Testing... testing...

So, for those insisting that we need to test everyone for the Pestilence before resuming any semblance of normal economic activity:

What's the false-positive rate on this test of yours?  Don't tell me it's zero, because it isn't.

Once you've hastily scaled up production to make and distribute something north of 330 million test kits (just for the U.S.!) all at once, plus the equipment and supplies for processing the samples expeditiously, what will the false-positive rate be?

When you deploy enough sample-collectors (of varying skill and fatigue levels) to get everyone tested (at collection points and labs of varying quality) in the space of a few days, what will the overall, sample-collection-to-result, false-positive rate be?

Multiply that by the number of uninfected people - we believe this to be at least 325 million, right?  What's that come out to, now?

Now what?  Do you round up all those who test positive and send them off to leper colonies for a month, thereby ensuring that all the false positives will become true positives?

We really need to stand the usual rule on its head.  The standard now appears to be: he who relinquishes power is obligated to have a detailed plan for everyone.  It should be: he who takes power is obligated to have a detailed plan for the consequences of the policies that he imposes.

Addendum:  And what's this about immunity passports?  I ain't no theologian, but this has a bit of a "that no man might buy or sell, save he that had the mark" quality to it, doesn't it?

Comments

[Rob]

I don't think you need to test everyone but we ought to test a LOT of the people, people from all over the country. It's hard to make accurate guesses without good data.

[Eric Wilner]

A large random sample from all over would, in principle, be a good thing - preferably testing both for current infection and for antibodies - BUT we need to know, with decent accuracy, what the limitations of the test are.
If we don't know what inaccuracies the test and the associated processes introduce, we'll get a large amount of data but it'll be wrong. Which is pretty much where we are now: lots of data, with enough contradictions that most of it must be wrong.
Now, here's a cute idea: are there frozen blood samples, from whatever sources, spanning the past few years? Might be interesting to run the antibody test on them - if it comes up positive, either the virus has been around longer than we think or there's something wrong with the test.