The Roles of Referendums in Deliberative Systems

ABSTRACT

This paper does not ask to what degree referendums match deliberative democratic criteria, because no institution can. Instead, it asks what roles referendums can play in a deliberative system. It argues that referendums can play constructive roles not just as legitimating tools, but also by providing focusing events that generate salience for everyday narratives of collective life. It also argues that small-scale deliberative processes work better when they are focused on such collective decision making moments.

KEYWORDS:

• Deliberative democracy
• deliberative systems
• referendums

Introduction

When deliberative democrats have evaluated referendums, they have tended to do so by applying a once-standard set of deliberative democratic criteria and looking for evidence that they are met in a particular institution, or a case, or both (eg, Parkinson, 2001). They have almost invariably failed to find much evidence. Referendums seem to embody everything that deliberative democrats set their faces against: agenda manipulation; brute majoritarianism; symbolic battles instead of reasoned dialogue; yes-no, take-it-or-leave-it questions instead of creative co-production … the list could go on. Some scholars have recommend creating new, parallel institutions to inject more deliberation into the process (eg. Gastil, Richards, & Knobloch, 2014; Uhr, 2000), but the need for such institutions only arises because referendums alone cannot do that job.

However, changes in deliberative theory over the last ten years throw new light on such issues. In the wake of the ‘deliberative systems’ turn (Dryzek, 2010; Parkinson & Mansbridge, 2012), theorists recognise that no institution can meet the classic deliberative democratic criteria at once, because they are qualities of a democratic system as a whole and not to be expected to co-occur in every institution or every social practice that makes up that system (Bächtiger & Parkinson, 2019; Chambers, 2009; Mansbridge et al., 2012).¹ Just as no part of a symphony is itself a symphony, no part of a deliberative democracy is itself deliberative democracy. Therefore, from a systems perspective, a lot of empirical deliberative scholarship has been applying the theoretical criteria at the wrong level of analysis, looking for symphonies in the parts of the music instead of standing back and looking to see how the parts contribute to the qualities of the whole. It might be that the deliberative (adjectival) quality of a democracy is produced by all sorts of institutional, discursive, and performative interactions between a variety of differently-situated agents, given different constraints, and is not simply the result of deliberation – the noun – in the right kinds of institutions (Chambers, 2018).

There is, of course, some scholarship that has looked at referendums in context, sometimes through a more systemic lens, particularly work on constitutional referendums by Levy, Kong, Orr, and King (2018), Suiter and Reuchamps (2016) and Tierney (2012); as well as new systemic perspectives on the reform of referendum processes, such as Cheneval and el-Wakil (2018). Increasingly, authors are insisting on the importance of context for any constellation of institutional design features (eg. McKay, 2018). However, while new work looks at linked or sequenced institutions, sometimes (but not always) in their social and political context, they still tend to focus on deliberation, the noun (eg., LeDuc, 2015; Levy et al., 2018), instead of the dispersed deliberative quality that a more thorough-going systemic perspective would suggest (Chambers, 2018; Landemore, 2018).

This paper takes that dispersed account seriously by examining referendums at a different level of analysis, using recent conceptual developments. It asks not how referendums match the classic deliberative criteria, but rather, ‘Can referendums, warts and all, contribute to the overall deliberative quality of a democratic system?’

The paper begins by setting out an account of deliberative systems that views them in dynamic and not merely formal terms, with three broad criteria concerning sensitivity to claims on public resources and attention; connecting claims to reasons – which include narratives of collective life – and weighing them against each other in a visible, recursive way; and making sure that agreements are decisive, consequential. Second, the paper discusses salient variations in referendum processes; and third, assesses referendums’ roles in deliberative systems in terms of the three criteria. The argument is that referendums can play constructive roles at both the start and the end of a mass deliberative process, connecting political actors with everyday political talk. It also argues that, given the right intervening institutions, referendum processes can connect citizens with communicative resources that they can deploy in their own deliberations and decision-making. Indeed, it argues that there are cases in which the intervening deliberative institutions work better in the shadow of a referendum vote. In other words, there is a need for additional institutions; but not for every function, and for different reasons than those offered to date.

