Post #1009: Degenerate curve.

Posted on February 15, 2021

Source:  Calculated from NY Times Github COVID data repository, data reported through 2/14/2021.

The seven-day moving average of new cases per day is now down 64% from the 1/8/2021 peak.



Technically, a straight line is a curve.  For want of a better term, it’s a degenerate curve.  A curve in name only.  So that last straight-line bit is our pandemic curve, for the time being.

I’m not exactly unhappy with the curve, as it stands.  Would not bother me a bit if this were to go on for a while yet.

But we’ve coming up on the five-week anniversary of the 1/8/2021 peak.  And it’s not curving.  Not even a little bit.


And it’s not merely that it’s a straight-line decline, it’s that it’s a nearly-uniform rate of decline across the states.  By eye, at least, none of the differences across the states matters in any systematic way.

COVID hygiene levels, mask mandates, existing base of infected individuals, vaccination rate, population density.  None of it matters, so far.


But that’s a pretty narrow view of it.  Heck, if I don’t label the graph, you will be hard pressed to identify the U.S. COVID peak from that of Canada, Great Britain, or Japan.  All started virtually the same day, all have nearly the same slope.





Yesterday’s Washington Post article, and countdown to the Super Bowl COVID-19 explosion.

As you know, I’ve been examining and analyzing this downturn for some time now.

Near as I can tell, the explanation of recent COVID trend is “seasonality”.  Which means, there’s no explanation at all.  It’s just a way of saying that it’s the time for COVID-19 to recede in the northern hemisphere this year.  No different, really, from the way that flu will go away, more-or-less all-at-once, at the end of each year’s flu season.

It’s as if God flipped the COVID-19 off switch sometime before Xmas, and the earth’s northern hemisphere has been coasting ever since.

When I read yesterday’s Washington Post article (Four reasons experts say coronavirus cases are dropping in the United States”), I was aghast for two different reasons.

First, it was clear that none of the experts had done any sort of detailed look at the data.  They were winging it.

Second, the level of ignorance of the commenters was appalling.  Which is to be expected.  But what was so shocking is that people will simply believe whatever gets repeated in the media, regardless of the lack of evidence behind it.  (If you think that sort of thing is limited to Q-Anoners and MAGA-heads, think again.)  And so, the myth of huge post-holiday surges was prevalent, much referenced, and much embellished.

None of the experts even thought to bring up the lag between infection and reporting.  Which — with the lags between infection, symptom onset, seeking treatment, reporting of test results, and then a seven-day moving average —  works out to be somewhere between 16 and 21 days, best guess.

That, by itself, immediately rules out most of the supposed “reasons”.  This linear, un-wavering decline in new infections started well before Christmas.  (See Post #989, because I have to keep reminding myself about that.)  It only showed up in the statistics on Jan 8, due to the lag between infection and the eventual reporting of positive tests.

Which mean this decline in infections began about a month before the Biden administration took power, and well before any material fraction of the population was vaccinated.  (Even ignoring that it takes six weeks for vaccines to be fully effective).  And, needless to say, that decline in new infections was occurring literally during the period when the news media (and the CDC) were hyping the huge Christmas surge in cases.

So, just by paying attention to that single detail, we can eliminate three plausible causes of the current decline.  And nobody thought to do that.

Then, by looking at the incredible uniformity of the rate of decline across U.S. states and across many widely separate cold-climate northern hemisphere countries, we can eliminate — well, pretty much all other explanations as well.

This sharp turnaround in new case growth is not due to (e.g.) a sudden change in COVID hygiene.  (And, objectively, things like propensity to wear a mask did not change materially over this time, nor did the cross-sectional differences in that change, which you may check for yourself at the Carnegie Mellon CovidCast project.)

And, to my chagrin, there’s no evidence that this decline is occurring because the U.S. is nearing herd immunity.  There’s no apparent difference across states, despite large difference in fraction of population diagnosed with and vaccinated for COVID.   And the states with highest fraction recovered or vaccinated are behaving no differently from others.  No matter how badly I would like to see that occurring.


And so, the very premise of the Washington Post story — Four reasons” — is just wrong from the get-go.  There is no cause-and-effect reason for this decline.  You give me a supposed reason, and I’ll easily show you why that’s not a good explanation of the facts.  (The emphasis there is on easily, meaning, none of the expert talking heads appears to have done even the slightest bit of homework before opining.)

Even for “seasonality” as a reason, that wasn’t defined, and so of course every citizen in the comments thought they knew that meant “cold weather”, when in fact, it just means that the disease has a defined season.  It became yet another term that everybody thinks they understand, but almost nobody understands what it actually means in this context.  (And of course, none of the commenters who conflated “seasonality” with “cold” ever stopped for a millisecond and said, hey, what’s happening in Hawaii?)

As an aside, I think that low-humidity indoor air is a significant risk factor for spread of respiratory infections.   Enough so that I’m running humidifiers constantly this year, and sent one off to college with my daughter.  And so, in terms of causality, I’d argue that low indoor humidity has a role in triggering flu season every year in the temperate climate zones.  But what brings it crashing to an end, all across the U.S., still has to be attributed to “seasonality”.  Which is another way of saying, that’s just the way it works.  See Post #996 for explanation of graph.

Based on the writing, and the comments, I think it’s only a matter of time before this entire decline is attributed to the impact of vaccination.  As long as the media increasingly cite experts who are willing to keep saying that, that’s what the popular belief will become.

And so I continue my lonely watch for the great post-Super-Bowl explosion of 2021.  Which I’m betting will not happen.  Because, contrary to popular belief, there was no post-holiday bump in cases for any of the prior major U.S. holidays.  The only thing I’m convinced of is that the whole idea of the much-hyped post-Super-Bowl new case explosion will have been forgotten by the media before those (hypothetical) cases should begin arriving in the data by the end of this week.

It’s a tough job but … .  Well, no, it’s not a tough job.  It’s more like I’m retired and have nothing better to do.  But I’m still going to carry through with it.  Maybe this time — because this occurs against a backdrop of declining cases — maybe this time we’ll have somebody in the U.S. media notice that there was no post-holiday surge in cases.  I’m not holding my breath waiting for it.