Real doubling time (2 days or 3 days or 4 days) will not be known until testing rate catches up with infection rate. Until then, the rate of new cases will simply be a reflection of increased testing rate. In other words, more people found to be infected because more people were tested.
Which is why I keep saying you have to look at the deaths. The # of seriously ill is an extrapolation from that as is the # of asymptomatic (+ mild illness). So if the 40 deaths in Seattle are one cluster and we discount that we have 30 deaths. Extrapolate that to estimate the # of seriously ill, the # of asymptomatic and mildly ill. Watch how the number of deaths increase. If the number of deaths is not increasing in two weeks we can call it and see what happens.
15k new cases since yesterday. (Many in places that were supposed to be safe due to climate, incidentally, but that’s a different point.)
You’re right, it’s currently not the rate of spread we’re seeing, it’s the spread of testing.
The difference between even a 2 or 4 day doubling time isn’t going to matter much if there isn’t containment. It would saturate the herd in a short time either way and take out our HCS.
if deaths are accurately attributed (a big question given some stories posted here and reddit), then the death rate becomes a lagging indicator of general doubling time, as long as the point(s) where systems are overloaded and triage begins are noted in the data. I would think a short moving average of daily death tolls provides considerable data on the state of infection 2.5-3 weeks prior, as well as doubling time.
china numbers on this are fabricated and at one point followed a very simple equation exactly daily. italy appears to make effort to report real numbers, and hopefully continues to do so even when the numbers get uglier.