Very little research has focussed on interaction between communities of systems, systems for which participation may be a better concept to understand human/machine interaction [1]. However, to be able to conceptualize the notion of ‘communities of systems and people’ and participation, an understanding of the implications is required. Participation in a community requires social interaction. Participants need to be capable of recognizing spatio-temporal trajectories of other members of the community-of-practice they share [2]. Participation in a community also presupposes the potential to negotiate and to accept responsibility and liability. Such potential is time sensitive both for users and owners of agents. Fundamental values for MAS design like autonomy, transparency, identifiability and traceability are time sensitive as well [3].
Witnessed presence is the topic of exploratory research in 2008/2009 on which this paper is based: exploratory research in which 24 experts (professionals in ICT and design, artists and system engineers) worldwide have been interviewed. These experts reflect on the notion of witnessed presence and trust in relation to daily professional practice. All of these experts share a concern to create trustworthy structures of communication coming from computer science, design, business, art, architecture, theatre, journalism and social science. The YUTPA framework [4] in which the relation between presence and trust is defined by four dimensions: time, place, action and relation provided the frame of reference.
In section 2 the theoretical foundation of these concepts is discussed and in section 3 results of this exploratory study are presented and the conclusion will be drawn that time design is crucial for trustworthy collaboration. Given these results, section 4 discusses implications of time design for agents and autonomous distributed systems.
1. Introduction Time Design for Building Trust
Multi-agent systems are most often designed for a purpose, to perform one or more tasks, to achieve one or more goals for human users. Current research on multi-agent system design focuses primarily on MAS functionality, structure and (emergent) behaviour. Very little research focuses on interaction between human beings and multi-agent systems. The current design paradigms within collaborative systems design presuppose delegation of tasks or mandation of systems to perform tasks.
