摘要：A long-standing debate in the IS literature concerns the relationship between technology and organization. Does technology cause effects in organizations, or is it humans that determine how technology is used? Many socio-theoretic accounts of a middle way between the extremes of technological and social determinism have been suggested: in recent years the more convincing explanations have been based on Giddens’ structuration theory and, more recently, on actor network theory. The two theories, however, may be seen to adopt rather different, and potentially incompatible, views of agency. Thus, structuration theory sees agency as a uniquely human property, whereas the principle of general symmetry in actor network theory implies that machines may also be actors. This rather fundamental disagreement may be characterized as the problem of agency. At the empirical level the problem of agency can be studied through ERP systems. These systems, though built and implemented by people, are thought to be wide-ranging in their effects on organizations, and offer good opportunities for the study of the interplay of human and machine agency. However these empirical stories also reflect the theoretical confusion. This paper argues that neither structuration theory nor actor network theory offers a particularly convincing account of the interaction of humans and machines, and that their different accounts of agency make them hard to integrate in any meaningful way. Comparing the two theories and their use in IS raises many important issues, questions and problems, which need to be solved if the IS discipline is to develop a consistent socio-theoretical vocabulary.
