Beyond institutions and the system: Network approach to deliberative democracy

Abstract The goal of this article is to advance a novel framework for studying deliberative democracy in real-world contexts. Currently, the theory is gradually moving from an institutional approach rooted in micro-deliberation toward a systemic one focused on macro-democracy. I argue that though this tendency has some unquestionable advantages, both approaches fail to tackle what I call “the problem of embeddedness”. On the one hand, different institutional models of deliberative democracy tend to isolate deliberation from the surrounding context. Therefore, institutions are under-embedded, detached from ongoing processes, which prioritizes action over impact. On the other hand, different systemic accounts of deliberative democracy tend to subject deliberation to the surrounding context. Therefore, institutions are over-embedded, engulfed in the ongoing processes, which privileges the outcome over input. Drawing on the social network theory, I develop a conceptual framework which integrates both approaches into a comprehensive perspective aimed at compensating for the above shortcomings. Focusing on relations and interactions between various nodes, this network approach tries to bridge public deliberation and democratic system. The latter is understood as a relational-pluralist structure in which horizontal and vertical ties are established not only between institutions but also among individual actors and entire networks.

Abstract: The goal of this article is to advance a novel framework for studying deliberative democracy in real-world contexts. Currently, the theory is gradually moving from an institutional approach rooted in micro-deliberation toward a systemic one focused on macro-democracy. I argue that though this tendency has some unquestionable advantages, both approaches fail to tackle what I call "the problem of embeddedness". On the one hand, different institutional models of deliberative democracy tend to isolate deliberation from the surrounding context. Therefore, institutions are under-embedded, detached from ongoing processes, which prioritizes action over impact. On the other hand, different systemic accounts of deliberative democracy tend to subject deliberation to the surrounding context. Therefore, institutions are over-embedded, engulfed in the ongoing processes, which privileges the outcome over input. Drawing on the social network theory, I develop a conceptual framework which integrates both approaches into a comprehensive perspective aimed at compensating for the above shortcomings. Focusing on relations and interactions between various nodes, this network approach tries to bridge public deliberation and democratic system. The latter is understood as a relational-pluralist structure in which horizontal and vertical ties ABOUT THE AUTHOR Marcin Zgiep is a graduate of the Institute of Political Science, University of Warsaw. He was a PhD student in the Institute for Social Studies, University of Warsaw. Mr Zgiep collaborated with the Centre for Deliberation affiliated at the Institute of Sociology, University of Warsaw. His research interests cover political philosophy and theory, democratic theory and practice as well as European integration. Mr Zgiep wrote numerous journal articles and book chapters, and participated in many conferences worldwide. Currently he is working on the project "Between the Theory and Practice of Public Dialogue: Models of Deliberative Democracy and Contemporary Democratic Institutions and Practices" supported by a grant for junior researchers from the Polish National Science Center. This article is intended to provide the analytical-evaluative framework for subsequent empirical research to be carried out on a selection of American and Polish cases.

PUBLIC INTEREST STATEMENT
The crisis of democracy is one of the key features of contemporary political landscape. Many competing remedies have already been proposed but none of them gained universal acceptance. One of them is deliberative democracy, a form of government in which people make decisions by means of public deliberation. The latter can be understood as rational communication aimed at solving problems. Currently, deliberative democracy has become a thriving area of research. However, existing theories and evidence from real-world cases seem to be inconclusive. We know little about how deliberation actually works in a democracy. Different approaches have been put forth, concerned either with individual democratic institutions (parliaments) or with largescale democratic systems (states). In this article, I propose a third way which tries to reconcile the above possibilities. Building on network theory, my goal is to focus on how ties between actors, institutions and networks shape public deliberation within a given system.
In this article, my goal is to advance a novel framework for studying deliberative democracy. Drawing on recent developments within this paradigm, I will integrate the existing approaches into a comprehensive perspective. Instead of looking at deliberation from the point of view of institutions or the system, I will rather emphasize the connections that exist or emerge within and between institutions located in the system. This "third way" to a large extent overlooked by current scholarship constitutes an alternative solution to the perplexing dilemma of choosing between micro and macro perspectives. The framework proposed here builds upon the well-known network model used in many scientific disciplines, e.g. sociology (White, 2008). Its goal is to supplement a categorical analysis with a relational one, that is to shift from studying elements to analyzing connections. If we apply this model to the realm of deliberative democracy, we can easily see that it has an advantage over actor-centered and institution-oriented micro-level approaches. The latter overestimate the value of individual actions and isolate deliberation from the wider context. On the other hand, system-centered and result-oriented macro-level approaches tend to ignore the role of the constitutive parts and treat institutions merely as a background. In both cases, we encounter the problem of embeddedness. Whereas in theories of micro-deliberation discourse is "under-embedded", in theories of macro-deliberation it is "over-embedded". This problem has farreaching implications. If we look at it from a standpoint of norms and facts crucial to the deliberative democratic paradigm (Habermas, 1996a), we can see that both approaches do not take these aspects seriously. Not only do theories of micro-deliberation tend to idealize the conditions of discourse, they also put too much emphasis on the properties of actors and institutions in determining those conditions. On the contrary, theories of macro-deliberation undervalue the latter by pointing to the significance of broader ramifications. Moreover, they insufficiently focus on how actors and institutions affect the large-scale processes. In contrast to both approaches, the network alternative stresses the importance of existing and emerging linkages that enable dynamic interdependence between actors and institutions in the system thus bringing a more realistic perspective into picture. In this light it tries to address the problem of institutionalization with which deliberative democracy has been struggling since its inception.
In general, the network approach treats various institutions as interconnected elements within the system. Institutions constitute its main building blocks, but they are not the only venues of deliberation. On the one hand, institutions comprise actors as individuals who deliberate together. On the other hand, these deliberations are part of the larger discourse. The network approach makes a crucial analytical distinction between three levels: intra-institutional (actor-centered), inter-institutional (institution-centered) and trans-institutional (network-centered). All parts are regarded simultaneously as "nodal points" and "networked environments" within the system understood as a "network of networks", that is an open-ended structure consisting of various relations and interactions. This perspective balances between micro-and macro-level approaches opening up a vast area of normative and empirical opportunities. It aims for the "hidden" meso level at which deliberation is at the same time independent from and dependent on the individual parts and the overarching whole.
The remainder of this article is structured as follows. The first section starts with an argument for the necessity of an institutional approach to deliberative democracy. Drawing upon the normative theories and empirical research I will show what are the strong and weak points of studying institutions. In the second section, I will introduce the recently proposed systemic approach to deliberative democracy which in one way deepens and in another simplifies the understanding of the institutional perspective. The third and last section is dedicated to the network approach to deliberative democracy. I will argue that it has the advantage over both approaches in that it integrates them into a single comprehensive framework. In the conclusion, I will indicate the possible avenues in which the theory and practice of deliberation can be developed further.

