Also elsewhere in this thread it's mentioned that the Florida and Santa Clara results could be entirely explained by high type II error in the test. The Florida test appeared to have a false positive rate of ~15% when independently validated, which is basically the infection rate they found. In other words, this is a specific form of base rate fallacy where the test accuracy is really low.
It seems pretty likely that the data will come out showing at least 10%, so it's literally impossible for it to undercount by an order of magnitude.
How do you think a random sample of inhabitants would be off by a whole order of magnitude, anyway? Can you explain the mechanism whereby that might happen? The only thing that comes to mind would be using a worthless test with a 90+% false negative rate.
* It will also show an undercount of at least an order of magnitude.
* Commenters will still pop up to explain why the results can't be trusted and which further studies are absolutely required before we believe them.