The ethical tension in research design is often characterized as that between individual and collective ethics. While adaptive clinical trials (ACTs) are generally considered to be more sensitive to individual ethics, the concomitant loss of statistical power associated with them is often used to justify randomized clinical trials (RCTs). This paper challenges this characterization of the central ethical problem in research design. It argues that the key consideration in clinical research hinges on the process of informed consent. When the research context is such that the subject is able to provide informed consent, RCTs can be justified and may be required. However, in desperate medical situations the process of informed consent is often undermined. It is argued that in such situations ACTs are ethically required. We introduce "the principle of interchangeability" and argue that it must be satisfied if research in desperate medical situations is to be justified.