[UPDATE, April 16. This post was a bad call. Sorry.]
I am posting this at 11 AM eastern time, and I am about to go on a bike ride. By the time I come back, I may regret saying this. But I think that the worst is over on the virus crisis. The trend of the 3DDRR is going to be down, perhaps even faster than my optimistic scenario.
If I am correct, then many questions remain. The big one is whether things would turn bad again if we opened up more sectors of the economy. I wish we had a better way to test that other than experimentally lifting lockdowns.
UPDATE, 2:20 PM. a commenter suggests that New York might be distorting the figures. I took totals for the U.S. and subtracted the numbers from New York. Then I calculated the 3DDRRxNY, that is, the rate for the U.S. excluding New York. It is lower than NY. And also declining. I’ll keep my eye on 3DDRRxNY, but so far I stick with my call.
This number is basically New York. Florida, Texas, and Georgia are just about to come on-line. This is a big country. Don’t count your chickens.
Intriguing idea I’ve heard lately: why not piggyback coronavirus sample testing on the jury duty system? Either literally test everyone in each new group of prospective jurors as they come in, or use the same system that randomly picks people for jury duty to randomly pick groups to report for testing, and give them compensation, a letter for their employer etc as we do for those who report for jury duty.
It’s not completely representative of the whole population and would likely not pass ordinary standards of scientific rigor. But it’d be a heck of a lot more representative of asymptomatic adults than the samples we’re testing now, and it would use mechanisms that governments at all levels throughout the country already have in place.
RIP Corona
Traditional public health methods to control pandemics? Do case tracking/testing, quarantining people testing positive and taking extra care for vulnerable people. Plus, S Korea’s added mask wearing to the traditional list; and it’s proven very effective.
China, state governors and ‘Contagion’ screen writers seem to find lockdowns attractive, but the Washington Post reported on 1/24/20 that US academics outside of CDC “couldn’t imagine” the benefits of the general lockdowns.
It is possible there was a hidden factor in South Korea. Japan didn’t do nearly as much or as soon, yet isn’t as bad off as much of Europe. Also, India ought to be on fire by now; why isn’t it? One common factor: mandatory TB vaccination in the first two, & a significant number of immunizations in India (far from universal though).
Might be worthy of investigation
https://www.fresnobee.com/news/coronavirus/article241791356.html
Coronavirus cases in Fresno County rose by five more people on Sunday, county health officials reported.
The total number of confirmed and presumptive positive results for COVID-19 in the county stands at 108.
Travel-related exposures remain the most common at 28, followed by community spread (24) and person-to-person contact (18). Another 38 cases remain under investigation.
The only death in Fresno County was reported Friday.
—-
Spread out ag town of 500,000. We are tracking this like the 2009 flu so far, but we will be behind the curve. Our spread rate out here is much smaller than NYC as we have fewer mass gatherings of people.
Not a coincidence but NYC is now reporting more accurate breakdowns on a neighborhood by neighborhood basis. What controls the spread between neighborhoods?
The fast spreaders were mass transit, cops, nurses working on a city wide basis. But those fast spreaders have already flattened via immunity and careful controls. Thus they are keeping the virus local to neighborhoods and likely targeting neighborhoods the are rising. It is being bottle up.
Hence, the distribution of neighborhoods will rarely see adjacent neighborhoods both have a sudden rise in cases. The one the other neighborhood will get extra lock down first. The effect is to make the arrival of the virus independent across neighborhoods and the distribution of the neighborhoods seeing a rise is very left centered, at most we get zero or one neighborhood rising and they are almost always over run by cops increasing lock down and the immune population.
https://www.politico.com/states/new-jersey/story/2020/04/06/rutgers-camden-coronavirus-tracker-suggests-new-jersey-curve-starting-to-flatten-1272598
Rutgers-Camden coronavirus tracker suggests New Jersey curve starting to flatten
—
I am almost certain that this is all about shutting down the fast spreaders, the mass gatherings.
I really appreciate Arnold giving us a framework to discuss this issue. I’ve turned his equation around:
3DDRR = 3DDRR = (C * r / D) ^ 3/n
In order for 3DDRR to fall any or all of 3 things on the right side have to fall
C the number of true cases (Transmission rate R0 could be falling). r could be falling because those most prone to death from the virus have quarantined themselves better. n could be increasing (the exponent falling) because those quick to die have died or have better quarantined themselves.
I think there may be limits to the effects on r and n from current policies. Removing the policies could cause r and n to reverse and cause havoc again. If r and n approach stability (constants?) then falling 3DDRR means falling true cases.
“An estimated additional 180 – 195 deaths per day occurring at home in New York City due to COVID-19 are not being counted in the official figures.” From Worldometer. Is this factored into your calculations? (People dying at home are not being counted in NY. I read that they are not counted in France either so the same kind of undercount applies to France vs. e.g. Italy.)
Including those deaths will affect the prediction of True infections. However, their inclusion will not affect the growth rate of infections unless the under count changes over time. The growth rate in cases, not the level of cases, has Arnold’s interest and mine.
Mark Levine, chair of the NY Health Committee, tweeted yesterday morning that…
This was included in a broader Twitter thread, which has more context.
The NYT reported on Sunday that this is happening nationwide.
I think this affects 3DDRR’s product (i.e. today’s 3DDRR is likely higher than 1.5). But I don’t think this undercuts 3DDRR’s usefulness, unless underreported deaths are increasing more quickly than reported ones.