When Do They Speak? Deliberation and Democratic Decision-Making in the European Union

The Democratic Deficit in the EU: A Review

A standard definition of the democratic deficit in the EU describes the partial influence of the addressees of decisions taken by the EU governing bodies on the contents of those decisions (Lodge, 1994). I do not want to take a side in the debate. My aim is to draw upon the existing literature to single out those common features that may ground an alternative assessment of democratic legitimacy in the EU.

Much of the scepticism about the democratic legitimacy of EU institutions assumes that reliable explanations of the democratic deficit are amenable to two phases: lack of representation (input) and inefficient policy delivery (output). Advocates of input legitimacy claim that the EU lacks the institutional mechanisms that guarantee communication between affected members and decision-makers (Bellamy, 2013; Hix, 2008; Kohler-Koch, 2010a). For proponents of output legitimacy, the legitimacy of institutions is measured against the problem-solving quality of laws and norms (Scharpf, 1999). They stress that the EU lacks democratic legitimacy because its policy outcomes are biased against particular interests of affected members and affected by structural barriers or ideological prejudices. On this view, what matters is the capacity of the EU to generate policies that actually ameliorate individual lives, resonate in collective identities or meet the technical views of the experts.

In an attempt to move beyond the input–output narrative, an alternative strand of literature has centred the debate on democratic legitimacy in the EU around the deliberative aspects of decision-making processes at the supranational level (Schmidt, 2006, 2010; Wimmel, 2008). ‘Throughput legitimacy’, Schmidt (2010: 7) says, ‘is a performance criterion centring on what goes on inside the “black box” of the political system’. Wimmel (2008: 11) stresses a similar point: throughput legitimacy encompasses all those practices that occur ‘after the citizens have exhausted their direct or indirect opportunities for participation and have usually delegated their preferences, [when] representative, or lobbyists enter the political scene [and] negotiate and decide in the complex arrangement of institutional procedures’. On this view, the legitimacy of the EU correlates with the necessary requisites of a well-functioning, pluralist and multilevel deliberative process. These requisites include the accountability of experts; openness to civil-society organisations, interest groups, social movements and international non-governmental organisations; transparent decisions and credible strategies of policy dissemination in the member states (Bistagnino, 2017).

Despite the lack of consensus and the wealth of reciprocally critical views on how these ideas can be tested empirically, it seems very difficult to deny that the phases of input, throughput and output are inevitably interconnected. And deliberation at the supranational level is a shared object of concern. Actually, input-based accounts cast light upon the demand for spaces of preference articulation and contestation; a throughput narrative focuses on the demands for accountable, plural and transparent ways to negotiate such preferences; and output-oriented scholars direct attention towards the mismatch between the interests of all those affected and collectively binding outcomes. When we look at the supranational level, output efficiency and deliberation are the primary legitimising mechanisms (Moravcsik, 2002; Schmidt, 2010), while political participation, activism and principal–agent representation are mainly centred at the national level (Hix and Høyland, 2013). As it is now, output policies and throughput processes at the EU level can affect the work of input mechanisms at the national level while downsizing the relevance of national deliberation and channelling the impact of national output policies (Schmidt, 2010: 7–10). In this way, in member states, decision-makers are not fully accountable for their policy directions, output policies are likely to be perceived as alien impositions, and pluralism of policy solutions might be contained. Together, this picture may strengthen the widespread belief in a democratic deficit in the EU.

With the growing attention to the problem, prominent scholars of European integration have challenged the thesis of a democratic deficit in the EU at different levels (i.e. Majone, 1994, 1996, 2000; Moravcsik, 2004; Trenz and Eder, 2004). For instance, Giandomenico Majone (1994, 1996, 2000) argues that there is no democratic deficit in the EU. This does not mean that EU policy-making is perfect, but that it should not be democratic in the usual meaning of the term. In his view, the EU is ‘a regulatory state, which comprises the creation of the single market, common services, product standards, and a monetary policy’ (Majone, 1996). To be democratic, this regulatory state should address and correct market failures to produce policy outcomes in which no one is made worse off. For him, greater professionalism and technical expertise, the media at both the EU and national levels, transparent decision-making and ex post review by courts are among the mechanisms that would make the public accept the EU as legitimate (Majone, 2000: 274–276).

Andrew Moravcsik argues that institutional checks and balances, indirect democratic control, the strengthening of the European Parliament and the inclusion of national parliaments in supranational legislation enhance deliberation and ensure that EU policy-making limits policy drifts. In his own words, ‘this myriad of institutional constraints not only render arbitrary and capricious action almost impossible but assures that legislation outside Brussels is likely to represent an exceptionally broad consensus among different groups at many levels of governance’ (Moravcsik, 2004: 349).

