> That’s why we’ve generally observed later variants of some viruses to be less fatal. It doesn’t do a virus much good to kill its host.

While this is somewhat true, there are important limitations to this model. If a virus has already done enough spreading, from an evolutionary perspective it doesn't matter much if the patient dies later. Another factor, observed in Delta variant, is that having a respiratory virus that is more contagious might also mean it is more deadly, because it is disrupting more cells to produce more viruses.