Institutional approach to deliberative democracy
The most basic understanding of deliberative democracy is associated with institutions. They can either be essentially deliberative or have certain deliberative features. While the first group of institutions encompasses mainly participatory institutions, e.g. deliberative polls, consensus conferences and citizen juries, the second group comprises representative ones, e.g. parliaments, constitutional courts and the media. However, there are examples which cut across this dichotomy. Participatory budgeting has some important deliberative features but is not a standard deliberative institution. In social movements, deliberation is only one among many forms of participation. Despite these differences institutions are necessary to generate communication among people in a given setting. They not only enable but even force participants to give an account of one's actions. In a stronger sense actors have to provide reasons for their past, present and future behavior. Therefore, institutions also structure communication in a way that actions have to be communicated/justified to others. In other words, they establish rules for collaboration that brings about better outcomes but requires collective efforts.
These institutional rules or norms can be either activated or challenged by participants. This dual perspective means that institutional design matters as much as a political practice. Designers should think about possible ex-post scenarios. If norms are not followed, the entire process can be deeply flawed. But if they are all wrong, rejecting them is a necessary thing to do. Though deliberation is affected by design and practice, it also impacts the process. In fact, there is a loop between institutions and deliberation. The latter can either strengthen the former or weaken it by means of critical discussion. Well-designed institutions enable actors to deliberate on how good the process is and how to further improve it.
1.1. Institutional models: deliberative liberalism vs. deliberative republicanism Thus, it is crucial to underline that it does not matter so much if we look at genuine deliberative institutions or institutions with deliberative features as long as there is much deliberation going on. We should rather turn to the particular design and practice without making too far-reaching assumptions. When it comes to political discourse, Steiner and his collaborators made a speech-act analysis of parliamentary debates in four countries trying to figure out which properties of political systems and issues contribute to high-quality discussions (Steiner, Bächtiger, Spörndli, Steenbergen, 2004). Looking beyond political discourse, Maia focused on mediated deliberation showing in what ways media provide space and constrain public debate among participants (Maia, 2009). Moving towards participatory discourse, Fishkin and his group conducted extensive studies on the deliberative polls (Fishkin, 2009), which point to variations in deliberation across many cases. Finally, as for radical discourse, della Porta did research on how social movements use different deliberative mechanisms in their internal processes (Della Porta, 2013). These few examples prove that any predefined classification is superfluous as it does not capture the myriad ways in which deliberative democracy actually works. Taking all this into consideration, I argue that deliberation is by nature rooted in institutions because it is the latter that makes communication possible and beneficial, regardless of any specific patterns. Not all discourse is organized in the same way but to some minimal degree it is present in all known settings. Non-political facilities like schools and workplaces have their own internal structures of decision-making just as political bodies (Fung, 2005). Non-public spheres like families engage its members in constant communication (Gerson, 2004). Even internal reflection would not be possible without external discourse (Goodin, 2003).
The question is then not whether and which institutions are deliberative, but how (well) deliberation is institutionalized. In the literature, one can find several classifications of deliberative institutionalization but the problem is that they draw upon the most cited models/studies and rarely combine normative theory with empirical research. My goal was to carry out an extensive literature review covering all published works since 1980s until now. Based on this broad corpus of texts I was able to distinguish four institutional models of deliberative democracy which refer to four contexts of contemporary politics. These models fit into two grand traditions of political thinking, namely liberalism and republicanism which matches the typology put forth by Habermas (Habermas, 1996a). Unlike him however, I argue that deliberative democracy does not constitute a separate "procedural model" but it rather tries to enrich and transform both traditions. That is why we can speak of liberal and republican models of deliberative democracy. The liberal model focuses on "discursive representation" in which deliberation is oriented towards public justification of political decisions. This "deliberative liberalism" is interested in protecting the freedom of citizens from any arbitrary interference. Deliberation is thus institutionalized to limit democratic power. Liberal model bifurcates according to two aspects: "deliberative ends" and "deliberative agents". When it comes to goals, we distinguish between will-formation and opinionformation. While the former is tied to decision-making, the latter is related to preference-shaping. As for the agents, we differentiate between political elite and symbolic elite. The former is responsible with making decisions on behalf of others, the latter-with shaping others' preferences. Therefore, democratic process comprises a core political process taking place inside representative institutions and concomitant public debate organized by media outlets. This internal differentiation leads to two variants of the liberal model: "political deliberation model" (PolDM) and "mediated deliberation model" (MedDM), which are oriented towards the performance of state and media as the loci of political power and public information, respectively. From an empirical perspective PolDM describes how political system works in order to make a binding decision. From a normative point of view it prescribes how it should works in order to make a legitimate decision. Therefore, this model involves institutions that take part in will-formation in which political elites make proposals in a given policy-area backed up by arguments. Power is thus tied to generating justifications in settings such as elections, political parties, parliaments, constitutional courts, government agencies, etc. Empirically, MedDM describes how the public sphere works in order to form a collective opinion. Normatively, it prescribes how it should works in order to form a reasoned opinion. Therefore, this model includes institutions that take part in opinion-formation in which symbolic elites take political standpoints on a given issue backed up by arguments. Information is thus related to generating justifications in settings such as newspapers, radio, television, the Internet, etc.
Republican model focuses on "discursive participation" in which deliberation is oriented towards public input into political decisions. This "deliberative republicanism" is interested in providing citizens with opportunities to challenge any arbitrary domination. Deliberation is thus institutionalized to control democratic power. Parallel to the liberal model, republican one bifurcates as well. When it comes to the deliberative ends, we distinguish between influencing will-and opinion-formation, and changing them, respectively. The latter is connected with participating in decision-making and preference-shaping, the former is related to contesting both. As for the deliberative agents, we differentiate between (self)selected citizens and (self)organized citizens, respectively. Whereas the former are included in making decisions and shaping preferences, the latter are involved in contesting both. Therefore, democratic engagement comprises conventional participatory process taking place inside citizen institutions and alternative political protest organized by social movements. This internal differentiation brings about two variants of the republican model: "participatory deliberation model" (ParDM) and "contestatory deliberation model" (ConDM), which are differently linked to the performance of society as the loci of social engagement. From an empirical perspective ParDM describes how civil society works in order to exert political influence. From a normative point of view, it prescribes how it should works in order to exert legitimate influence. Therefore, this model encompasses institutions that contribute to will-and opinion-formation in which (self)selected citizens discuss issues and recommend solutions to key social problems. Engagement is thus associated with processes of generating inputs in settings such as public committees, NGOs, public consultations, deliberative polls, participatory budgets, initiatives, referendums, etc. Empirically, ConDM describes how counter-public sphere works in order to make a social change. Normatively, it prescribes how it should work in order to make a reasoned change. Therefore, this model is comprised of institutions that challenge will-and opinion-formation in which (self)organized citizens oppose existing arrangements and struggle for bringing about alternative ones. Engagement is thus linked to generating inputs in settings such as social movements organizations, street demonstrations, strikes, sit-ins, occupations, etc.

