This paper argues that in order to allow for the comparison and sound assessment of possibly controversial or uncertain information on the web, the semantic web effort requires increased capabilities for the social reasoning about web ontologies and other knowledge contributions. As an approach to this issue, we propose formal means for the logical representation of heterogeneous, possibly inconsistent opinions, and various attitudes towards statements on the web using metamodeling techniques, namely modal description logic. Doing so, we integrate concepts from speech act theory and distributed artificial intelligence with approaches to web semantics, aiming for a social semantics of web content.