Documentation, interaction, and conversation

Article Documentation from the perspective of an AI-oriented interface designer. W hen I learned that I had been awarded the Joseph T. Rigo prize for achievement in system documentation, I felt greatly honored, and also a bit puzzled. In my research and writing, I have never seriously addressed the problems of documentation as a professional ac. tivity. My image of what a professional organization on documentation must do was dominated by the common stereotype of documentation as a rather static and formalized effort. The very word "documentation" carries with it an aura of rigidity. An "undocumented" person needs to go through bureaucratic rigors before getting "documented" and thereby become a real person in the eyes of the state. A system or product can be criticized as not "completely documented" if some detail is missing in the instructions. And certainly within the culture of the software industry, documentation is that burdensome chore that managers are always trying to force onto recalcitrant and otherwise productive programmers. But when I learned more about the award and the organization and its journal, I realized that the field as represented by SIGDOC has gone well beyond that outsider's image. It provides a forum in which to reflect on the larger issues of communication that underlie the practices of documentation. In looking further, I became curious about just what mix of perspectives one finds in the documentation field today. It is certain that there isn't unanimous consensus about how to approach the problem of documentation, and it seemed likely to me that there would be some of the same tensions that pervade other fields that deal with language and communication. Fernando Flores and I drew a contrast between two traditions in the understanding of language: "rationalistic" and "hermeneutic". The rationalistic tradition emphasizes formal representation and its correspondence to truth. The hermeneutic tradition emphasizes interpretation and its dependence on background. In that book, we presented the implications of this difference for i