Institutional approach and the problem of (under)embeddedness
Much research has been done so far on particular institutions with large body of work dedicated to the so-called "mini-publics". Recently the latter has become the subject of empirical scrutiny (Grönlund, Bächtiger, & Setälä, 2014;Warren & Pearse, 2008). Though there are examples of normative analysis (Lafont, 2015) their prime concern was to elucidate the deliberative dimension of political institutions (Cohen, 1997;Lascher, 1996;Mendes, 2013;Rummens, 2012). Still we have some empirical studies targeting more elite-centered settings (Davis, 2009;Ferejohn & Pasquino, 2002;Steiner et al., 2004). Although there is not much systematic, historical and comparative research available, more and more case studies are conducted on the performance of deliberative institutions worldwide. Major problem with existing findings is that due to differences in research design they cannot be easily correlated and compared. The potential of generalizing the results is to a large extent limited because researchers focus on various (aspects of) deliberative institutions. Their results do not bring us closer to the truth about how deliberative democracy actually works. One of the illuminating examples comes from the area of democratic innovations in which scholars have elaborated a framework for analyzing and evaluating the performance of participatory institutions (Geissel & Gherghina, 2016). This framework encompasses three dimensions of legitimacy: input pertaining to participation by and representation of the people, throughput referring to procedures and processes through which inputs (preferences) are converted into outputs (decisions) and finally output concerned with arriving at problem-solving outcomes. Each of above elements is scrutinized according to two equally important aspects: ex ante institutional design and ex post political practice (Geissel & Gherghina, 2016). On the one hand, we ought to know whether the institution was properly designed in order to enable the deliberative decision-making in the first place. It is important to ask if the designers had established norms which provided space for actors to participate, deliberate and decide on the issues. On the other hand, we should know whether actors practiced the deliberative decision-making according to these principles. It is vital to ask whether the norms were activated while participating, deliberating and deciding on the issues. To answer both questions, we apply the selected criteria, e.g. "Inclusive participation"; "Quality of deliberation"; "Impact on final results", and indicators, e.g. who can take part/who took part, rules for deliberation vs. practice (DQI), institutionalized connection to constitution-making body vs. real impact on constitutions etc.
The problem with this approach is that though being seemingly comprehensive it actually models institutionalization according to the ideal of participatory deliberation, adjusting realworld institutions to the predefined scheme. Furthermore, it only vaguely tries to map the possible ties between a given institution and its surroundings. Though it points to the latter, it is treated merely as a background for what takes place among the deliberating actors. External context is absent from scrutiny thus rendering the framework incomplete, partial and ultimately one-dimensional. None of the existing accounts provides an accurate picture of deliberative democracy focusing rather on democratic deliberation as its constitutive element. 1 That said, it is crucial to say that the problem of abundance does not pose a real dilemma. Far more critical is the already signaled "problem of (under)embeddedness", that is the isolation of institutions from their environments. Although institutional approach generates a large amount of data on their deliberative performance, we still know little about how these sometimes wellresearched institutions are embedded in particular contexts. This problem emerges only if we are interested in the democracy at the large scale but even at the local level political outcomes result from the work of several institutions. Studies of parliamentary discourse like those conducted by Steiner's group focus entirely on the internal deliberation between MPs without taking into account how parliaments interact with other institutions such as constitutional courts, the media, NGOs, and SMOs. The same applies to participatory institutions such as DPs which are even more separated from any external setting or process. This creates a dilemma between depicting an easily discernible but tiny fragment of political reality and widening but blurring that picture. Choosing the first strategy means setting fixed boundaries between discourse and its context which is neither desirable nor possible. Normatively, it would mean that deliberation should affect only participants engaged in direct communication. In consequence, it would largely reduce the essential goal of deliberative democracy to reinterpret and transform current democratic arrangements. Empirically, it would mean that democracy is based on a set of micro-deliberations taking place independently from each other. It is far from the truth even if certain institutions work on the margins of the system, e.g. in enclaves (Sunstein, 2002). For a genuine deliberative democracy we need a perspective in which we are able to assess how deliberation impacts and is impacted by all relevant institutions and analyze how the latter are related to and interact with one another in creating space for decision-making. Institutional approach points rightly to the foundations of discourse but eventually fails to deliver such a perspective. And without it we are not able to make any definite and substantial analysis and evaluation regarding deliberative democracy. Although the problem of abundance seems likely to be mitigated it cannot be said the same about the problem of embeddedness.