This is not an overall evaluation of referendums – a purely deliberative framework would be inadequate for that task (Warren, 2017). Nor does the paper spend much time on pathologies of referendums – that is a topic on which much ink has been spilled. Instead, it focuses on using a systemic perspective to reveal some novel arguments in favour of the role of referendums in deliberative systems.

Deliberative Systems

‘Deliberative systems’ describes a family of approaches to deliberative democracy that have developed since the mid-2000s – the phrase was coined by Mansbridge (1999), and its two major theoretical variants have been set out by Dryzek (2010) and Parkinson and Mansbridge (2012). While deliberation is commonly defined as ‘justification through public argument and reasoning among …  citizens’ (Cohen, 1989, p. 21), or as ‘changing …  judgements, preferences and views [through] …  persuasion rather than coercion, manipulation, or deception’ (Dryzek, 2000, p. 1), deliberative systems theorists argue that these are distributed features of democratic systems and need not – indeed, cannot – entail all the relevant parties sitting in a room together ‘deliberating’.

In Bächtiger and Parkinson’s (2019) recent elaboration of the approach, a deliberative democracy involves agents, practices and institutions that (1) set the agenda by systematically listening to the full range of the public’s claims on public attention and resources; (2) connect those claims with reasons and weigh them against competing ones, in visible (Rummens, 2012) and recursive (Mansbridge, 2017) ways – that is, with systematic reflection back to claim makers and citizens more generally, in terms they themselves use; and (3) reach binding decisions or are directly consequential in ways that are subject to formal legitimation. These requirements can be described as ‘plugging in deliberation’. To be democratic, a deliberative system needs to be both plugged into the experiences, narratives, deliberations, claims, even the symbols and language of the relevant demoi at one end; and plugged into the ‘power socket’ at the other, by being decisive in some way.

That last feature continues to be a controversial one in deliberative theory. On the one hand there are those like myself (Parkinson, 2012b) and Thompson (2008) who argue that a deliberative system must have direct political and policy consequences, backed by resources and enforcement, if it is to merit the label ‘democratic’. There are others, notably Goodin and Dryzek (2006) and Niemeyer (2014), who argue that consequentiality takes a great many forms, not all of which can be reduced to a single decision moment with direct impacts. Somewhere in between is Chambers (2012) who says that deliberation must at least be ‘decision-oriented’. But there is less disagreement than meets the eye, because these claims operate, once again, at different levels of analysis. It is not inconsistent to say that a deliberative system must be directly decisive while at the same time arguing that some individual moments within it need not be.

The precise role of deliberation – the noun – in a deliberative system is not settled either. Some argue that a system gets its deliberative quality by means of institutions that create the special conditions for near-ideal inter-personal deliberation, the institution injecting, as it were, deliberation into the system (Owen & Smith, 2015). I describe this as the ‘additive’ view of deliberative quality. The alternative is the ‘summative’ view, which sees deliberative quality as emerging from the interaction of citizens in multiple settings and in multiple ways. These views are not mutually exclusive: indeed, deliberation between citizens is a necessary condition to ensure that the system responds to people as citizens and not simply as individual consumers (Bächtiger & Parkinson, 2019, p. 109). It is just not a sufficient condition to give a political system a deliberative democratic quality.

This helps address another point of debate in the literature about what kinds of communication are valuable in deliberative democracy. Most authors include in their definition of deliberation a variety of communicative modes such as protest, public rhetorical performance, rituals of greeting and respect, and so forth (Bächtiger, Niemeyer, Neblo, Steenbergen, & Steiner, 2010; Ercan, 2017; Sass & Dryzek, 2014), in response to criticism from difference democrats and others that the deliberative theory privileged the communicative practices of the already-powerful. From a systemic perspective, this expansive response may have been unnecessary: recognising the democratic value of getting angry and shouting in the face of injustice neither requires us to redefine deliberation to include getting angry and shouting, nor diminishes the value of deliberative modes at appropriate times and places.

