> That doesn’t feel like good science where you verify a hypothesis you have _before_ starting the experiment.

I'm pretty sure that's a requirement for starting research, to have a hypothesis.