Concerning those 3% to 7% estimates, we won't know exactly what this means until we have the published paper. Even then we will have a respected researcher's opinion, expressed in numbers, which will need to be supported or modified by much more work.

Don't put too much confidence in numbers, especially when dealing with small samples and laboratory techniques that may still have systematic errors. Statistical confidence bounds are based on having systematic error identified and under control, so that you can use error distributions characterized by only two numbers: mean and variance. We are scarcely at that point. Our ignorance is such that we simply don't know what distributions are appropriate, and so use the least demanding.

In most such instances, there are shortcomings of test assays. In this case, we have plenty of evidence that it is easy to get false negatives. As tests improve, this alone suggests percentages will rise. A second factor is that we are learning about when and where to look for this virus in human patients. Improvements here will have the same effect as improved sensitivity.

Finally, we are at the early stages of identifying the particular subsets of the general population which show higher rates of incidence. Experience shows that having a better idea of these subsets gives you a much better chance of estimating overall rates of incidence from small samples, because you are less likely to overlook a small group which will dramatically change the general result.

These arguments, taken together, strongly suggest later estimates of general incidence will be higher. This is highly troubling to officials trying to craft a response to a crisis. Even a 3% rate translates into about 10 million infected people in the U.S. A 7% figure would push the number over 20 million. If the factors mentioned above make even that conservative, we could be looking at 30 million.

How you deal with an epidemic of lifelong infection of such proportions is a very good question. We need reliable tests implemented yesterday. We need to understand transmission. We need ways to stop or slow the spread. We need effective treatment options.

This is a wicked problem to hand anyone, even without a legacy of mishandling.