One persistent confusion about estimands is the notion that, by stating a target population, we are somehow assuming that studies (trials in particular) involve random sampling from that population.