Now let us try to bracket disagreement and isolate the points of contact across different accounts of democratic legitimacy in the EU. First, there is widespread recognition that decision-making procedures in the EU are somehow deliberative (Moravcsik, 2004: 349; Schmidt, 2010). The degrees of inclusiveness, accountability and efficiency of such deliberative mechanisms are instead an important object of disagreement. Second, it is commonplace to accept that the EU is a multilevel system in which decision-making procedures consist both of national and supranational processes (Bartolini, 2005). Usually, disagreement is about the distribution of decision-making power across relevant actors at the two levels. Third, there is an often neglected epistemic element across the literature on the democratic deficit in the EU. Scholars acknowledge that decision-makers at the supranational level might benefit from differences across the EU. In this case, disagreement tends to be on the designation of the best actors to introduce and combine these interests in collectively binding decisions. None of the available accounts goes as far as contesting the existence of such a mix of interests. In addition to the deliberative, multilevel and epistemic components, I hold that the democratic legitimacy of the EU may be weighed against the kinds of decision it makes, given the presence of many, and sometimes otherwise neglected, interests.

Together, these four elements help me define the conceptual framework of my article. In the following, I shall maintain that the EU has an architecture to reach collectively binding decisions that is multilevel and somehow deliberative and may be responsive to different degrees to the available interests.

Some Assumptions

For a long time, political theorists have been hesitant to uphold democracy at the supranational level. Deliberation at the supranational level multiplies problems of scale that already may undermine deliberative mechanisms at the state level (Follesdal, 2014). Nevertheless, the presence of several imperfections of supranational deliberation does not prevent us from thinking of the EU in deliberative terms. This does not mean that we need to construct an idealised account of the EU, but rather it demands a certain degree of consistency with the existing literature on deliberation beyond the nation-state (e.g. Dryzek, 2006; Eriksen and Fossum, 2000; Parkinson and Mansbridge, 2012; Smith and Brassett, 2011). It implies a recognition of the EU efforts to make the decision-making process conducive to on-average-good decisions that are inclusive, transparent and accountable (e.g. COMM, 2001, 2015, 2016; Schmidt, 2006, 2010). It also implies a careful construction of my argument in light of the vast literature on deliberation in non-ideal political scenarios (e.g. Bächtinger et al., 2010; Fung, 2005; Goodin, 2005). For these reasons, in this section, I make my assumptions explicit.

The first assumption of this article is that a legitimate and democratic EU is desirable. Arguments for and against a democratic deficit in the EU do not question the desirability of the EU as a democratic and legitimate regime. Disagreement is mostly about the expectations we should have about the legitimacy of a democratic supranational system (Kohler-Koch and Rittenberg, 2007; Steffek and Nanz, 2008: 9). Now, if a legitimate and democratic EU is desirable, and if we keep in mind its multilevel and somehow deliberative architecture, then the improvement of deliberative mechanisms may be one of the ways to make the EU more legitimate and democratic. To help do so, I focus on the communicative processes that precede and follow voting on collectively binding decisions. This entails two things. First, single decision-making processes are the units of analysis. Here I consider the legitimacy of EU decisions, not the overall legitimacy of the EU. Second, in line with the so-called deliberative turn in political theory, I agree that accountability may be a source of legitimacy (i.e. Chambers, 2003: 309; Dryzek, 1996: 475–476), but I add that a legitimate decision is also a decision that solves a problem by meeting collectively constructed criteria (Anderson, 2006, 2008; Risse and Klein, 2007: 74; Scharpf, 1970: 21; Steffek, 2015: 265–266), criteria that are the result of inclusive deliberation (Anderson, 2003, 2006). Even if it is possible for technocrats to determine what meets such criteria, there may also be shifts of interest in the constituency (Harmon, 1969). Therefore, insofar as there is a correct answer to be found to some questions, one of the possible ways to improve the democratic quality of governments and ensure that outputs serve general interests (Scharpf, 1970) is to extend the range of participants in intersubjective processes of communication (Anderson, 2006; Landemore, 2012; Page, 2008) in which, in front of other views, they can share and rethink their self-interest (Mansbridge et al., 2010). Inclusion may ensure participation in the construction of standards, and pluralism would favour the discovery of biases and prejudices (Anderson, 2006, 2008). In other words, there are both epistemic and democratic reasons to uphold a strategy of inclusive deliberative mechanisms (Mansbridge et al., 2012). This is the third assumption of my article. This strategy is not new in the EU. For academic, political and administrative actors, as Sabine Saurugger (2010) documents in great detail, inclusion of civil-society groups in decision-making processes became a political imperative at the European level during the late 1990s and culminated in the White Paper on Governance (Follesdal, 2003), the introduction of Article 47 on ‘participatory democracy’ in the constitutional treaty and the introduction of Article 8b of the Lisbon Treaty. Within this context, for several years, EU institutions have seen civil-society groups as a panacea to crises in democratic legitimacy (Greenwood, 2007, 2012, 2017; Kohler-Koch, 2009; Kohler-Koch and Quittkat, 2013).

