Visualizing social networks, especially an overview emphasizing their structure, i.e., communities and their interconnections, is known to be a challenging problem. In this paper, we present a set of design rationales to build such overview visualizations of social networks and our solution called Jasper. We evaluate its performances against two of the most widespread visualization techniques (ma-trix and node-link diagram) in a human-computer controlled experiment based on community-related tasks. While none of the techniques emerge as the overwhelming winner, Jasper appears to be one of the best method for each task; a fact sustained by the marks given by the users. Overall, Jasper can be seen as an all-encompassing solution for quickly producing legible and compact overviews of large social networks on a single modern computer.