In an exponentially increasing epidemic, there are 2 factors at work. One, like you say, is that there is an unknown number of undetected cases. The other is th

In an exponentially increasing epidemic, there are 2 factors at work. One, like you say, is that there is an unknown number of undetected cases. The other is that there is a lag time between detection and resolution of up to 6 weeks. Comparing the known cases today to the resolved fatalities from several weeks ago doesn't work. The undetected case issue can be addressed by intensively studying smaller populations and extrapolating to the general population. The Diamond Princess or other cases where every person in a defined population was tested and monitored would be a good place to start. The exponential growth issue can be solved with cohort analysis, group people by the date of their onset of symptoms and track the resolution for each group. With a few more weeks of data we should start to see these studies, then we'll know just how serious this is.