But what do I mean by civil-society groups? To the observers familiar with the debate about civil-society groups and democratic deliberation inside (e.g. Bouza Garcia, 2010, 2015; Greenwood, 2007, 2017; Kohler-Koch and Quittkat, 2013; and outside the EU (e.g. Brassett and Smith, 2010; Chambers and Kopstein, 2009: 370; Cohen and Arato, 1994; Dryzek, 2006; Habermas, 1996: 370; Hendriks, 2006), it might be problematic to accept a unitary definition of civil-society groups. In the following, I will try to remain as normatively agnostic as possible with respect to specific conceptions of civil society. In so doing, I will refer to the mainstream account within EU institutions. As the Glossary of EUR-Lex says, the expression ‘civil society groups’ refers to ‘all forms of social action carried out by individuals or groups who are neither connected to, nor managed by, the State’. Examples of such groups include trade unions, employers’ groups, non-governmental organisations and grassroots organisations.

Often, inclusion of civil-society groups seems to be the equivalent of their institutionalisation (Bouza Garcia, 2015; Greenwood, 2007: 347; Warleigh, 2001). Within the EU, institutionalisation may have two meanings: first, it may mean direct engagement with institutions and officials (Greenwood, 2017; Kohler-Koch and Finke, 2007; Schmidt, 2010); second, it may also entail the creation of a European public sphere in which ‘national public authorities, civil society, and the European institutions need to work together’ (COMM, 2006: 5). By focusing mainly on the legitimation of supranational institutions, the latter approach neglects the possibility that including civil-society groups may corrode the vitality of those groups (Kohler-Koch and Quittkat, 2013: 179) and democratising action outside the hardware of institutional practices (Dryzek, 2000). In this article, I acknowledge scepticism concerning the ways civil-society groups might mediate between the EU and citizens. Kohler-Koch shows that civil-society groups struggle to achieve equal and effective participation, and therefore they contribute little to the democratic legitimacy of the EU (Kohler-Koch, 2012: 814, 820). However, I hold that, if civil-society groups bring interests to the table that otherwise would be neglected, their presence in institutionalised deliberative processes is conducive to more legitimate decisions on both epistemic and democratic terms. The idea that civil-society groups always bring otherwise neglected interests to the table is disputable on both conceptual (Steffek and Ferretti, 2009) and empirical terms (Mahoney and Beckstrand, 2011; Steffek and Nanz, 2008). ¹ Nevertheless, my view follows existing accounts that see civil-society groups as bringing added epistemic value to the democratic decision-making procedure (Anderson, 2006; Stevenson, 2016). In the same vein, within the context of EU politics, such a perspective echoes the Commission’s attention to civil-society groups as potential agents of both input and output legitimacy (COMM, 2001; Greenwood, 2007, 2017; Kohler-Koch, 2009; Kohler-Koch and Quittkat, 2013). On such a view, EU institutions ought to include civil-society groups in the decision-making process because these groups bring to the fore interests of constituencies that are not necessarily represented in the Commission, in the European Council and in the European Parliament. And inclusion of different interests in the decision-making process is seen as conducive to decisions that are less affected by the most vociferous groups (Steffek, 2015: 287).

The Experimentalist Model of Democracy

I think an alternative interpretation of legitimacy in the EU may be grounded on the constructive function of deliberation and its capacity, through the inclusion of different interests, to produce decisions that are good on average. This reading brings us to the context of epistemic arguments for democratic legitimacy. Over the past three decades, people (e.g. Goodin and Spiekermann, 2018; List and Goodin, 2001) have defended democratic mechanisms to aggregate knowledge as the best way to achieve correct decisions. Meanwhile, other theorists have argued for the epistemic and democratic benefits of inclusive deliberation (e.g. Cohen, 1997; Estlund, 1997; Habermas, 1996, 1997; Landemore, 2012; Marti, 2006). Within this tradition, the experimentalist model of democracy is a deliberative model with an aggregative component. Voting expresses beliefs about what the correct solutions are according to public interest. Then, through an analogy between democracy and scientific practice, Elizabeth Anderson (2006, 2014) draws upon John Dewey (1976) to claim that democratic and inclusive deliberation is like a thought experiment: we advance hypothetical and biased responses to problems, we try to imagine the consequences of these alleged solutions from our own point of view, we correct biases through deliberation, we attempt to foresee their impact and we have favourable or unfavourable reactions to them. In this way, as Anderson (2003: 250) writes, properly constructed procedures, which would include different interests for shared ends, can make democracy promote ‘government responsiveness to the people’s needs’.

The experimentalist model of democracy is not the only non-static model of deliberation in non-ideal political conditions (Curato, 2012; Goodin, 2005; Parkinson, 2006). Nicole Curato (2012: 426) and Robert Goodin (2005: 188) show that the applicability of deliberative virtues is linked with the specific requirements of particular stages of deliberation. However, these models do not consider when a deliberator participates in deliberation, but rather they depict how a deliberative agent should or could behave at different phases of the process (Curato, 2012: 429; Goodin, 2005: 194). The underlying assumption is that different agents participate from the beginning to the end of the deliberative process. John Parkinson (2006: chap. 7) also offers a sequential model of legitimate large-scale deliberative democracy in which the deliberative process is divided into four stages (define, discuss, decide, implement). This account has many affinities with the experimentalist model of democracy. For instance, both models give citizens a prominent role in the deliberative system, are open to various kinds of deliberation in formal and informal settings, subordinate the enactment of collectively binding decisions to talks and extend the perspective to the agenda-setting phase and to post-decision moments. Nevertheless, the experimentalist model of democracy adds an epistemic component to Parkinson’s four-stage model. Specifically, according to Dewey and Anderson (2006), a decision is legitimate when (a) it is the result of an inclusive and egalitarian process and (b) it respects collectively constructed standards of appropriateness.

