How do various actors use digital media to influence the flows of political communication and what are the social consequences of their practices? Treating media as a space of actions and building on network theories, this dissertation argues that public attention and opinion on social media can be understood and analyzed through networks of social interactions among social actors. Ultimately, this approach traces the multifaceted flows of attention and opinion to the sources of their origins in an increasingly complex media system and holds the potential of revealing patterns of interactions between networks of social actors, and between social networks and institutional networks like news media. Results show that embedded in networks of online affinity relations (i.e., following relationships), social actors within a Twitter "flock" exhibit homogenous attention and opinion patterns, and that such flocks interact with each other and can influence news media coverage. Moderate and center-left news media network still possesses significant power in driving the attention and setting the agenda for partisan news media networks and Twitter flocks, though there are some bottom-up flows of communication from Twitter flocks to news media networks. Also, the interaction between partisan news media networks and partisan/activist Twitter flocks gives rise to partisan media ecosystems, with the conservative and progressive media ecosystems reacting to each other differently. Furthermore, activism discourses on Twitter can originate from vastly different Twitter flocks situated in networks of communications and exhibit varying attention dynamics. These results speak to the continuing splintering of the public into passionate and engaged networked publics. Such networks aggregate attention and synthesize opinions, exerting direct influence on powerful legacy news media via networked visibility and power. However, homogenous networks on social media and the oppositional reactiveness between the partisan ecosystems reflect a deepening partisan divide in the digital media system. If the fundamental battle about the definition of the norms of society, and the application of these norms in everyday life, revolves around the shaping of the human mind, communication is central to this battle. … [The] process of communication operates according to the structure, culture, organization, and technology of communication in a given society. The communication process decisively mediates the way in which power relationships are constructed and challenged in every domain of social practice.-Manuel Castells, Communication Power Shaping the human mind and underlying power and counterpower, communication is central to any society (Castells, 2013). For social actors who strive to maintain or challenge power, their success hinges on the ability to communicate ideas to the public effectively, including attracting public attention and molding public opinion and ultimately driving communication flows. As those actors influence the process of communication in society, they can challenge those in power and thereby alter existing power relationships. At the same time, actors and processes of communication are shaped by existing power structures, particularly the media system that enables and constrains actors and their communications. While existing studies have examined the media system in the early decades of the 21st century by focusing on shifts in its structures, norms, cultures, and audiences (e.g., Benkler, 2006; Chadwick, 2017; Prior, 2007; Stroud, 2010), there is yet a systematic framework for understanding and tracing communication flows to reveal the dynamics of the media system and power relations. Treating media as a space of actions (Couldry, 2012), I examine how social actors use social media to signal attention and express opinion, and how their practices are related to communication flows in the larger media system. Building on network theories (Castells, 2013; McPherson et al., 2001; Latour, 2012), I propose a network approach that positions social actors on social media in the networked space of social interactions and places social interactions on social media in the networked space of social actors. I apply computational methods, including network sampling and analysis (Chen, Zhang, & Rohe, 2019) to detecting networks of social actors and interactions so as to track public attention and opinion expression; employ time series modeling (Wells et al., 2019) to formally analyze their interaction with news media; and use natural language processing to map out the patterns of public opinion and media content. Results provide empirical support for the network approach and reveal complex patterns