THIS
As I have been pointing out repeatedly for a while, by making all the curves on that kind of chart start at N deaths, you are implicitly making a pretty strong assumption, namely that once they hit N deaths countries are at a similar stage of the epidemic.
But you don't actually know that! It *could* be that, if it's taken longer for one country to reach N deaths than for another, it's just because the virus started circulating longer in that country and/or because stochastic factors delayed the spread for a while.
But it could also be that, for whatever reason, something different is going on in different countries/regions and, even if the curves look similar when you make them start at this arbitrary point, the epidemic is going to unfold very differently. I think we just don't know.
In more general terms, this kind of decision just introduces degrees of freedom into the analysis, which people can use to make the data fit whatever their favorite theory is. I think it's best not to massage the data by making that kind of largely arbitrary decisions.
By the way, in my latest blog post, I make a related but even more important point.