On this view, democratic deliberation has three phases: talk, vote and feedback. As Anderson (2006: 9) writes, discussion is needed to make certain private preferences a matter of public interest. Parties transmit information about private interests at the beginning of the process, but such interests are a result of incomplete, unsure, partial and incomplete information. Inclusive deliberation should alter such interests in desirable ways. Namely, in voting the solution that advantages or benefits a person or a group, parties take into account the views of others (Anderson, 2006: 11). Democratic deliberation begins with talks on the character and scope of possible objects of disagreement. It continues with exchanges on the set of possible solutions to such an object of disagreement. These interactions should identify and correct implicit biases. Once a decision has been made (voting phase), deliberation continues (feedback phase). According to Anderson, in the talk and feedback phases, deliberation benefits from the inclusion of significantly different interests. When a large number of different voices participate on an equal footing in the definition of issues of common concern, the system is more likely to overcome particularistic interests. When members with different interests can deliberate on a matter of public interest, they can individuate systemic biases and perhaps correct them.

Within the experimentalist model of democracy, when parties do not participate directly, they have a representative who enters the relevant deliberative phases after a ‘pre-deliberative stage’ or ‘caucus of the likeminded’ in which parties develop something like a common self-interest (Mansbridge, 2006: 117). Deliberation helps clarify and maybe tone down this common concern for the sake of what is genuinely in the public interest. Within the framework provided by an issue that is truly of public interest, the process of reaching a decision is also a process whereby individual preferences may change. In a similar way, post-voting deliberation is supposed to enable citizens to clarify information and to sharpen their own preferences. So the benchmark for evaluating the merits of a democratic system should be that decisions are both suitable according to independent criteria and designed by the affected members according to their views on a given issue.

In this vein, Anderson emphasises that the exclusion of some relevant interests from deliberation may undermine the hoped-for democratic legitimacy in significant ways. When the interests of some less powerful affected members are left aside, the claim that deliberative processes are dealing with a problem of public interest is undermined (Anderson, 2006: 15). It is therefore crucial to begin by identifying problems that, being of public interest, may be assigned to democratic action. Public (or general) interest is an idea that has attracted significant attention in empirical and theoretical approaches to politics (Moravcsik and Sangiovanni, 2003; Scharpf, 1970; Steffek, 2015). According to the experimentalist model of democracy, whether a problem counts as of genuinely public interest is determined in part by external criteria and in part ‘by whether citizens or their representatives affirm its place on the public agenda through procedurally fair decision-making processes’ (Anderson, 2006: 10). If fair, such processes should entitle all groups, regardless of their size, to define the agenda and compromise on what is truly of public interest (Anderson, 2006: 16). ² Within this context, as Anderson argues, three conditions should be respected: (a) the solution to problems of public interest requires joint action by citizens and (b) demands action through the law; moreover, (c) the observation that a private preference is widely held among those affected does not necessarily turn the issue at stake into a problem of public interest. Here, deliberation is needed in order to ‘articulate proposals to make certain concerns a matter of public interest’ (Anderson, 2006: 10). Likewise, inclusive deliberative procedures help acknowledge systematic biases, correct them and take steps to assess possible solutions to disagreement in a different way (Anderson, 2014: 6). ³ For instance, Anderson (2014: 6) writes, wishful thinking may lead policy-makers ‘to exaggerate the effects of measures designed to produce some good, and hence to adopt those measures even if they don’t really work’. But the fundamental test of solutions to disagreement is ex post, based on our experiences upon their adoption. To test a norm, Anderson (2014: 6) writes, ‘we act in accordance with it and consider whether we can live with the consequences’. By testing norms in practice, affected members may come to change their views of the problems they are meant to solve, in light of the experiences of living with them (Anderson, 2014: 6–8). On this view, ex post dissent has a fundamental importance because methods of self-correction can both increase the epistemic power of democracy and ensure that decision-making is deliberative rather than imposed. For this reason, citizens must be ‘free to dissent not just at the voting stage, but after a decision is made … Without an opposition to remind the public of continuing objections, and to pose alternative’, nothing would force decision-makers to reconsider their decisions (Anderson, 2006: 17–18).

