There are a few hypotheses suggested by the higher-count events in VAERS, as well as the standard ones checked in postmarket studies (effects on pregnant women and children in particular). I agree that the mechanism is a good way to do hypothesis generation, but it's not a solid argument for the safety of a treatment. It is really only an antidote to specific, hallucinatory claims, like the line about it "changing your DNA."