ABSTRACT
The last four chapters have generated a range of empirical observations. We have explored, for example, some normative dimensions of the organisation of calls to public information services; differences in the ways in which exchanges are closed in the human corpus and the WOZ data; and finally we have begun to examine some of the ways in which subjects in the WOZ study dealt with what they interpreted to be problems in the exchange with the system. In this final chapter we want to draw upon some of these results to make some observations about the simulation studies, conversation analysis and human-computer interaction. We will organise our discussion in terms of the three key concerns of this study: wizards, and simulation studies more generally; machines, and the use of CA in the design of speaking computers; and humans, or the importance for HCI of taking account of the kinds of everyday communicative practices people will rely on in their dealings with apparently intelligent artefacts.