Therefore, an epistemic point of view substantiates the idea that, being the result of inclusive deliberation, a legitimate decision must be perfectible and approximate standards of appropriateness. This does not mean that, for Anderson (2006), alternative means to collect different interests and reach appropriate decisions would be equally desirable. Deliberation remains a necessary condition for legitimacy. According to Anderson, it ensures the inclusion of different views and parity among the participants. But the epistemic standpoint is a corrective that helps evaluate outcomes of deliberative procedures for standards that transcend pure accountability and whose validity extends well beyond the voting phase. ⁴ At the normative level, this suggests a differentiated model of evaluating the legitimacy of collectively binding decisions. First, a decision must be the result of procedures that involve affected members in the definition of problems and epistemic criteria of evaluation. Second, a decision is legitimate when it aligns with collectively constructed criteria of appropriateness and the individual experiences of all those subject to it. Then when a democratic regime makes a decision in this way, we can think of it as having legitimate authority with respect to its citizens.

Captain Hook Politics and Garbage-Time Politics

In its essential form, as we have seen, the experimentalist model has three phases: talk (between t₁ and t₂), voting (t₂) and feedback (from t₃ to tₓ). To this scheme, we should add t₀ (prior to t₁), when parties express their point of view on a certain issue and bring the issue to the attention of a wider public. From this perspective, for a decision to be legitimate, between t₀ and t₁ all affected members must agree that the issue under scrutiny is of public interest; between t₁ and t₂, all affected members must deliberate on possible solutions to disagreement; at t₂, they must vote on a collectively binding solution. This decision is legitimate because inclusive deliberation ensures that members have polished reciprocal interests up to the point at which all possible solutions in the array of available choices are unbiased.

However, the legitimacy of the t₂ outcome is still contingent on its being put into practice. At t₃, the implementation of the t₂ outcome and the experience of living with it may cause the articulation of new interests. If this is the case, deliberation is required to rethink the t₂ outcome in light of the new available interests. The resulting outcome, let us call it the t₂ revised outcome, then may be the object of further deliberation when new interests arise. In this view, the significance of new interests at t₃ is twofold. First, if included, they strengthen the comparative epistemic merit of the t₂ revised outcome against possible alternatives (i.e. the t₂ outcome, the t₂ partially revised outcome, the t₂ more-than-partially revised outcome and the like). From this perspective, the t₂ revised outcome would be the best possible synthesis of a considerable amount of different interests. Second, the exclusion of new and relevant interests may undermine the hoped-for epistemic merit of collectively binding solutions. In this case, it is not beyond a reasonable doubt that the t₂ revised outcome is better than the t₂ outcome, the t₂ partially revised outcome or the t₂ more-than-partially revised outcome.

This picture helps us clarify why, according to the experimentalist model of democracy, including civil-society groups is a proxy for the construction of legitimate democratic decisions. That is, if we hold that including civil-society groups brings otherwise neglected interests into deliberation, inclusion is conducive to better outcomes since deliberation among parties with different interests, especially those interests that would be otherwise neglected, may help unveil particularistic and implicit biases before and after voting on collectively binding decisions. Moreover, deconstructing decision-making processes into a series of deliberative phases helps us grasp why the moment in time at which civil-society groups bring in otherwise neglected interests is decisive for evaluating the legitimacy of collectively binding outcomes. Accordingly, for a decision to have full legitimacy, civil-society groups ought to be included in all deliberative phases. This means that the process should be structured in ways that allow civil-society groups to convey different interests arising from the experience of living with a t₂ outcome.

Specifically, when deliberation does not account for otherwise neglected interests at t₁, civil-society groups may have room to contest the legitimacy of collectively binding decisions on epistemic terms at t₂ and between t₃ and tₓ. The exclusion of some of the available interests may let some implicit biases persist. This might also occur when civil-society groups are allowed to participate in the decision-making process, but t₂ outcomes systematically neglect the interests they defend.

So, in the deliberative process, the experimentalist model of democracy identifies three critical junctures – (A), (B) and (C) (Anderson, 2006: 10–16) – at which the legitimacy of collectively binding decisions is at stake in democratic and epistemic terms. The first juncture is at t₀, when biases and power may control the agenda and the definition of general interest. Here the expression of different points of view is a necessary condition to define what counts as a problem of public interest (Anderson, 2006: 16). The first test (A) for legitimacy is the inclusion of all different viewpoints in the definition of what is actually a problem of general concern. The idea is that a rich and diverse articulation of proposals to make certain concerns matters of public interest is a way to counterbalance the presence of more powerful and vociferous groups (Anderson, 2006: 9). Within this framework, when decision-making procedures do not consider matters that are truly of public interest, solutions cannot even claim legitimacy, regardless of the degree of inclusion all along the policy life cycle.

The second test for legitimacy is the inclusion of all affected interests in the definition of policy proposals (B). This assessment occurs just prior to the voting procedure. Legitimacy also depends on the inclusion of all affected interests in the evaluation and reassessment of collectively binding decisions (C). This is the third juncture. The combination of (B) and (C) introduces four new scenarios. Notably, if, at (B) and at (C), affected civil-society groups recognise that otherwise neglected interests have been excluded from deliberative procedures, and other affected parties cannot reject this claim, institutions-relative-actions are illegitimate. Conversely, if, at (B) and at (C), affected civil-society groups cannot reject the claim that otherwise neglected interests have been included in the deliberative process, institutions-relative-actions are legitimate.

