Apparently, its origin is unclear, and its usage polemic. So this is a thread about its origin, why, how it’s used, and a lesson about processes vs. goals.
Back in March, ppl had no idea what was happening. They took cases at face value. One of the big goals of the article “Coronavirus: Why You Must Act Now” was to highlight how official cases was meaningless.
Then, ppl realized cases were not the entire picture. Testing was crucial too. No tests, no cases — but lots of hidden infections. So they started reporting cases and tests.
But these are meaningless numbers in a vacuum, so they sought a ratio.
The 1st one ppl recommended was tests per million ppl.
That made no sense to me. It meant testing only depended on your population and not on the size of your outbreak. So very early on (March 10) I already calculated positivity (“positive test rate” I called it)
In April, everybody was reporting tests and tests per million, but nobody was reporting positivity. Only on May 12 did the WHO start recommending 5% positivity
So in April there was no guidance on how to judge the quantity of tests: neither a single valid metric, nor what to aim for. That’s why we came up with one.
At the time, it didn’t even have a name, so we didn’t use one. But we quickly suggested around 3%.
Why?
Because that’s what good countries were doing.
Not a single country that was controlling the virus hd more than 3% positivity
We debated long and hard what the number should be. At the end of the day, we stuck with 3%. Here’s why:
You’re either in a path to control the epidemic or you aren’t. If you are, your positivity should be going down slowly as cases go down and tests up.
But if you aren’t, your positivity might be 1% one day and 3% the next. You want an EARLY warning that you’re going in the wrong direction. 3% is earlier than 5% or 10% (other thresholds used) but is substantial enough that no country with good #COVID mgmt would have passed it
So 3% is a good threshold for general alert about the status of the epidemic in your region.
That’s why I suggested the team at COVIDActNow to use 3% on their website, which they did, and I believe that started circulating afterwards & becoming the standard.
Since then, they’ve researched it further and reached the same conclusion.
This is my narrow perspective of what happened. I’m sure other things were happening at the same time that I was missing.
What that tells you is 3% shouldn’t be seen in a vacuum. It needs to be in context with the trajectory. But that’s a nuance most ppl will miss. To keep it simple, you need a single number. Hence 3%.
But is this a good threshold for school closure? Here’s my pbm with that.
Ppl bundle all children in 1 bucket. By now, we shouldn’t. Childcare kids are least at risk of contagion and falling sick. They’re also doubly valuable in school: education & parental work
Pre-K should be the last thing to close. And after that, kindergarten, elementary, and middle.
Only high school and higher Ed should be considered for closure, since they can become superspreaders.
With all processes, ppl lose sight of the goal they were trying to achieve, and start following the process mindlessly. This is a good example.
Instead of just closing all schools at 3% positivity, govs should do all they can to prevent getting there
When they do get there, they should start closing clubs, bars, restaurants, gyms, hotels... in that order. At the bottom of the list should education be. We should, as a society, strive to keep educating our next generation, so they’re less stupid than us.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
We’ve been lucky though. In the 1918 pandemic, the 2nd wave was likely driven by a mutation that was both more infectious AND fatal.
We already knew this was happening back in March. This image is from The Hammer and the Dance. The only thing we didn’t know then is which variant was going to prevail. Details.
Capitalism vs socialism, markets vs gov... Most ppl think 1 is great and the other trash. That’s simplistic.
They’re tools adapted to different situations. We must understand them to know when to use them. Thread.
[1/
Capitalism is great. It uses natural selfishness to push ppl to be as productive as possible, promising them wealth. The + you produce for others, the + you get.
That is achieved by incurring both the cost and benefit of your initiatives.
Here’s the pbm
[2/
It creates a huge incentive to increase your benefits in ways that worsen society.
This happens in many ways. Eg:
1. Information asymmetry
You want cheap & delicious food. But what if it has ingredients that cause cancer? The producer knows it, but doesn’t tell you.
[3/
One of the key arguments of Herd Immunity apologists like @ScottWAtlas or Anders Tegnell is that you can't stop the virus. That means it only stops killing people when 50%-80% of the population has caught it (66% in Manaus). nature.com/articles/d4158…
If it had taken us 5 years to get a vaccine, it might have made sense: it might be too hard to control the virus this long. But now we can guess that by mid-late 2021, enough ppl might be vaccinated to stop it.
As a Stanford University alumnus, I am appalled at @ScottWAtlas's defense of herd immunity, and disappointed at @Stanford's response. But the nuances are important to get right. Thread:
First, @ScottWAtlas. The @washingtonpost reported his herd immunity position. The quick idea: (a) It's impossible to stop the virus, (b) it doesn't kill that many people anyway, (c) Sweden has succeeded letting it run, (d) lockdowns are too expensive
[2/