Paper long abstract:

I created a monster. A tool that people use to share the most unfathomable and arguably useless visualizations of their research. Nobody can do anything to stop it, and I think this is for the better. Such a monster is not viable. Yet each time it collapses under the weight of its own problems, it is reborn again following the needs of its users. This paper will explain this paradox by exploring the intersection of design, code and social science research.

Gephi is a rather successful open source network visualization and analysis software widely used in social network analysis (and in other sciences like biology). It receives approximately 20,000 downloads per month. I created its prototype back in 2006, though now a community of developers lead by Mathieu Bastian develops it. Over the years I specified or contributed to specifying most of the software's features, and I continue to give dedicated teachings (currently in the Sciences Po médialab) to students and researchers interested in using it. It is particularly through these trainings that I have learned many things about methods in the social sciences: how a simple tool can enable, bend or block a research enquiry, and how engineers can be inventive to help researchers obtain satisfactory findings. In this paper I will explain how we discovered that some unexpected features were decisively useful to social sciences. Technique and method are two sides of the same coin: building a tool is implementing a methodology.