Abstract
Hypertext - advanced software for organizing information according to webs of conceptual, rather than symbolic, links - has recently provoked humanists to reconsider post-structuralist semiotic concepts. Debates about the design and uses of hypertext, among both software developers and humanists, reflect a conflict between two problematic views of text: as a medium for social interaction, and as a replication of a cognitive structure. Post-structuralist critical theory (PSCT), in challenging concepts of authorship and univocal meaning, argued that semiotic products were more closely connected to each other than to `reality' or to their `original' producers. PSCT's notion of `intertextuality' captured this hyperactive, social aspect of language products. Theories of social construction of scientific knowledge (SCSK), I argue, have crucially relied upon similar, essentially semiotic concepts such as inscription devices, discourse repertoires and the textualization of heterogeneous resources. As SCSK's practitioners have articulated their programme, they have covertly imported cognitive abilities into ostensibly social processes, creating a kind of theoretical hypertension which surfaces in the similar debates over hypertext. Questions about the status of artificial intelligence, which concerns the capacity of a purely symbolic/syntactic structure - a hyper text - to perform as a social actor, sharply expose the tension between cognitive and social that underlies many of SCSK's key concepts.