Accidentally stopping extremely useful and completely safe genetic research with incautious bans on "gain of function" research would be a disastrous consequence of the very reasonable desire to avoid an accident that could cause millions of deaths.

However, what is lacking in your defense of such research is an account for why we would ever really need to perform a dangerous gain-of-function experiment. If we do not understand to any satisfactory degree the mechanisms by which a virus attains its level of virulence, we can learn an awful lot through loss of function experiments. Once we do understand, practically any gain of function experiment can be conducted in a loss of function background where only the critical step under investigation is assessed. For instance, if you want to understand the entry of viral particles into host cells, you can perform gain-of-function experiments on viral surface proteins in a background that is replication deficient. It does mean that you might have to use a more sophisticated assay to measure the effect--cell lysis or organismal death might not be a good indication any longer--but rather than requiring elaborate process-based safety mechanisms to be followed in order to avoid disaster, the experiments are safe by design.