Video communication systems are most often used for short, synchronous and highly-engaged face-to-face interactions. Previous work on mediaspaces has demonstrated the potential value of long-term video links for casual awareness and informal interaction. Yet, few video systems manage to effectively support both general awareness and face-to-face interactions. As a notable exception, MirrorSpace provides users with simple ways of choosing the level of engagement that best suits their needs from a continuous set of possibilities. I believe this notion of variable degree of engagement is a particularly important one for mediated communication and that we should develop what I call multiscale communication systems [1, 2].
As part of a research project funded by France Télécom R&D between 2005 and 2008, Sofiane Gueddana and myself have designed a series of domestic communication systems based on this multiscale approach . Pêle-Mêle  is the first prototype we developed. It is a multiparty video communication system that combines simple computer vision techniques, spatial and temporal filtering of the video streams and an original layout to support a variable degree of engagement. Using both asynchronous and synchronous mechanisms, Pêle-Mêle supports different forms of communication ranging from casual awareness to focused face-to-face interactions. This video (22 Mb) shows an early version of the system. The use of a later version of the system over a long period of time got us interested in the effect of communication pace on attention allocation, which we also studied in a more controlled context [5, 6].