Tools, combined with methodologies, enable researchers nowadays to work on material that was formerly believed to be inaccessible to sociological and political research. This is the case here of a photograph of the content of many sites in France concerning the European Constitutional Treaty, refused in the 31 May 2005 referendum. The author analyses the corpus of data and the actors by means of a tool, the crawl, and describes the necessities of its configuration. He then examines the way in which the communities of 'yeses'? and 'no's'? were constituted and fuelled, and the interactions between the two.