The final point to make is that there are at least two rather different ways of conceptualising deliberative systems, two ‘generative metaphors’ (Schön, 1978): the metaphor of the network, featuring sites of deliberation and decision making, and communicative linkages between them; and the metaphor of stages or sequences. The first focuses on usually-formal relationships between institutions but overlooks dynamics, agency, and the fact that communicative linkages are themselves ‘sites’ with their own incentives and influences. Such views also lack an account of a ‘decision’ moment. The second corrects the dynamics oversight and includes some ideas of agency – the sequential view is all about actor-centred change over time through sites with their own logics, cues and incentives – but tends not to capture wider contextual factors, or the chaotic nature of the ideational ‘soup’ in which ideas and actors swim (John, 2003). Indeed, it fails to capture the thought that many institutions are not formally constituted, but exist as informal yet patterned scripts of understanding and action such as discourses (Dryzek, 1996), a theme Chambers (2018) and Landemore (2018) develop when arguing that deliberative scholarship itself needs to be less exclusively focused on formal institutional design. There are many variations on these broad themes, and scholars urging colleagues to go even further in applying more dynamic, interpretivist, models of political change (Boswell & Corbett, 2017) – or just less normatively-blinkered ones (Spada & Ryan, 2017). The point for now is to note that the tendency, alluded to in the introduction, to think of deliberative systems as sets of linked yet static and somewhat formal institutions is reasonably common, but misses out much that matters on more dynamic, communicative accounts like that set out in Bächtiger and Parkinson (2019).

To sum up, a deliberative democracy may well depend on some people sitting and talking carefully amongst each other, connecting and weighing arguments in small-scale, citizen-focused or representative forums; but it also depends on going out and sharing, building, listening, engaging, cajoling, performing, representing, jockeying for positions, accusing, scrutinising, creating, seizing moments and muddling along, to build coalitions of people who eventually say ‘yes’ or ‘no’ in the face of a powerful or well-institutionalised status quo. The ability of agents to do that will be constrained by institutional and structural factors, particularly the degree to which governance is susceptible to, or insulated from, such pressures – that is, real world agents and institutions sit in a ‘participatory ecology’ (Zaremberg & Welp, 2019). These systems will be deliberative, in a summative sense, to the extent that they connect claims with reasons, listen, and reflect in a visible way; they will be democratic to the extent that the grounded perspectives of those affected are included and dignified, and those people either make or endorse decisions. But those are system-level judgements, not simply ones to be made at the level of individuals, venues, or institutions considered in isolation.

The general implication of these points for the study of referendums is that they need not themselves match the classic criteria for deliberation in order, nonetheless, to have value in advancing deliberative and democratic values in the system as a whole. Before going into the details of how that works, however, we need some definition of ‘referendum’.

Varieties of Referendum

Butler and Ranney (1994, p. 1) define referendums as mass public votes on an issue, as opposed to mass votes to choose representatives. This definition covers a variety of practices with a variety of labels that range from citizen-initiated votes (called ‘initiatives’ in the U.S. and Switzerland but ‘citizen-initiated referendums’ in New Zealand, for example) to government-initiated ones, or even votes that are automatically triggered by constitutional decree. That might seem to be unhelpfully general, given that referendums are not a single institution but come in many varieties, with different institutional pressures and incentives (Chambers, 2018; Cheneval & el-Wakil, 2018). Nonetheless, for the purposes of this paper, the key differences relate to the threefold functions of a deliberative democratic system, particularly the first and third: that is, the ‘listening’ or agenda-setting functions; and the decision functions. The second, ‘connective’ and recursive function, becomes more important in the next section.

