The Diamond Princess cohort has 705 positive cases, of which 4 are dead and 36 serious or critical. In China, the reported ratio of serious/critical cases to deaths is about 10:1, so figure there will be 3.6 more deaths. From this we can estimate a case fatality rate of 7.6/705 ~= 1%. Adjust upward to account for cases that have not yet progressed from detection to serious, and downward to account for the fact that the demographics of cruise ships skew older. There are unlikely to be any undetected cases in this cohort.
You're right, adding deaths+.1*serious the way I did seems incorrect. But, since not all of the serious cases have recovered yet, that would seem to imply that the serious:deaths ratio is worse in the Diamond Princess than it is in China, which would be pretty strange. It's not clear to me that the number of serious cases is as up to date as the number of positive tests.
So, widen the error bars some more I guess?
This post is a container for my short-form writing. See this post for meta-level discussion about shortform as an upcoming site feature.