A systemic approach to deliberative democracy
Existing institutional models of deliberative democracy tend to track the focal points of structure and dynamics of deliberation. Though they capture the factors determining the deliberative performance of institutions they fall short of a more complex picture of deliberative democracy. What the institutional approach leaves out is the systemic account of the process of which an institution is a part. By a systemic account I mean here the relations and interactions in which that institution is embedded. While relations mean existing ties between actors and institutions defining their status, interactions can be defined as actions undertaken by two and more entities aimed at preserving or changing the status quo. They can be horizontal and vertical, depending on whether they exist/emerge between evenly placed actors/institutions or between hierarchically situated ones. In this light deliberative democracy is construed as an overarching system that cannot be reduced either to single, large forum (Elster, 1997) or to a set of small forums (Gastil, 2014;Gastil, Deess, Weiser, & Simmons, 2010). Conversely, it consists of all arenas which contribute to the deliberative quality/potential of the entire structure. According to the recently influential theory of deliberative systems, it is precisely the connection between the forums, or deliberative sites, that matters. "To understand the larger goal of deliberation, we suggest that it is necessary to go beyond the study of individual institutions and processes to examine their interaction in the system as whole" (Mansbridge et al., 2012).
While institutional approach begins and ends with micro-deliberation, systemic one goes beyond it and tries to capture the macro-deliberation which can be understood as a communicative superstructure emerging from various sources. The latter matter as much as they affect the former in one way or another. It means eventually that the quality/potential of a particular instance of institutional deliberation is dependent on the quality/potential of systemic deliberation. Deliberative performance of a given institution hinges upon the deliberative performance of the system. This changes the way in which analysis and evaluation is carried out. First, analysis is focused on how the relations and interactions between institutions affect the system. What is crucial here is how micro-deliberation feeds into macro-deliberation. Studying these emergent patterns can bring us closer to answering the question what kind of institutional arrangements and collective practices are beneficial/detrimental for/to deliberative democracy. Second, evaluation targets mass communication as the effect of direct communication between actors. The key aspect is how well the system performs based on the overall performance of all institutions. Evaluating the final outcome gives us a better position to judge the real quality of processes underlying the emergence of deliberative democracy. In the end, institutions are important insofar as they provide inputs into the system but it is the latter that generates outputs in form of democratically generated decisions/opinions.
From the above picture, we could draw an inference that micro-deliberation is irrelevant in systemic accounts as is macro-deliberation in institutional ones. Both statements are not entirely true. As argued in the previous section, institutional deliberation points to the external context but does not develop any comprehensive perspective that would capture the ensuing relations and interactions. Same happens in the case of systemic deliberation. Here too we can observe an attempt to connect both levels but the starting and main orientation point is a contrario, the external context. Systemic deliberation treats institutions as "black boxes" rather than discursive settings. The emphasis is thus put on the macro-level landscape with only a slight reference to the micro level. Whereas in the institutional approach, the system was absent or played a minor role, in the systemic approach the opposite is true, that is institutions constitute merely a background. They are treated as functions which induce deliberativeness of the system and thus the quality of democracy. Neither do they need to be essentially deliberative nor they have to initiate and sustain deliberation in the system. Institutions are expected to generate positive outcomes in terms of systemic reflexivity, reciprocity and responsiveness. They are functional if they manage to play at least some of these epistemic, ethical and democratic roles, and dysfunctional if they cannot or, in a worst-case scenario, exacerbate the situation.
Contrary to institutional approach, systemic one is still largely an under-studied area. Majority of works have been highly theoretical with an ambition however to set out an analytical framework for further studies to be carried out. There is not much empirical research done on how deliberative systems work in different structural and cultural settings (Felicetti, Niemeyer, & Curato, 2016;Jonsson, 2015;Maia, Laranjeira, & Mundim, 2017;Parkinson, 2006). Nor is there an attempt being made at a comparative analysis and assessment of those systems in a diachronic and synchronic manner. This creates a worrisome deficit of informative data that can show relevant patterns of systemic deliberation. This problem of scarcity will likely be solved with more and more research conducted both on the working of the individual (case studies) and multiple deliberative systems (comparative studies). Two main accounts proposed by Stevenson and Dryzek (Stevenson & Dryzek, 2014), and a group led by Mansbridge (2012) are a good starting point for future research but they need refinement as well. This is crucial in light of some unavoidable drawbacks pointed out by several scholars who argued for a revision of their key assumptions (Boswell, 2013;Dryzek, 2017;Ercan, Hendriks, & Boswell, 2017;Owen & Smith, 2015;Rollo, 2017;Tamura, 2014).

Systemic approach and the problem of (over)embeddedness
Major problem of the systemic approach seems to be the same problem of embeddedness with which the institutional approach has been struggling. This time however the subordination of institutions to the context poses a challenge, not the separation of the two. Whereas in the institutional approach deliberation was under-embedded in the system, in the systemic perspective the problem emerges on the opposite pole, that is deliberation is over-embedded. The system has both normative and empirical priority. On the one hand, indirect, dispersed communication in mass democracy is valued higher than face-to-face symbolic encounters. What it means practically is that assessment is aimed at the level of system even if the particular instances of microdeliberation are still relevant for other non-functional reasons. Participatory institutions are often seen as limited in their impact on larger processes. One of many examples are the two provinces of Canada, namely British Columbia and Ontario, in which citizen assemblies were used. Though they were praised for high-quality deliberation, they did not achieve the goal of changing the electoral system. If the systemic approach is the right one, then we will perceive those experiments in only one way, that is as a failure. 2 This conclusion however provides strong reasons to oppose deliberative democracy.
On the other hand, systemic deliberation is regarded as the end-result stemming from the division of labor between the institutions. An analysis is thus focused on the functioning of the system that is determined by how those institutions contribute to its deliberative quality. As in the case of systems analysis the emphasis is put on the transformation of inputs into outputs (Easton, 1965). Many cases of participatory budgeting, especially in Brazil, provide an argument against such reduction. They show that deliberation has also non-instrumental value in that citizens benefit from participating in PB bodies (assemblies, councils) regardless of their impact on the final budget (Avritzer, 2006). If we look at these examples primarily from a systemic perspective, we tend to prioritize the democratic outcomes over deliberative intakes.
Therefore, I argue that contrary to what proponents of the systemic approach claim, they are not in fact interested in the interdependence of institutions but in the system itself. It means that micro-deliberation which is the fundamental building-block of deliberative democracy is to a large extent neglected. Macro-deliberation overlooks institutions and consequently actors by eliminating boundaries between different discursive settings. This perspective of "messiness" does not take into account that discourse simultaneously occurs within and beyond individual sites, that is it is both generated by a given subject and diffused across a variety of locations. By means of oversimplification systemic account reduces deliberative democracy to the trans-institutional/personal realm which is normatively and empirically untenable. Whereas institutional approach makes micro-deliberation unnecessarily superior, systemic one tries to counterbalance it to the advantage of macro-deliberation. While in the first case, we can speak of deliberation without democracy, in the second one, democracy exists without deliberation. In the end, neither of them can provide a plausible account which is rooted both in norms and facts.