On the question of agenda-setting power, it is common to distinguish between referendums that are formally government-initiated or citizen-initiated. While the normative literature focuses almost entirely on the legitimacy virtues of the latter – although see Bellamy (2018) – the former is by far the most widely-used, most often to decide constitutional or territorial issues (Ruth, Welp, & Whitehead, 2017). Sometimes governments refer to the people questions on which the legislature cannot reach agreement, or as an internal power play, or as a responsibility-shifting tactic on controversial issues. For example, it has been argued that former British Prime Minister David Cameron’s promise to hold the Brexit referendum was in part an exercise in government-initiated responsibility shifting, as tried to control dissent within his own party, long riven by disputes over EU membership, as well as fear of losing votes to the UK Independence Party (Copsey & Haughton, 2014).

However, it is increasingly recognised that non-governmental referendums are almost never initiated by ‘ordinary citizens’ (Smith, 2009, p. 117). This is largely a matter of the resources necessary to qualify a proposal for a ballot, let alone win an argument, given time limits and signature thresholds. For example, in California proponents have 150 days to gather signatures from five per cent of those who voted in the last gubernatorial election. At the time of writing, that was 1,444 signatures a day, which has necessitated a well-organised signature-gathering industry that pays voters to sign petitions. That is in addition to the necessary teams of lawyers, volunteers, political marketing consultants, lobbyists and networkers who can only be afforded and mobilised by those who already have extensive, effective, and well-funded networks. As a result, initiatives are launched or quickly taken over by big lobby groups, second-tier political parties, even factions within mainstream parties. Even Switzerland, with its relatively low signature hurdle sees its direct democratic politics dominated by established networks because of the organisational demands of gathering even a comparatively modest number of signatures: 18 months to gather 100,000 signatures for an initiative proposal, for instance (Kriesi & Trechsel, 2008; Parkinson, 2001).

Social media have changed this dynamic less than one might think. While some systems, like the UK, establish formal, funded, authorised campaign groups, a great deal of campaigning goes on outside those formal organisations. For example, the Scottish independence referendum of 2014 is sometimes held up as a remarkable example of grass-roots mobilisation in action, with organisations like Women for Independence and local Yes campaign groups playing a crucial role in organising online, disseminating information and deliberating in both online and face-to-face ways, quite independent of the symbolic politics conducted by the official campaign groups. However, none of the grassroots groups started ex nihilo. They all drew on existing networks of climate, social justice and political reform campaigners. As a result, we need to be cautious of claims that referendums of any sort provide a direct agenda-setting route where there has been a failure of (formal) representation (eg. Cheneval & el-Wakil, 2018, p. 299): referendum committees and unofficial campaigning groups too are representatives who mediate public discussion, and are thus also unlikely to capture the full ‘pool of perspectives’ (Bohman, 2012) without institutionalised assistance.

Turning to the third function, there is much room for variation in what a referendum decision entails, but in no case is this an uncomplicated matter. There is a standard distinction in the literature and public discourse between binding and non-binding referendums, at least outside the United States and Switzerland where referendums are held on fully-drafted laws or constitutional amendments. In a great many cases – and not just in Westminster-based systems featuring parliamentary sovereignty – referendum questions are broader statements of intent that require both enabling legislation to hold, and a parliamentary or even inter-governmental processes to write, debate and pass the actual law changes or treaties afterwards, allowing much more room for interpretation between conflicting ideas of a measure’s broad intent and the implementational detail. But even in the ‘fully drafted’ systems, there are cases of multiple referendums on the same topic, with different groups promoting different drafts or counter-proposals, the winner being the one that wins a plurality of votes. McKay (2018) commends such features, but they can easily mean that a majority of voters were opposed to the eventual winner (Bellamy, 2018), while in many jurisdictions courts can, and do, overturn results on constitutional grounds.

