[criticaltinker]

Unlike drawing a card from a deck, you don't need to choose drugs randomly. You can use existing safety and efficacy profiles to narrow the search - and in the case of SARS-CoV-2 you're likely looking for combinations of categories, e.g. antiviral, corticosteroid, and antithrombotic.

I agree though, the search space can explode quickly. Assuming Ivermectin is useless might be fair given the lackluster benefits seen in RCTs using it in single-drug regime. P95 is 2 sigma - typically in RCTs you'd be looking for stronger statistical significance, to reduce the risk of false positives as you allude to.

You certainly have a valid point, but hopefully you agree that it doesn't need to stop us from continuing to evaluate treatments using existing and widely available medicines.