Abstract
Much of the current research in HCI is carried out using experimental environments based on word processors, database search, or other conventional office automation. While this approach meets many needs it lacks the power required for investigating many unconventional situations. Complex multi-agent real-time systems are not typically found in offices and cannot easily be investigated in typical word processing or office automation contexts. The paper refers to four environments where multi-agency exists in a real-time environment: flight systems, plant control, telephone networks, and complex office systems. Consideration is given to the requirements of an alternative experimental environment which could allow HCI research to explore a wider range of issues.