Those institutional setting issues only tell part of the story, however, because referendum proposals do not have to win or be binding on governments to nonetheless influence policy and politics. This is clearly the case in Switzerland: while its consensual institutional structure means that initiatives lose much more often than win, the weak position of the executive relative to parties, cantons and formally-recognised interest groups means that government action to head off initiatives often comes at the price of negotiated law changes anyway (Kriesi & Trechsel, 2008). For example, various groups in Switzerland have launched more than 30 anti-immigrant petitions, a dozen of which have forced a vote, but until 2009 not one secured the necessary double-majority in the Swiss system (a majority of individual voters in a majority of cantons). Nonetheless, winning the vote is not necessarily the point in the Swiss system: setting the terms of deliberation is. And that is one reason why one needs to take a longer view than just focusing on the Swiss People’s Party (SVP) wins in 2009 and 2014. They launched anti-immigration initiatives repeatedly from 1992, most of which ‘failed’, but the failures shifted public opinion and forced law changes, sometimes significant ones, while the SVP grew from minor status to become the largest party in both houses of the Swiss Federal Assembly on the back of these campaigns. The same is the case with the Brexit vote: while many commentators stressed the non-binding nature of the vote after the fact (eg., Grayling, 2016, and responses), that bore little weight in the face of the discursive power of the ‘express will of the people’, pushed by single-issue campaigners who changed the discourse.

All this goes to the issue of context, which should stand as a constant caveat over abstract analyses such as this one. We can see this in a relatively crude fashion just by comparing how referendums work in consensual systems versus majoritarian ones, to use Lijphart’s (1999) broad distinction between types of democracies. In consensual Switzerland, the winner’s share of initiative votes is greater than 70 per cent in more than a third of cases; 70–75 per cent is the most common winning total. In majoritarian California, a quarter of all decisions are sharply divided, with the winners getting less than 55 per cent of the vote; half win less than 60 per cent, three quarters less than 65 per cent (Parkinson, 2001). In other words, the macro political and communicative system in which referendums are embedded exerts an independent influence on the argumentative dynamics of a referendum, not just the particular configuration of the device itself. This further underlines the importance of making system-level judgements about referendums, to which the paper now turns.

Referendums in a Deliberative System

This section considers each of the three criteria set out above for a system to be deliberative and democratic, applying a dynamic, sequential view rather than a networked account. The analysis starts with the origins of narratives and claims on agendas, considering the degree to which a referendum process can help or hinder the capturing of the variety of the pool of perspectives, and why and how that matters; then the connective processes that link narratives and claims with reasons followed by the weighing of competing proposals; and then plugging into the power socket, the legitimation processes and the degree to which they result in agreements that can be, and are, implemented.

Plugging into Inclusive Public Narratives

One of the key requirements in any democracy is that institutions be sensitive not merely to the formal claims that are made against the state in the public sphere, but also to the narratives of common life that underpin those claims. This is the ‘everyday talk’ of deliberative systems (Mansbridge, 1999), the stories people tell each other in bars and at bus-stops, on social media or talk-back radio. They generally concern personal experiences of collective life: what happened to grandma when she broke her hip and had to go to hospital; experiences of bullying at school or the workplace; what it feels like to deal with a fragmented, part-privatised welfare state in which just getting by requires us to learn different, ever-changing processes from different, ever-changing suppliers. In standard, ‘responsiveness’ accounts of electoral democracy this is one of the roles of elected representatives: they are the conduit between their constituents and formal decision-making power. There are other routes too: interest groups, new social movements, and social media ‘curators’ (Bennett & Segerberg, 2012); investigative journalists, ombudsmen and administrative watchdogs; and also the vast apparatus of ‘user’ consultation constructed by public bureaucracies around the world, especially from the 1980s as the relationship between state and citizen was reconstructed.

Citizen-initiated referendums are frequently promoted as another effective route, although generally not so much for the narratives that matter to the relevant demos, but for their formed claims – that is, their assertions about what the state should or should not do in response to felt conditions. They are explicitly advocated and defended on the grounds that they allow citizens to propose changes to the law, constitutions or policy directly to their fellow citizens, and have those fellows make a majority decision, without having been filtered out by party politics, bureaucracies, or interest-based lobbying – although it must be said that a lot of the literature is somewhat polemical, affected by the same fervour that affects other reformist strands of democracy studies (Floridia, 2017). On this view, government-initiated referendums are, by definition, less desirable: they provide opportunities only for ‘citizens vote’ participation, and not ‘citizens propose’.

