I disagree with a lot of what your cousin said. First off, you don’t randomize to create equal baseline severity, you do it so that there isn’t an unknown factor that disproportionately affects one group more than the other. Second, if hydroxychloroquine worked, we would likely see some signal of benefit in the well-designed studies, and we have not.
Because there has been no benefit seen, a very small, but loud, group has moved the goalposts - well, it needs to be started earlier (first 4 days, first 24 hours, prophylactically, etc) or it needs to be combined with azithromycin or it needs to be combined with azithromycin and zinc, etc.
One of the arguments of the Yale scientist is that the death rate dropped in a region of Brazil around the time they received a shipment of hydroxychloroquine. Obviously a ton of other factors could have influenced that, and actually the death rate also dropped in the US when we greatly decreased our use of hydroxychloroquine.