I'm fine with the original, though I think putting the percentages on the bars does help remove any misunderstanding. If the aim is to show relative numbers of deaths for different age ranges perhaps a pie chart would be better?
The trouble is, for comparing between different ages ranges you want to zoom in to the region where the data diverges. Perhaps, a very obvious flagging up of the 15% would be better. It depends who your audience is though. Plus, although you don't want to be alarmist, you don't want to under sell the risk either.
I disagree with taking the percentages off. Otherwise you're implying that death rates are zero for anyone under 50. (It's also hard to tell if the death rate is zero without any comorbidities.) There's a balance between being reassuring and erroneously removing all fear together.
Andy, well done with this chart and presentation. We need to be more level-headed and less emotional as we attempt to make sense of Covid-19 and its global impact. Yes, Covid-19 is serious and we should be concerned but a combination of quantifiable data and sagely advice (wash hands, etc.) do more justice to the public than anything else.
But that was Alice's original point in her tweet. She thinks, reasonably, that too many people don't have the data literacy skills to realise that the bar in the original isn't representing 100%. I've seen this kind of error often: people misread bars because they don't read the axis. Also: I agree 15% mortality rate is very significant. If in this video I came across as saying "15% isn't very much", that was not my intention!