I have already noted that citizens’ agenda-making power in referendums is greatly overstated. But if we take a deliberative systems point of view, the distinction between everyday narratives and public claims is important, both in a positive and negative fashion. On the positive side, it is not the case in a real deliberative sequence that narratives always precede claims, for all participants, over the entire time period of the referendum campaign. We know that specific policy solutions precede problems, that they lead independent ideational lives, and that the job of political entrepreneurs is to successfully connect ‘their’ solution to some felt condition out there in the world (Kingdon, 1984); that is, to connect claims with the everyday narratives that ground them. So, it may not matter so much where a given claim originates – whether it is really citizens’ initiative, or that of some sectional interest, or the government itself – so long as the claim can be redeemed in discourse that grounds it in commonly felt conditions, as well as facts, interior states and social norms. In short, a referendum of whatever kind can act as a spur for actors throughout the political system to pay attention to everyday political talk, and attempt to make a case that their claims are validated in terms of that talk.

This helps make sense of what appears to be a straightforward deliberative failure: the success of the Leave.EU campaign in the UK in the Brexit referendum of 2016. A traditional deliberative analysis would focus on the Leave campaign’s breaches of sincerity and epistemic standards as it made claims that were simply not redeemable by reference to facts. The systemic analysis, by contrast, suggests also looking at the degree to which Leave’s claims were redeemed by reference to widely-shared narratives about lived experience, and sure enough, we have empirical evidence that this was the case. Shaw, Smith, and Scully (2017) show that the Remain campaign (a) spent most of its time responding to themes determined by Leave and thus allowed Leave to set the agenda; and (b) spoke in the language of analysts and experts. Leave demonstrated active listening, one of the things that Bächtiger and Parkinson (2019) claim gives democracy a deliberative character. Leave often showed that they were hearing (some of the) people by repeating claims back to them in language they themselves used or valued.

Of course, whether those claims were true was a separate question. A deliberative democracy also requires that claims be connected with facts, evidence, logic and values, and not just everyday narratives. I will discuss that more in the next sub-section, but the point for the moment is that deliberative democratic systems do not just depend on the ‘better argument’ judged against purely epistemic standards, but also against the standards of claims grounded in narratives of everyday life. Not only that, but referendums make the latter particularly salient – they are won and lost on the ability to make such connections because it is the people who live and breathe those narratives who make the decisions, even if they do not determine the agenda. Focusing on ‘the facts’ or lamenting our post-truth world is only half the story.

Still, if this is an endorsement of a role for referendums in deliberative democracies then it is a pretty half-hearted one. First, one can concede that it may not matter where a specific claim originates, as above, but also argue that it matters very much indeed where the generality of claims originates. If the only issues that are debated are those promoted by well-resourced, powerful interests, then we are not listening to the full pool of perspectives, one of the criteria for a democratic deliberative system. But second, the requirement that claims be grounded in the people’s narratives rests on a big empirical ‘if’. In practical politics narratives of collective life are created by those with an interest, a point about the manufacture of consent advanced by writers as different as Lippmann (1922) and Gramsci (Simon, 2002). This implies a requirement that the narratives of collective life be normatively prior to the claims made on the basis of them. Now, such a requirement presents both a philosophical and empirical challenge. The philosophical problem is that it appears to rely on some idea of authenticity of experience, ‘the thing itself’, that is somehow prior to its expression in either narratives or claims. For reasons of space I cannot argue that out, but there are answers to that problem found in, for example, Benhabib (2002) and Fricker (2007), who argue for a norm that subordinate peoples ought to be the ones to come to views about what is authentically theirs and what is an imposition from outside – a presumption of the rightful source of narratives and claims, not a presumption of epistemic truth – recognising that peoples are not homogenous, that identities and practices are contested, shape-shifting with context. But the empirical difficulty remains: we need some means by which the marginalised are given effective opportunities to make, endorse or challenge narratives and claims, and the hot political environments of referendum campaigns are not generally ones in which that is likely to happen unaided. Other institutions are needed for that function, to which we now turn.