Network approach to deliberative democracy
Considering the shortcomings of both approaches to deliberative democracy, I propose here a perspective which emphasizes its intrinsically relational aspect. I contend that existing theories do not fully develop the latter, emphasizing instead either the macro-level performance or macrolevel effects. In so doing they have become an easy target of critique by those who oppose deliberation on grounds that its significance in a democracy is overrated. What is missing here, I believe, is how the deliberative micro-level is related to the democratic macro-level (Goodin, 2008). This issue has not been fully addressed by deliberative democrats. The gap between institutions and the system may however be overcome by integrating the institutional and systemic approaches in order to arrive at an alternative "network approach". The latter focuses on resolving the problem of embeddedness by investigating links between institutions within the system. Contrary to the institutional approach, this network perspective enables close examination of contexts in which actors and institutions deliberate together exchanging various resources, both material and symbolic. Deliberation is needed to sustain inflows and outflows of preferences between the sites. This is important for two equally valid reasons. First, deliberation allows to generate opinions and reasonsas inputs into the decision-making. Actors as participants express their values and interests which are then pooled together and discussed in order to form a collective viewpoint. Institutions are the tangible embodiments of this creative-critical process. Second, deliberation enables to weigh up prevailing opinions and reasons, which accounts for throughputs in the decision-making. It countervails the enforcement of certain values and interests at the expense of others by means of a discursive confrontation. Therefore, the free and equal flow of preferences in various directions guarantees that all relevant opinions and reasons are present within a given context. In consequence, all values and interests are either indirectly represented or directly included.
Contrary to systemic approach, the network perspective captures institutions which generate and shape ties among different groups of actors. Deliberation is needed to sustain adequate impacts and uptakes of preferences between the sites. This is important because it enables to influence opinions and reasons as outputs in the decision-making. Deliberation allows to uphold or change the existing values and interests via public discourse linking together all actors and institutions from across the system. In fact, the latter does not constitute a qualitatively separate entity to which everything is reduced. The system is rather a constellation of all elements, both discursive (Dryzek, 2001) and non-discursive which are scattered around on a certain area, be it local community, nation-state or international space. As such it does not have clearly defined borders-the latter are only analytically differentiated. The system is thus a dynamic structure which stability is challenged by ongoing changes between its parts. It is rather discontinuous as the reconfigurations among actors and institutions bring about new conditions affecting its elements. The system is both dialogic and dialectic because deliberation is concomitant with nondeliberation, discourses co-exist with actions, openness with closure. However, these dialogic and dialectic moments are just stages which endlessly intertwine in the process. Face-to-face communication ends with an action taken up by the participants, which triggers another direct encounter. Opinions and reasons are formed, mediated, re-formed, circulated, deformed etc. Any kind of action is not an isolated act but constitutes a response to a preceding act and a reference point for a succeeding act. They are all parts of an interaction among multiple subjects. Furthermore, all actions can either be the result of a discursive event or be discursively "encapsulated". In the first scenario actors deliberate on an issue and come to an agreement, compromise or any other arrangement followed by a decision-making mechanism, e.g. voting. In the second scenario actors simply communicate their actions. In the end the system is the product of such collective interactions based upon individual actions and consists of all transactions made throughout the decision-making. These outcomes are binding for all actors and institutions in two ways. They either constitute enforceable legal ("hard") acts or promissory speech ("soft") acts with varying levels of legitimacy. In this light the system is placed somewhere between overarching whole and loose set of elements. Institutions with their internal (actors) and external arrangements (contexts) constitute interdependent nodes linked to one another. Therefore, by focusing on relations and interactions the proposed framework maintains equal distance towards institutions and the system while keeping them as core elements of the entire structure.
Similar attempts to move beyond institutional and systemic approaches have already been made (Benhabib, 1996;Cinalli & O'Flynn, 2014;Habermas, 1996b;Hajer, and Wagenaar, 2003;Knops, 2016). However, neither of them tries to construct a genuine network-analytical perspective, focusing instead on a vague normative concept of subjectless communication (Habermas and Benhabib), a general concept of network society (Hajer and Wagenaar) or a social network analysis used to map deliberation in public sphere (Cinalli and O'Flynn). In his recent article on "Deliberative networks" Knops comes close to the idea of merging networks, deliberation and democracy together. Nevertheless, the latter seems to be left out of the equation. Although he shows how the systemic approach can be compensated with a more relational aspect ("vicarious deliberation"), different institutional arrangements are absent from his account. Despite explicit reference at the outset, Knops overlooks also different communication styles. The case study of implementing Poll tax in UK narrows down the scope of discourse to reason-based discussions in a small circle of government officials. It does not extend the range of deliberative network to other than Conservative political parties, media, social movements, etc. Nor does the analysis try to capture the more radical voices challenging the decision. Contrary to the goal of rendering deliberative democracy more normative, Knops fails at delivering a framework that would actually do justice to the potential of discourse. He also tends to focus more on the systemic perspective and its problems than on the institutional one. Therefore, his account does not fulfill criteria for a network approach to deliberative democracy.

