In an immune suppressed individual treated with convalescent plasma, this treatment changed the viral population to favor a viral strain that was able to escape the antibodies.

In a related way, implementing large scale vaccination while large numbers of infected people interact with immunized people sounds like an efficient way to discover vaccine-escaping strains. This vaccine-escaping selection should be more than offset by the fact that vaccination will result in fewer infections and lessen the opportunity for more, worse strains to evolve.

Still, it would be better if we could avoid exposing a huge population of immunized people to a huge population of infected people. It's obvious everything would be better with fewer infected people, but it's still dawning on some that benefits include fewer strains and maybe more effective vaccination impacts.