The diamond princess' fatality rate was less than 1% - lower than the 1-3% used to drive early projections. Not to mention the fact that the cruise liner is disproportionately older, and the whole vessel recirculates its ventilation. This was, quite possibly, the worst population and location to have an outbreak and yet the death rates there were lower than the figure used for country-wide projections.

Effectively every reputable source is observing that the early death rate projections were substantially inflated. China's reported death rate is 0.5% - and they're the ones with the most experience with the virus. Germany only recent started random testing. They previously had a fatality rate of 2%, but after random testing they found that 14% of the population already had antibodies - thus lower the death rate down to 0.37% [1]. There's a vast difference between asking "out of the people who got sick enough to go to the hospital, what portion people infected with coronavirus died?" and "what portion of people infected with coronavirus died".