We cast the psychology of human-computer interaction (HCI) in terms of task analysis and the invention of artifacts. We consider the implications of this for attempts to define HCI in terms of a priori conceptions of psychology. We suggest that artifacts can be considered theory-like in HCI, and observe that they do play a theory-like role in the field as practiced. Our proposal resolves the current methodological perplexity about the legitimacy and composition of the field. We conclude that HCI is a distinct son of science: a design science. © 1989 Taylor & Francis Group, LLC.