Clear-cut designs have a number of methodological virtues, with respect to internal and external validity, which I illustrate by means of informal causal analysis. In contrast, a more uniform experimental practice across disciplines may not lead to progress if causal relations in the human sciences are highly dependent on the details of the context.