>If that's the definition, then that doesn't sound evil to me. It doesn't sound like informed consent is impossible.

It's impossible because the risks are largely unknown, so there's nothing much to be informed about. Anyone who wants to take on such unknown and potentially enormous risks is probably acting irrationallty, and it's unethical to exploit such people.

But this is where we get to the "written in blood part". The history of medicine is full of cases of people being horribly exploited in early stage testing. That's why we have the rules. Not because of abstract arguments about informed consent, but because we've seen what happens if we don't. It's all very well to imagine people giving informed consent to this sort of testing and imagine scientists acting responsibly and ethically in this sort of scenario; but we know that this isn't actually what happens.

>What's the thing similar to murder that is allowed? The equivalent of "FDA approved middle-stage testing"? Otherwise this analogy falls apart half-constructed.

The point of the analogy is simply to show that "the rules ban X" doesn't entail "the meaning of 'X' is defined by the rules". So e.g. if someone says "I don't fish in this lake because the rules ban it", that doesn't mean that they take the verb 'fish' to be defined by the rules in question. The same goes for FDA rules and 'early stage testing on humans'.

"Early" just seems like an extremely relative term to me, which is why I ask for examples. I guess not to you in this context. I don't think we're going to resolve this.