Deliberative democracy and social network theory
The tendency to describe deliberative democracy in relational terms stems from the unresolved micromacro dilemma which was particularly crucial for modern sociology. Networks allowed researchers to think about social life as built upon interrelated subjects (Granovetter, 1973) or objects (Latour, 2005). Thus, it created a third possibility of conceptualization: beyond individualism and holism, rational utilitymaximizing individuals and structured norm-oriented society, respectively. Neither of those approaches could explain the socialization process as the former presented an under-socialized account and the latter put forth an over-socialized one (Granovetter, 1985). On network terms individuals are not as rational and utility-maximizing beings, and society is not as a structured and norm-oriented entity.
Socialization is an open-ended process of transforming persons and their environment. Relations and interactions-the building-blocks of networks-are located between levels of an individual and that of a society. As Mark Granovetter claimed, "the analysis of processes in interpersonal networks provides the most fruitful micro-macro bridge. In one way or another, it is through these networks that small-scale interaction becomes translated into large-scale patterns, and that these, in turn, feedback into small groups" (Granovetter, 1973(Granovetter, , p. 1360. In his theory of "the strength of weak ties" these interpersonal networks are central to the understanding of how links between individuals create the social fabric. Contrary to obvious interpretations, ties between people do not have to be necessarily strong. Though social bonds exist among closely related persons ("ego networks"), that is relatives and friends, it is the social bridges that integrate many small networks into a larger structure. Both however contribute to the life of a community: strong ties are important for (re)producing various forms of resources within different groups, weak ties are crucial in distributing them further ones. By applying this perspective to the case of deliberative democracy and making necessary adjustments, we can see how direct communication among actors within institutions and indirect action between institutions within the system result in a picture which is empirically plausible and normatively beneficial. Whereas deliberation generates (better) solutions to the problems that a particular society faces, democracy (evenly) spreads ideas among a greater number of people.
According to researchers working in this area, networks simultaneously affect and are affected by discourse and action. Relations and interactions enable certain types of behavior while constraining others. But they do so not due to the power of certain individuals or the whole society but by the very nature of interconnectedness which makes people dependent on one another. As Amirbayer and Goodwin contend, "The point of departure for network analysis is what we shall call the anticategorical imperative. This imperative rejects all attempts to explain human behavior and social processes solely in terms of the categorical attributes of actors, whether individual or collective" (Amirbayer & Goodwin, 1994). It does not mean however that all elements are equal to one another. Surely networks are not egalitarian structures. While they can provide opportunities for some people, they can also greatly limit others. First and foremost, we need to be aware that certain individuals are members of groups with high status. Second, access to such groups is formally or informally restricted to all but certain people. In realworld, it can happen that both cases are true for persons who possess a substantial amount of material and non-material capital (Bourdieu, 1986). The so-called elites are those who benefit from being members of powerful and affluent groups as well as having access to others. It is the strong and weak ties that define the person's capabilities and not only the traits of his/her character or the aspects of the social structure. This leads to a conclusion that network science which is regarded primarily as an empirical science, has a covert normative dimension too. Studying the network structure can bring us closer to an idea of democracy as a rule of the people in which each person has an equal say. Knowing where different individuals and groups are and how they behave in the decision-making we are able to design, develop and implement measures to rectify occurring problems. Focusing only on institutions and the system will give partial answers to the question of how to bring about more justice. After all the latter is itself a complex ideal regulating how particular elements should be arranged.
Network theory is thus crucial for a deliberative democratic theory because it allows to study deliberation in terms of relational and interactional fields of discourse and action. The dominant approaches followed essentially the categorical imperative with its emphasis on properties of actors, institutions or the system in providing favorable conditions for deliberation. The latter result from norms which have been established and activated in the process due to participants' capacity to talk together (deliberative virtue), institutional design aimed at sustaining discussions (deliberative performance) or features of the system enabling authentic discourse (discourse quality). Conversely, the key idea of the network approach is to capture how these actors and institutions are embedded in relations and interactions which enable and constrain individual and collective behavior, and in turn the functioning of the system itself. Favorable conditions of deliberation stem from norms emerging from and being negotiated in those relations and interactions. By investigating the actual ties between communicating actors we are able to analyze and evaluate the patterns of discourse within institutions. However, it is crucial to make a step forward and look also at how the latter engage in (re)shaping the discourse in different locations within the system. This move extends and complicates the unfolding picture by going beyond individual sites and embracing cross-site deliberation between directly connected institutions and indirectly linked networks. On the other hand, we also need to account for the systemic patterns affecting existing relations and interactions.
In this light deliberative democracy can be defined broadly in terms of networks linking institutions to the system. This conceptualization aims at overcoming the chasm between the two by striking a balance between interconnectivity and centrality. Juxtaposing institutions, systems and networks enables us to think of the democratic system as a deliberative structure composed of open systems of interlocking networks linking various institutions. The latter are constituted by ties between their participants who discuss given issues and communicate the collective outcome. Deliberative institutions work at the micro level which is linked upward to the system at the macro level. However, this linkage is not direct but is rather mediated through deliberative networks at the meso level. Internal actor-centered environment feeds into external institution-centered environment and further into public network-centered environment which then feeds back. Democratic system is thus the political space where all horizontal and vertical ex-ante relations, and ex-post interactions between actors, institutions and networks are present (see Figure 1).

Network-based deliberative democracy: a framework for analysis and evaluation
Going beyond formal conceptualizations of the network approach, I will proceed now to describe a framework for studying deliberative democracy. The guiding principle here is what I call "relational pluralism" according to which democratic system will be grasped in radically relational and pluralist terms. All its elements can be disaggregated into particular nodes with ties between them. Even the system itself is not a single entity but rather a set of all possible nodes and ties. In fact, as it is usually perceived in network sciences, democratic system accounts for a semi-open structure comprising all directly and indirectly linked networks. The only non-relational element here is the actor understood as an individual person or a group of people who can act (participation) and be acted upon (representation). It does not mean that all higher-level nodes such as institutions and networks are per se relational. In fact, they have a dual nature depending on the scale in question. Therefore, at the microlevel institutions are regarded as networks of actors but at the meso level they account for nodes in the networks of institutions. Following this logic, at the macro level these networks are nodes tied up to one another in the system which constitutes a network of networks. Hence, the entire framework is built around four key elements: actors, institutions, networks and the system. Except for the former all of them are internally differentiated and externally interrelated structures. First, institutions form relations and interactions between actors who participate and deliberate in a direct communication. Second, networks are composed of relations and interactions between institutions which represent their actors and deliberate in a mediated communication. Finally, the system is based on relations and interactions between networks which represent their institutions and deliberate in a public communication. Actors are thus represented "directly" by institutions in which they participate and "indirectly" by networks. This picture tries to find an equilibrium between individual and collective orientation emphasizing both the bottom-up, micro-deliberative and top-down, macro-democratic perspective. Furthermore, the hierarchical, vertical structure of the system is counterbalanced by the flat, horizontal structures within it.
Each "deliberative level" of the framework is scrutinized along three broad "aspects": architecture, discourse and authority which are further developed into specific "criteria" and "indicators". First and foremost, the Architecture aspect describes how participation and representation are designed in and shaped by institutions, networks and the system. Ideally, actors, institutions and networks should be able to take part directly and indirectly in free and equal deliberation. Moreover, participation and representation ought to be based on or aimed at strengthening political efficacy and inducing common efforts. Next, the Discourse aspect is focused on how deliberation is designed in and shaped by institutions, networks and the system. Ideally, actors, institutions and networks should be able to take part in critical discourse by initiating and changing the agenda, and framing the discussions. What's more, deliberation ought to be based on or oriented towards sustaining respectful talks and generating sound reasons. Finally, the Authority aspect refers to how decision-making is designed in and shaped by institutions, networks and the system. Ideally, actors, institutions and networks should be able to take part in collective opinion-and will-formation by meaningfully contributing to making decisions. Furthermore, decision-making ought to be based on or directed at providing political equality .

