’s now been fixed.
