This article discusses past developments and future prospects of cognitive engineering in human-computer interaction (HCI). One aim is to shed some light on design problems that may arise as we extend our scope of HCI design from the immediate interface between a single user and the system, to the indirect interface between the system and the task world, and to the organizational interface for cooperative work between people and systems. Also discussed are methodological problems associated with the attempt to build more or less general models of the user in HCI. Lastly, implications of future technological advancements in HCI for a cognitive science view of human communication are discussed in a somewhat futuristic manner.