Linking Claims with Reasons

It has been shown many times that, because referendum campaigns rely heavily on information production and dissemination through traditional news media, increasingly fragmented online and socially-generated sources, as well as directly from campaigners to citizens; and because of the cultural, technical, and political-economic features of those media (Maia, 2009; Street, 2001); it is highly unlikely that all the relevant claims and counter-claims will get heard, let alone tested, in the course of a mass public debate. However, it is worth turning to another case for a moment, because it illustrates two points. First of all, in some cases it is the everyday talk that provides the relevant epistemic basis, with academic and other research following behind, strengthening the claims based on those narratives (cf. Chambers, 2012). And second, there are ways of enhancing the deliberative qualities of referendums with other institutions and innovations.

The case is a series of processes that have led to two major referendums in Ireland: the first in 2015 which legalised same-sex marriage, the second in 2018 which removed a constitutional ban on abortion. Each was preceded by sizeable citizens’ forums – the 2012–14 Convention on the Constitution and the 2016–17 Citizens’ Assembly – whose tasks were to make recommendations on constitution changes, law changes and even the standing orders of the Oireachtas (parliament). I will not go into all the details here – the former is well described by Suiter, Farrell, and Harris (2016), the website of the latter is https://www.citizensassembly.ie/en/ (accessed 12 October 2018) – but it is worth noting the following:

1. both featured randomly-selected citizens, the latter entirely, the former with 66 lay members out of 99 in total;²

2. both had relatively broad remits, especially the Convention whose task was nothing less than redesigning formal Irish political practice;

3. both made a large number of recommendations across a range of constitutional and policy issues, relatively few of which were accepted by the Oireachtas – some were outright rejected, others were kicked into the long grass of internal reviews and procedures.

The first significant point, however, is that the Convention in particular was the result of a long build-up of frustration with and anger at Ireland’s political elites, particularly the Oireachtas and the Catholic Church. The former was blamed for the series of economic disasters that struck Ireland in the wake of the Global Financial Crisis of 2008, but it went beyond that particular frustration and spread to a broad sense of distrust of and distaste for a ruling elite that was seen to be out of touch and indeed contemptuous of the Irish people. The Church was bundled into this not only because of the long, constitutional relationship between church and state in Ireland, but also because of the Church’s own scandals and disasters, most notably a string of child abuse scandals and cover-ups, as well as the fact that Ireland remained one of the few developed countries that forbad divorce and contraception (Barr & Corráin, 2017). The anger boiled over at the 2011 election which reduced the dominant party in Ireland, Fianna Fáil, from 41 per cent of the seats in parliament to a mere 17 per cent. So, the Convention and its successor, the Assembly, took place against a backdrop of long-standing communicative action that undermined support for and confidence in the two dominant institutions in Irish society. The ‘facts’ were the stories of sexual predators in the priesthood and elite attempts to cover that up; and the economic predators in banking and finance, property and politics; and narratives that showed it was the rest of the people who were carrying the can.

