Yeah, interesting. I suppose the logic is that given that the mutation is random, it's primarily the capacity to spread that defines how prevalent it is in the population (I'm mainly thinking of its population here, not ours). In a pure numbers argument, it would be hard for a less deadly variant to become more prevalent, in part because the percentage of deaths is so low compared to the overall population (again, of virus, not hosts). Presumably it would have to spread faster/more easily and be less deadly/harmful to get a foothold. But one would still expect that probably the most benign strain would ultimately be the most prevalent, since presumably a greater percentage of asymptomatic cases would also enable easier transmission. But yeah, clearly I'm not a virologist!