It is 2% dead if you assume that all today infected will recover in the end. It is much more pessimistic when you do not have such assumption but project current dead to number of infected about 2-3 weeks ago.