Scaling down deliberation: micro-discursive level-actors and deliberative institutions
Let me now focus on the specific criteria and indicators in each of the above aspects. Despite certain nuances they are to a large extent unified. There are three different types of architecture, discourse and authority at each scale: Internal, External and Public, respectively. Starting with micro level, that is network of actors, we can distinguish between three aspects: "Internal architecture", "Internal discourse" and "Internal authority". Altogether they capture the dynamics and structure of institutional deliberation. The "Internal architecture" aspect encompasses four criteria: 1) Inclusive participation, 2) Individual autonomy, 3) Individual control, and 4) Institutional environment. First criterion is indicated by Access to an institution, i.e. to what extent actors can participate in it. It can vary from no access if only certain individuals or groups can take part to full access when participation is open to everyone. Second criterion is indicated by Position of actors, i.e. to what extent they are autonomous within an institution. While weak position means that participants have little opportunities for meaningful action, strong position implies they can act freely and unconditionally. Third criterion is indicated by Influence exerted by actors, i.e. to what extent the latter can control an institution. It stretches from weak influence when there is little ownership of its structure and dynamics to strong influence when participants are in charge of the process. Lastly, fourth criterion is indicated by Collaboration among actors, i.e. to what extent they can work together despite existing differences. No collaboration entails either no intention to collaborate or distrust and even hostility among them. Full collaboration means that participants not only intend to but actually collaborate.
The "Internal discourse" aspect includes four criteria: 1) Institutional agenda-setting, 2) Institutional framing, 3) Inclusive communication, and 4) Rational argumentation. First criterion is indicated by Agenda open to change by actors, i.e. to what extent they are able to propose new and shape existing discussion topics. It extends from closed agenda when there is a pre-defined set of items to open agenda when participants can add, remove or modify them. Second criterion is indicated by Consideration of all options, i.e. to what extent actors frame the issue in a comprehensive manner. Flawed consideration means that participants do not take in account certain viewpoints or aspects of an issue. Careful consideration, on the other hand, tries to capture all relevant perspectives. Third criterion is indicated by Mutual respect between actors, i.e. to what extent they are willing to communicate openly an. It ranges from no respect if participants disregard other viewpoints to full respect when they tend to listen to one another. Lastly, fourth criterion is indicated by Reason-giving between actors, i.e. to what extent they provide arguments to back their positions. Weak reasoning means that participants make claims without supporting them with good reasons. Conversely, strong reasoning implies they do not only express but also explain why their standpoints are superior to others'.
Finally, "Internal authority" aspect has two criteria: 1) Scope of individual power, and 2) Distribution of individual power. First criterion is indicated by Input into decision-making/opinionformation, i.e. to what extent actors have a final say in an institution. It can vary from no input if participants do not contribute to making decisions/forming opinions to full input when they make a significant contribution to the process. Second criterion is indicated by Equal power among the actors, i.e. to what extent they have an equal say in an institution. Unequal power entails differences in authority between the participants. Equal power means that there is either no significant inequality among them or the latter can be easily overcome.

Scaling through deliberation: meso-discursive level-institutions and deliberative networks
Moving to the meso level, that is networks of institutions, we differentiate between three aspects: "External architecture", "External discourse" and "External authority". Altogether they capture the dynamics and structure of network deliberation. The "External architecture" aspect encompasses four criteria: 1) Direct representation, 2) Institutional autonomy, 3) Institutional control, and 4) Network environment. First criterion is indicated by Access to a network, i.e. to what extent actors are represented by institutions in a network. While no access means that only certain individuals or groups are represented, full access implies that representation refers to all. Second criterion is indicated by Position of institutions, i.e. to what extent they are autonomous within a network. It can vary from a weak position if they are substantially constrained to strong position when there are no serious impediments. Third criterion is indicated by Influence exerted by institutions, i.e. to what extent they can control a network. Whereas weak influence means little opportunity to affect the latter, strong influence entails high chance of impacting it. Lastly, fourth criterion is indicated by Cooperation among institutions, i.e. to what extent they can operate together despite existing differences. No cooperation means no intention to cooperate or high levels of distrust and even hostility. Full cooperation equals not only intent but actual and substantial effort to cooperate.
The "External discourse" aspect includes four criteria: 1) Network agenda-setting, 2) Network framing, 3) Institutional dialogue, and 4) Political justification. First criterion is indicated by Agenda open to change by institutions, i.e. to what extent they are able to create new and alter existing discursive patterns. Fixed agenda substantially limits the scope of discourse in a network. Flexible agenda, on the other hand, is equivalent to a wide range of communicative possibilities. Second criterion is indicated by Consideration of all opinions, i.e. to what extent institutions frame the issue by taking into account different interests. While narrow consideration ignores certain claims, wide consideration tries to weigh all valid preferences. Third criterion is indicated by Mutual respect between institutions, i.e. to what extent they are open to dialogue. It ranges from no respect if there is no space for communication to full respect when they respond to one another attentively. Lastly, fourth criterion is indicated by Reason-giving between institutions, i.e. to what extent they provide justifications for their decisions/opinions. Weak reasoning entails a lack of rational basis for their actions. Strong reasoning means that institutions act in a justifiable way.
The "External authority" aspect has two criteria: 1) Scope of institutional power, and 2) Distribution of institutional power. First criterion is indicated by Impact on decision-making/opinion-formation, i.e. to what extent institutions have a final say in a network. It can vary from no impact if they do not influence the process of making decisions/forming opinions to full impact when they substantially affect it. Second criterion is indicated by Equal power among the institutions, i.e. to what extent they have an equal say in a network. Unequal power privileges certain institutions in terms of authority. Equal power tends to level the playing field among them.