The experimentalist model of democracy, I argue, introduces two other scenarios. First, at (B), civil-society groups cannot reject the claim that otherwise neglected interests have been included early in the deliberative process, but, at (C), civil-society groups recognise that such interests have been excluded from the deliberation prior to voting, and other parties cannot reject this claim. This is what I will call Captain Hook politics. ⁵ Second, at (B), civil-society groups recognise that otherwise neglected interests have been excluded from deliberative procedures, and other parties cannot reject this claim, but, at (C), affected civil-society groups participate in the revision of the t₂ outcome. This is what I will call garbage-time politics. Despite including otherwise neglected interests in the deliberative process, Captain Hook politics and garbage-time politics lead to illegitimate decisions.

According to deliberative democrats, it is not new to think that those decisions that are not demonstrably in the public interest cannot aspire to legitimacy at all (e.g. Habermas, 1996; Steffek, 2015). I agree with this view. Exclusion from deliberative decision-making processes is also an important object of study (e.g. Kohler-Koch and Quittkat, 2013; Steffek and Nanz, 2008). Much less has been said on the other two cases. For this reason, by applying the experimentalist model of democracy to the EU, I shall demonstrate that Captain Hook politics and garbage-time politics have both explanatory and normative power within non-ideal deliberative settings. On the one hand, real-world examples show that these two notions help capture existing democratic pathologies in the EU. On the other hand, Captain Hook politics and garbage-time politics suggest lines of action to improve legitimacy of democratic decisions and democratisation.

Captain Hook Politics and Garbage-Time Politics in the EU

For a long time, scholars have debated the deliberative potential of EU politics (Christiansen et al., 2004; Curtin, 2003; Eriksen and Fossum, 2000; Joerges and Neyer, 1997; Risse and Kleine, 2010; Steffek, 2015) and the epistemic merits of supranational decisions (Beetham and Lord, 1998; Majone, 1996; Scharpf, 1970, 1999; Steffek, 2015). Moreover, in the past few years, the EU has tried to provide venues for deliberation all along the policy life cycle, from agenda setting to input and feedback mechanisms of monitoring and evaluation (COMM, 2001, 2015, 2016). For these reasons, the EU provides a laboratory for the first application of the experimentalist model of democracy at the supranational level. ⁶

In dialogue with analyses on the (actual or alleged) exclusion of certain interests from the entire decision-making process (e.g. Friedrich, 2011; Greenwood, 2017; Kohler-Koch, 2010; Steffek and Nanz, 2008), an application of the experimentalist model of democracy to the EU can help identify other possible obstacles that may undermine the search for legitimacy of supranational democratic institutions through deliberative practices. From the perspective of this article, when its multilevel, somehow deliberative architecture includes all affected interests in the talk and feedback phases, and the collectively binding decisions align with collectively constructed criteria of appropriateness, EU decisions are legitimate. That preferences and interests change for the better in the actual practice of deliberation is still a contested matter (Kuyper, 2018), especially in the context of EU politics (Christiansen et al., 2004). However, many scholars (Eriksen and Fossum, 2012; Lord, 1998) recognise in the EU institutional structure ample room for interest transformation. In the framework of a consensus democracy (Lord, 1998: 47–48), significant resources are invested into procedural strategies to work out disagreements over the interests that participants bring into play (Eriksen and Fossum, 2012).

Crucial for the present approach to democratic legitimacy is also the moment at which civil-society groups bring otherwise neglected interests into the deliberation process. Therefore, in the application of the experimentalist model to the EU, it is necessary to identify the three phases. By drawing upon the literature on deliberation in the EU, I hold that talk encompasses the deliberative mechanisms, at both the national and supranational levels, prior to ratification that aim to build consensus and legitimise policy decisions (e.g. Curtin, 2003; Eriksen and Fossum, 2000; Risse and Klein, 2010; Schmidt, 2006, 2010). This view follows arguments about the so-called Community method of EU governance (e.g. Friedrich, 2011; Schmidt, 2006, 2010; Steffek, 2015). Here, deliberative mechanisms include formal debates inside EU institutions, readings by the Council, readings by the European Parliament, opinions from the European Economic and Social Committee and the Committee of the Regions, and commission hearings, intraparty talks, institutional provisions facilitating interest group access, media activity, lobbying, public hearings, political talks among activists, mini-publics and assemblies of experts, online public consultation, committee hearings within member states and online public consultations (Schmidt, 2006, 2010). In this way, talk is an all-encompassing phase that includes a number of deliberative practices with different intents and actors in which civil-society groups are seen as potential agents of both input and output legitimacy (Kohler-Koch, 2009; Schmidt, 2006, 2010; Steffek, 2015).

