Georgia's numbers for July 21

new cases today  3413  yesterday 2452     7 day rolling average  today 3574   yesterday 3572    total 148,988  trend is still up

deaths today 78               yesterday      3      7 day rolling average today 29     yesterday 21           total 3254 
I commented yesterday I thought there was a reporting error.  Even so, the 2 day average us 40.5.  That's the largest number reported since May 21.  One jump does not make a trend, so we'll say that it is steady
hospitalizations today 3255  yesterday 3183     7 day rolling average today 2990    yesterday 2917     trend is up

positivities   US 8.5   Georgia  15.1  slight decline.  Too soon to call a trend

DeKalb County has mandated masks while out in public.  Kemp's lawsuit against Mayor Bottoms is still pending (two judges have recused themselves), yet CEO Thurmond has deemed a mandatory order necessary with a warning on first offense and $250 fine on subsequent offenses,  DeKalb is the 4th most populous county in Georgia with 750,000+

I loved a sign I saw once:

NO Smoking on these premises.

If you are observed smoking, we will assume you are on fire and take appropriate emergency actions.

I know you don't put much store in statistics, but I thought you might find it interesting if we took a look at New York's (I couldn't find a breakout of just New York City) coronavirus trend since the day before (July 14) this article was published) to see if there was an identifiable upward trend.

On June 14 (the day before the article), New York had 694 new cases  with a 7-day running average of 745.
On July 21, -- 36 days later -- New York had 519 new cases with a 7 day running average of 725

The trend has been stable to down.  There was a spike on July 14 of 913 new cases. 

The 7 day rolling average went above 745 on July 16 -- 753 -- July 17 -- 751 -- and July 18 -- 755, then settled down. 

Based upon this information (see link below), the protests don't seem to have caused a noticeable spike in new cases in New York, although I am aware that you generally don't trust statistical data not your own.  I also suggest that had the protests caused a significant increase in cases, some media -- conservative, middle, or liberal --  (as you say, it's a low trust society) would have written it up, but I haven't found such an article  

https://www.nytimes.com/interactive/2020/us/new-york-coronavirus-cases.html

I use rolling or moving averages all the time in designing instrumentation and control algorithms to quickly react to sudden changes. Some of those algorithms were shamelessly plagarized from some stock and commodity market analysis code but I'm normally dealing with weight, displacement, voltage, current, etc.

However, those kinds of figures are pretty useful about covid-19 cases trending up or down, as long as the sample rate is also kept constant or the ratios track variable sample rates.

One thing that throws off trends is the delay in test results, or the lag from test to results, and what criteria are used to actually do tests.

If there is an upward trend in asymptomatic spreaders, there might be a decline in actual tests because people that don't feel sick may not go get tested, which lets actual cases spread under the radar.

The reverse happens if most of those contagious are actually getting sick enough to get tested.

But a long enough moving average, with a baseline that is about 1.5x the test to result interval, is a pretty good indicator of long term trends. It's still an approximation.

It's kind of like inferring piston ring blow-by and engine bearing wear by monitoring temperature, RPM, and oil pressure. With enough observation points and a long enough trend, one can notice that the oil pressure used to be 45 PSI at a 600 RPM idle at normal operating temperature a year ago, while this year it is only 32 PSI and noticeably moves about 3 PSI at the camshaft rotation speed, like some bearings are letting a lot more oil flow through than they did when everything was newer and tighter.

And monitoring the oxygen sensor outputs, MAP values, and gas mileage, which are widely different sample rates but normalized over a long enough interval, one can infer that the engine has less horsepower, lower compression ratio, and is burning more oil than it used to.

The only way to be really sure is to take out the engine, disassemble it, and measure everything, and of course at that point the crank goes to the machine shop to be turned 0.010 undersized, the block is miked out and probably bored 0.030 oversized, then new pistons, rings, bearings are put in, a new front and rear main seal, new cam bearings, new timing chain, new sprockets, new valve lifters, and the heads are fully rebuilt to fix any valve and seat and guide wear as well.

And once you put it all back together and stick it in the car, it ought to run like a new engine.

But indirectly one can get a pretty good picture of how the engine is wearing by measuring all the other stuff.

That's why I chose to use the 36 day span from the date of the article to the present.  It was long enough a period to spot any noticeable increases in new case numbers. 

In working with the numbers, I spotted nothing which appears wonky.  The numbers were within a limited enough range  throughout to give one confidence that the method of reporting was consistent -- no extreme ups, no extreme downs, just that one spike.   The 7 day rolling average varied only a little during that 36 day period, so I think one could safely and realistically state that the protests were not significantly spreading the virus.

https://www.nytimes.com/interactive/2020/us/new-york-coronavirus-cases.html

Georgia's numbers for July 22

new cases  today  3314    yesterday  3413    7 day rolling average today 3495    yesterday  3575    total 152,302
There was an article in yesterday's newspaper entitled "Testing for Covid_19 overwhelms Georgia" in which they report that some of the testing centers have had to close early because the testors were sick with coronavirus, and that labs are backed up by about a week, so I'm not sure how accurate the numbers are right now.  There is a three day decline in number of new cases, with the qualifications I've just stated.  Too soon to say there is a trend.

deaths today  81   yesterday 78    7 day running average  today 35    yesterday 29   total 3355
This number jumped from 3 on Monday to 78 on Tuesday to 81 on Wednesday.  I certainly hope this is not a developing trend.

hospitalization numbers were not reported.  There were 3155 on Tuesday with a 7 day running average of 2976

positivities   7 day rolling average   US  8.5   Georgia 15.1 

MerlinsDad said:

I think one could safely and realistically state that the protests were not significantly spreading the virus.

and the fraction of the population actually participating in protests is probably relatively small, and they are happening in fairly open areas. Then who knows - tear gas and smoke and flame from burning police cars and dumpsters might kill the virus pretty effectively.

rocketman says:  "Then who knows - tear gas and smoke and flame from burning police cars and dumpsters might kill the virus pretty effectively."

not to mention being beaten by batons and shot with rubber bullets. 

Georgia's numbers for July 23

new cases today 4286    yesterday  3616     7 day rolling average for today 3314    yesterday 3495     total 156,588  trend appears to be stabilizing in the 7 day

deaths  today 25   yesterday 81  7 day rolling average for today 37  yesterday 35   total 3360   trend is still up

hospitalizations today 3157   yesterday 3077    7 day rolling average today 3179   yesterday 3032    trend is still up

positivities   7 day rolling average  8.5   GA  15.1

and occasionally from being run over.