摘要:Complex, transnational issues like the tobacco epidemic are major challenges that defy analysis and management by conventional methods, as are other public health issues, such as those associated with global food distribution and climate change. We examined the evolution of indoor smoke-free regulations, a tobacco control policy innovation, and identified the key attributes of those jurisdictions that successfully pursued this innovation and those that to date have not. In doing so, we employed the actor-network theory, a comprehensive framework for the analysis of fundamental system change. Through our analysis, we identified approaches to help overcome some systemic barriers to the solution of the tobacco problem and comment on other complex transnational problems. Understanding complex policy issues requires a socioecological approach, 1 which we used to describe a tobacco-use management system 2 that not only determined tobacco use but also served other purposes (e.g., profit for companies involved). The model proved useful for defining key problems facing tobacco use management and for making recommendations for the future evolution of the management system. However, systems thinking says little about the processes involved in producing system change. We analyzed the evolution of smoke-free places, using the actor-network theory (ANT) to explicate the processes involved in producing significant system change. ANT 3 – 5 offers a comprehensive model of the innovation process, be it for producing new tobacco control policies or other sociotechnical changes. However, before we describe ANT in detail, it is important to understand the context within which ANT is applied. Systems change their structures and processes (how they do what they do) within limits defined by their response repertoire. This repertoire becomes more limited as the system optimizes its capacity around responses that have previously been successful. 6 (p205–207) Consequently, the potential for structural or procedural change becomes constrained over time. In addition, systems are typically unable to change their functions (what they do) from within, especially when their activities take place within a context characterized by polarization and stalemate, 7 , 8 such as the tobacco control domain. 2 , 9 Fundamental, innovative change requires action outside existing system structures. External actors include internal people (or groups) acting outside their system roles and external people or groups who, while varying in the intensity of their relationship with the system, believe it is not delivering. Management increasingly recognizes advantages in incorporating the views and actions of external actors when analyzing problems or redesigning systems. 10 – 13 These actors are not structured into any system, but act as a functionally coordinated network with respect to a potential change. 14 – 16 Policy innovation involves at least 4 stages to create a reformed system: establishing the issue or problem ( autonomization in ANT terms), developing possible solutions, contesting possible solutions, and institutionalizing, implementing, and refining the preferred solution. Most extant theories of change or innovation focus solely on 1, or at most 2, of these stages. For example, systems thinking 17 – 20 focuses on problem identification (but not how it becomes autonomized), and on potential solutions, whereas (social) diffusion theories, 21 , 22 knowledge management theories, 23 , 24 and approaches derived from complexity theory 25 focus on the last 2 stages. 26 ANT is a broader umbrella theory that can explicate all 4 stages; it encompasses Westfall et al.'s 27 extended US National Institutes of Health (NIH) categories—translation 1 (basic science to trials), 2 (trials to practice) and 3 (practice-based refinement)—and uses concepts and strategies associated with the more focused theories in doing so. At each stage of an innovation process, the resolution is not completely determined by evidence and internal coherence, but is codetermined by the capacity of the parties to keep their arguments alive, which involves persistence and power. This capacity is influenced by the different framings 28 , 29 competing networks use (e.g., protecting health vs protecting choice), leading to the development of different narratives or stories. The evidence, the theories that make sense of the evidence, plus the implications and possible outcomes, are woven together by the protagonists into a discourse intended to persuade others to their point of view. As explained in the following sections, ANT provides a theoretical framework for understanding, and a guide to the prerequisites for, successful change in complex systems; in this case, through policy innovation.