Voting is the adoption of a policy measure at the supranational level. Feedback incorporates ratification, implementation, monitoring and evaluation of a collectively binding policy measure. According to a series of documents recently published by the European Commission (COMM, 2015, 2016), feedback includes data collection, screening of sectoral legislations, repealing outdated legislations, open public consultations, fitness checks (COMM, 2015: 10–11) and exchanges with ‘high level experts from business, social partners, and civil society appointed through an open and transparent process as well as experts from all 28 Member States, the European Economic and Social Committee and the Committee of the Regions’ (COMM, 2015: 13). A broader view of feedback mechanisms, which I share, also includes practices that are not purely deliberative, such as media commentaries (Parkinson, 2006), post-decision activism, petitions, strikes, civil-society groups’ mobilisations and advocacy (Fung, 2005).

In this view, t₀ may be a moment at which affected parties pressure supranational institutions to consider an issue to be of public interest, and it may also be the moment at which the European Commission formulates a proposal. Here, the phase between t₁ and t₂ includes all deliberative practices, inside and outside EU institutions, until an act is adopted or not adopted (t₂). The growth of an international and vibrant epistemic community, together with EU efforts to ensure pluralism, transparency and accountability prior to ratifying collectively binding policy solutions (COMM, 2001; Curtin, 2003; Greenwood, 2007: 346; Klüver, 2013; Knodt et al., 2011; Kohler-Koch and Quittkat, 2013: 152; Quittkat, 2011; Saurugger, 2008), could give credit to the thesis of input legitimacy, with the EU deliberation, between t₁ and t₂, reducing systemic partisan, functional and territorial biases. A scan of the literature produces a wealth of evidence against this thesis (Greenwood, 2007: 347; Mahoney and Beckstrand, 2011). For instance, certain powerful interest groups may highlight some aspects of a proposal to strategically shape a legislative debate in their favour (Kluver et al., 2015). Elite groups may have the capacity to monopolise policy-making inputs (Christiansen et al., 2004). And the Commission may have a conservative bias that tends to give more prominence to the usual stakeholders (Quittkat, 2011).

The phase between t₃ and tₓ encompasses exchanges, not necessarily deliberative, at the national and supranational levels that discuss the t₂ outcome. EU attention to feedback mechanisms is a relatively new phenomenon (COMM, 2015: 9, 2016). For a long time, post-decision political mobilisation inside and outside national parliaments could not enter the institutionalised decision-making process at the supranational level.

Moreover, civil-society groups may affect only the early stages of the decision-making procedures – Captain Hook politics. Dawid Friedrich (2007, 2011) shows that, in the EU, as the decision-making process draws to a close, interests of non-state actors are often ignored. For instance, within the context of migration policies, the directive on family reunification revealed little uptake of civil-society concerns. Early in the process, the Commission wanted to achieve an encompassing directive, which was applauded by civil-society groups. However, over time, civil-society groups saw their stakes disappear and the second amended proposal was restricted in all areas (Friedrich, 2007).

Garbage-time politics occurs when civil-society groups are denied access to institutions-led deliberative processes prior to voting, but they channel otherwise neglected interests in post-decision dissent outside institutions. For example, as Dür and Mateo (2014) write, in 2011 and 2012, 31 countries and the EU signed the Anti-Counterfeiting Trade Agreement with the aim of establishing stricter standards for the transnational enforcement of intellectual property rights. The agreement received support from many sectoral business associations, but, in the negotiation phase, it excluded a collation of citizen groups. Consequently, the opposition to the agreement really took off with the signing of the agreement, including street protests and an online petition (Dür and Mateo, 2014: 1202). This post-decision mobilisation brought the issue of Internet freedom to greater public attention so that, in front of strong public opposition, a series of member states stopped the process of ratification in February 2012.

The idea that the absence of adequately inclusive channels to convey otherwise neglected interests may undermine the legitimacy of EU decisions does not necessarily convey an overall-pessimistic vision of the democratisation process in the EU. As Trenz and Eder argue, protests and mobilisation against the architecture of governance at the supranational level prove the self-constituting dynamic of a European public sphere. On this view, the diffusion of a democratic-deficit discourse contributes to the expansion of transnational spaces of political communications and the democratisation of the institutional system (Trenz and Eder, 2004: 7). Likewise, when backlashes and violations of normative principles occur, protests and transnational communication intensify with the result of facilitating and speeding the democratisation of the European system (Trenz and Eder, 2004: 20). So exclusion of civil-society groups which try to bring otherwise neglected interests into the deliberative process may not come at a very high cost to the process of democratisation of the EU. After all, as a wave of scholars have been arguing for a long time, civil-society groups convey their interests at the civil-society level, thereby contributing to the self-constitution of an engaged public sphere (Della Porta and Caiani, 2007). Nevertheless, I believe that the exclusion of otherwise neglected interests from talks may come at a high price in terms of the legitimacy of EU decisions in the short run, at least. By moving outside institutionalised deliberative channels, civil-society groups may mobilise the otherwise neglected interests explicitly against supranational policy solutions. The TTIP (Transatlantic Trade Investment Partnership) is a case in point. It concerns foreign trade, a policy area that falls under the competence of the EU, but, since negotiations are kept secret and within the European Council, formal political opportunities for civil-society actors have been limited (Caiani and Graziano, 2018). However, over the years, civil-society groups opposing the TTIP have become increasingly vocal through a number of initiatives at the national, supranational and global levels (Caiani and Graziano, 2018). Now the TTIP has come to a dead end, mostly because of the oppositional politics run by the Stop TTIP organisation outside formal deliberative EU procedures (Caiani and Graziano, 2018: 19–20).

