Why would you test something for efficacy before you're sure it's safe? You'll end up testing a bunch of harmful substances that don't actually do anything.

Maybe because the real standard is "the benefits out weigh the risks" you might want a drug that has a 10% chance of harm if it has a 90% chance of benefit, for a condition that is terminal if untreated.

Fair point, although it also seems nonsensical to test safety for human consumption before having a robust hypothesis for desired effects.

It's not true that safety is fast and easy to check.

Is there utility in knowing the lethal dose of a drug before you administer an efficacy study. Libertarians wonder…