(I have very little biology knowledge, so treat this with a bag of salt.) My understanding of the protein spike mutations is that they consist of a number of d

(I have very little biology knowledge, so treat this with a bag of salt.) My understanding of the protein spike mutations is that they consist of a number of deletions - and deletions that have shown up repeatedly in a viral genomics database (convergent evolution across related viruses). These seem to me to be a number of host (human) specific mutations that are actively selected for when one of the spike sharing viruses jumps to humans. The fact that these mutations are deletions (are they?) means there's a very real limit to the degree of change - there are only so many human specific optimisations that can occur, and once they are made it should be significantly more difficult to reverse them. Once all of the optimisations have been made, the virus's attack vector will be left static, and it will be left without the protein structure that it depended on to infect the species that it jumped from. The virus won't be completely backed into a corner, but it will be limited to a significantly reduced rate of change - which should make it easier to control. So how wrong am I?