As my examples illustrate, civil-society groups may use the exclusion of otherwise neglected interests from institutionalised deliberative processes strategically. They can challenge decisions ex post by displaying neglected interests and the evidence at their disposal. If it was used productively by supranational institutions, this dissent would lead to revisions, a rethink of collectively binding decisions and perhaps re-legitimatisation of the whole process. Without adequate deliberative channels, post-decision display of otherwise neglected interests may encourage resentments, suspicion and scepticism about institutions-relative-actions and the standards of evaluation.

From the point of view of EU institutions, at the empirical level, it is important to identify possible barriers that prevent civil-society groups from advancing their interests in public deliberation before and after voting. At the normative level, the present analysis suggests that, at the supranational level, an understanding of institutionalised deliberative processes that begin prior to voting and continue after through multiple formal spaces may increase the legitimacy of supranational decisions.

From the point of view of civil-society groups, at the empirical level, it is important to identify particularistic biases and systematically neglected interests in supranational decision-making processes. This would allow strategic positioning and perhaps entry into institutionalised processes. At the normative level, the present analysis suggests that inclusion prior to voting is not sufficient to consider a decision as fully legitimate. If the idea is strengthening the legitimacy of supranational decisions, inclusion of otherwise neglected interests in post-decision deliberation is benign in overcoming pathologies such as Captain Hook politics and garbage-time politics, but, considering the other barriers to participation (e.g. Friedrich, 2011; Kohler-Koch and Quittkat, 2013; Quittkat, 2011; Steffek and Nanz, 2008), such an enterprise is not necessarily an easy one.

Viewed through these lenses, if they find a more legitimate EU to be a desirable outcome, civil-society groups should be vocal in their demand for inclusion and in asking for deliberative processes prior to and after voting so that a large number of otherwise neglected interests can contribute to the entire policy life cycle. At the same time, this analysis also supports the belief that the exclusion of civil-society groups from institutionalised processes of decision-making may be the germ of a more intense democratisation. Excluded civil-society groups may convey their interests outside institutions, take steps to challenge single decisions on epistemic terms and defend their interests with more flexibility and using a larger array of unorthodox rhetoric and practical instruments, thereby contributing actively to creating conditions that allow for disagreement on specific policies without reversing the commitment to democratic values.

Conclusion

In this article, by looking at decision-making processes in the EU, I have argued that the experimentalist model of democracy can contribute to contemporary disputes about the inclusion of civil-society groups in deliberative democracies. At the explanatory level, it is possible to evaluate the legitimacy of EU decisions by looking at three phases: talk, vote and feedback. From this perspective, it is possible to identify and explain obstacles to democratic decision-making. At the normative level, the fundamental idea is that, in conditions of pluralism and disagreement, for a decision to be legitimate, deliberative decision-making processes must be structured so as to allow the inclusion of affected interests before and after voting.

I have argued that there are three ways for a decision to be illegitimate: exclusion of affected interests from all deliberative phases, Captain Hook politics and garbage-time politics. Much has already been written about wholesale exclusion of civil-society organisations from policy-making processes in the EU. Therefore, my analysis has focused only on Captain Hook politics and garbage-time politics: in these cases, some otherwise neglected interests are included, but they are included with bad timing. The intersection of these two deliberative pathologies shows that the absence of post-decision institutionalised deliberative channels may affect the legitimacy of collectively binding decisions.

While bad for the legitimacy of institutional decisions, Captain Hook politics and garbage-time politics may be functional for the purposes of civil-society groups and, more generally, for the process of democratisation. ⁷ Civil-society groups may direct otherwise neglected interests towards the civil-society level or explicitly against institutions-relative-actions. This move may be a cause of disenfranchisement, suspicion and mistrust, which is bad for legitimacy. At the same time, I have argued that, by bringing otherwise neglected interests outside institutionalised processes, civil-society groups may contribute to strengthening democratic discourses.

In this article, I have used EU politics as synecdoche for real-world deliberative decision-making processes with an aggregative component. However, my perspective is relevant for other settings, such as local policy-making (Anderson, 2008) and national political systems (Anderson, 2003). In all contexts, an experimentalist model of democracy would offer a way to see different interests not simply as sources of bias but as resources for constructing more legitimate policies through inclusive deliberation. Against this backdrop, my argument for inclusion all over the policy cycle would help us reconsider the many and diverse roles that interests can play at different stages of a democratic decision-making process.