The last ten years have seen the growth of linkages between many of the world s cap-and-trade systems for greenhouse gases (GHGs), both directly between systems, and indirectly via connections to credit systems such as the Clean Development Mechanism. If nations have tried to act in their own self-interest, this proliferation of linkages implies that for many nations, the expected benefits of linkage outweighed expected costs. In this paper, we draw on the past decade of experience with carbon markets to test a series of hypotheses about why systems have demonstrated this revealed preference for linking. Linkage is a multi-faceted policy decision that can be used by political jurisdictions to achieve a variety of objectives, and we find evidence that many economic, political, and strategic factors - ranging from geographic proximity to integrity of emissions reductions - influence the decision to link. We also identify some potentially important effects of linkage, such as loss of control over domestic carbon policies, which do not appear to have deterred real-world decisions to link. These findings have implications for the future role that decentralized linkages may play in international climate policy architecture. The Kyoto Protocol has entered what is probably its final commitment period, covering only a small fraction of global GHG emissions. Under the Durban Platform for Enhanced Action, negotiators may now gravitate toward a hybrid system, combining top-down elements for establishing targets with bottom-up elements of pledge-and-review tied to national policies and actions. The incentives for linking these national policies are likely to continue to produce direct connections among regional, national, and sub-national cap-and-trade systems. The growing network of decentralized, direct linkages among these systems may turn out to be a key part of a future hybrid climate policy architecture.