Scaling up deliberation: macro-discursive level-networks and deliberative system
Finally, at the macro level, that is at the level of the network of networks, three aspects are important: "Public architecture", "Public discourse" and "Public authority". Altogether they capture the dynamics and structure of systemic deliberation. The "Public architecture" aspect encompasses four criteria: 1) Indirect representation, 2) Network autonomy, 3) Network control, and 4) System environment. First criterion is indicated by Access to the system, i.e. to what extent actors are represented by networks in the system. It stretches from no access if only certain individuals or groups are represented to full access when representation refers to all. Second criterion is indicated by Position of networks, i.e. to what extent they are autonomous within the system. Whereas weak position entails substantial constraints to their functioning, strong position eliminates any major obstacles. Third criterion is indicated by Influence exerted by networks, i.e. to what extent they can control the system. Weak influence accounts for a greatly limited impact on the latter. Strong influence means a good chance to affect it. Lastly, fourth criterion is indicated by Coordination among networks, i.e. to what extent they can function together despite existing differences. No coordination means no intention to coordinate efforts or high levels of distrust and even hostility. Full coordination entails not only intent but actual and substantial effort to coordinate actions.
The "Public discourse" aspect includes four criteria: 1) Public agenda-setting, 2) Systemic framing, 3) Public dialogue, and 4) Public justification. First criterion is indicated by Agenda open to change by networks, i.e. to what extent they are able to form new and transform the existing discursive landscape. It can vary from structured agenda which narrows the boundaries of discourse to unstructured agenda which enlarges the communicative space. Second criterion is indicated by Consideration of the common good, i.e. to what extent networks frame the issue in terms of general welfare. While partial consideration tracks only private interests, impartial consideration tends to identify the public interest. Third criterion is indicated by Mutual respect between networks, i.e. to what extent they sustain dialogue across the system. It extends from no respect if they do not communicate between one another to full respect when communication transcends their borders. Lastly, fourth criterion is indicated by Reason-giving between networks, i.e. to what extent they provide justification for the final decision/opinion. Weak reasoning entails illegitimate basis for action. Strong reasoning means the latter is rightly justified and thus legitimate.
The "Public authority" aspect has two criteria: 1) Scope of political power, and 2) Distribution of political power. First criterion is indicated by Impact on final decisions/opinions, i.e. to what extent networks have a final say in the system. It can vary from no impact if they do not affect the decision/ opinion to decisive impact when they are involved in making/forming it. Second criterion is indicated by Equal power among the networks, i.e. to what extent they have an equal say in the system. Unequal power occurs when there are differences in authority between them. Equal power tends to neutralize these disparities. Table 1 below summarizes the key elements of the above-described framework: By applying the above framework it is possible to carry out comprehensive analysis and evaluation of real-existing democracy through the deliberative lens. Network approach allows to study individual sites and collective processes arising from the former gradually and progressively. Decisions and opinions emerge bottom-up but this emergence is structured by various factors. Networks of institutions are the cornerstone of the entire structure because they describe and prescribe how interpersonal discourse in small groups is and should be translated and integrated into large-scale transpersonal discursive patterns. Hence, the network perspective combines institutional and systemic approaches without overlooking any relevant dimension.
Let me illustrate how this framework is useful in analyzing and evaluating real-world cases. I will take two examples matching the institutional models defined above. The first "liberal" case focuses on the response of the US government to the economic crisis of 2008. The second "republican" case deals with the implementation of participatory budgeting in the city of Warsaw (PBW), Poland in 2014. From network-deliberative perspective, it is important to start the analysis by mapping all key actors and institutions that either were involved or had an interest in the decision-making. In the American example, these are the President and his administration, the Congress (House of Representatives and Senate), judiciary with the Supreme Court, government agencies, political parties, media outlets, interest groups and social movements. Polish example differs here as it centers on the PBW internal structure: the Council, district teams, district coordinators, discussion meetings and public consultations. But the list of actors/institutions extends further to include the Mayor and her office, City Council, district councils, district boards, local parties, local media, citizen groups, social movements. Next step is to analyze how all these actors/institutions performed independently before proceeding to track the relations and interactions between them. In this way we can see what kind of deliberative networks emerge and how the ensuing linkages affect and are affected by the dispersed institutional deliberations. In the American case of particular importance is the way and extent to which Obama administration managed to shape discussions and influence the Congress, both major parties as well as the wider public. As for the Polish case, main focus should be placed on the discourse and action within and outside the PBW structure, that is between the internal and external actors/institutions. Finally, the last step is to track relations and interactions between networks themselves in order to see which parts of the democratic system were dominant in the process. The key aspect of the American example is the location of the Obama administration in the tight web of influences leading to the final outcome. In the Polish example it is crucial to define the position of PBW within the existing discursive and political structure of urban politics.

Conclusion
In this article, I proposed a network approach to deliberative democracy which integrates the institutional and systemic approaches in order to provide a novel and robust conceptual framework. By "robust" I mean here a perspective which is rooted in norms and facts, thus enabling analysis and assessment of real-existing democracies. I argued that deliberative institutions and deliberative system are not sufficient to provide an adequate basis for deliberation as they struggle with the problem of embeddedness. On the contrary, network approach comes to terms with the inevitability of trade-offs between microdeliberation and macro-democracy. However, it does not reject both approaches but ensures that their advantages are maintained while their shortcomings are mitigated. Although network approach compromises the potential of institutions as genuine arenas of deliberation and the potential of the system as a genuine venue of democracy, it retains the core of both in order to advance a third "middle-range" alternative which was vaguely present in the literature but never sufficiently theorized. By focusing on relations and interactions connecting the institutions, it paves the way towards the concept of deliberative democracy as a fully fledged network. Therefore, it is able to bridge micro-deliberation and macro-democracy plausibly showing how they interlock and overlap. Under these approach institutions and the system presuppose each other. The false independence of both realms is neutralized by pointing to the space where they are in fact interdependent. The missing link between micro and macro levels which resolves the problem of embeddedness is provided by networks located at the meso level. Thanks to this, we are able to see how institutional deliberation is part and parcel of the system and vice versa, how patterns of systemic deliberation are re-created in institutions. The networks qua deliberative systems play a key role here as "communicative funnels" which structure the discourse of institutions and transmit it to the democratic system. It is the latter that makes the essential political superstructure responsible for making legitimate decisions and forming reasoned opinions.
Knowing how deliberative democracy works at different levels and how they are interconnected, we are able not only to conduct empirical research but also design institutions, networks and even entire systems. First and foremost, the proposed framework can be valuable in formulating hypotheses and testing them in various settings. Although it needs further elaboration, it constitutes a good starting point for specific (case) studies delving into the nature of deliberative interdependencies. Furthermore, future research can focus on links between different deliberative scales. Not only do we know little how institutional deliberation and systemic deliberation coincide, there is also scarce data on how local, national and global deliberation interpenetrate one another. Finally, a network approach can be useful in designing institutions, networks and systems. So far only the first area has been investigated. However, it is crucial to go beyond designing individual sites of deliberation and enter a more difficult and challenging domain, that is how these different sites can be linked together and how these links are further connected to form an overarching mechanism. Not only is this important in creating new settings, it may also help in bringing together existing ones which are located on the edges or even outside current structures. Therefore, if we can succeed in substantiating the network approach with theoretical and empirical evidence, then deliberative democracy might finally "come of age" (Bohman, 1998), this time however, in order to become truly, and not seemingly a "revolutionary idea" (Fung, 2005).

Correction
This article has been republished with minor changes. These changes do not impact the academic content of the article.

Citation information
Cite this article as: Beyond institutions and the system: Network approach to deliberative democracy, Marcin Zgiep, Cogent Social Sciences (2019), 5: 1639876.

Notes
1. See Chambers (2009). Proposal put forth by Gastil in his book "Political Communication and Deliberation" comes very close to this goal. But even he does not realize the potential of contestation which has become integral part of the paradigm in recent years. By overlooking contestatory deliberation his framework falls short of including radical forms of discourse present in a democracy. Gastil (2008). 2. see Chambers (2017). Recent Irish case contradicts this argument.