The second point is that the Convention and Assembly provided institutional focal points for: (a) clarifying and shifting the agenda of what were extraordinarily broad yet elite-written remits; (b) connecting claims with narratives and evidence, using the people’s own language, and thus demonstrating reflective listening; and (c) doing so in public, generating reasonably extensive media coverage for what can seem arcane processes. The fact that claims were aired and validated in public forums like this may have become crucial later on in the process, because – and this is based only on anecdotal evidence from a few people involved – it armed other citizens with examples, stories, arguments and counter-arguments that could be deployed in the referendum campaigns (Suiter et al., 2016). That is, they provided citizens with a repertoire of pre-tested, clarified, validated claims that were clearly anchored in everyday talk. This is an important variation on the ‘recommending force’ claim of Fishkin (1997) and MacKenzie and Warren (2012). This is not a claim about the drawing of conclusions based on a trust decision or a judgement about descriptive representativeness; it is a claim about making clarified, tested arguments available, through public performance, to a wider audience who then go on to use those claims in their own everyday deliberations, and eventual decision making. It is about making rationalised inputs available for citizens’ judgement and decision, not a delegation of judgement, and not just the open slather of public discourse dominated by the powerful discussed in the previous sub-section.

Compare the Irish approach with the Scottish independence referendum debate. In 2007, the nationalist Scottish government launched a formal consultation process called A National Conversation (Scottish Government, 2007), and once a question had been agreed with the UK government, formal Yes and No campaigns were created, in line with requirements of the Political Parties, Elections and Referendums Act 2000. However, those processes created something of a partisan, mediatised, ritualised combat that generated a great deal of heat and not a lot of light. Around 18 months out from the vote, pressure mounted on the Scottish government to do something about the resulting information vacuum, but the response was largely to stand back and allow think tanks and democracy activists to fill that vacuum. The activists organised workshops and conferences, engaged with academics and journalists, mobilised networks through social media (some drawing on networks from prior campaigns on social justice issues), and created new networks of people in towns, cities and villages across the country, often focused not so much on independence per se, but on what kind of country they wanted Scotland to be, on issues of welfare and justice (distributive as well as retributive), as well as procedure, identity and symbolism. The result was an astonishing outburst of conversation at bus stops and kitchen tables, in market squares and online forums, in pubs and workplaces, on the details of policy and the possibilities of an independent Scotland.

However, in interviews, some activists conceded that they never managed to engage more than about 40 per cent of the voting population of Scotland, having started in 2007 from a position in which support for independence was around 24 per cent (Scottish Social Attitudes Surveys, available from http://www.scotcen.org.uk/, accessed 12 October 2018). There are many reasons for this, but my conjecture is that the bottom-up approach, while generating a lot of everyday talk, failed to provide a neutral, institutional focal point in which those narratives could be connected to reasons and the resulting clarified arguments ‘dignified’ by their setting (Parkinson, 2012a), and thus made available in a way that made them easy to be integrated into the communicative repertoires of the wider public.

Decisiveness

But nor could the Irish ‘minipublics’ have been successful – broadly rational, widely engaged, generating legitimate collective agreements – on their own. Randomly-selected citizen assemblies have a great many virtues, but lack some key ingredients when it comes to binding collective decision making, particularly a lack of bonds of authorisation and accountability between the relatively small number of deliberating insiders and the vast majority of citizens looking on (Parkinson, 2006). What referendums add, that no other device can, is an act of mass public consent, which then makes legitimate the exercise of power to enact that decision. But it is also the fact that because referendums make decisions, they act as focal points, the anticipatory effects of which greatly strengthen the salience of any preceding minipublic. Scots would not have debated the nature of Scottish society in market squares and online forums without the fact of a looming referendum. Consider, by contrast, the case of Iceland’s constitutional reform efforts of 2009–13. This case is frequently misdescribed as ‘crowdsourcing’ the constitution, but that is not accurate, despite the fact that lay citizens were on the Council that was charged with drafting the proposed document and the highly participatory, online nature of some of the contributions:

Despite open access and the existence of a robust secretariat staffed with many experts to assist the Council, the Council was not able to plough through all the extensive public input systematically, as it only had four months to complete its task. Some Council members never looked at any of it; it was entirely up to each member to navigate through the information and decide what to take in or consider. Furthermore, members never felt obligated to adopt inputs they did not agree with. (Bergmann, 2016, p. 24)

Not only was the process far less participatory than commonly presented, Bergmann shows that it was not effectively ‘plugged into the power socket’ either. There was relatively little salience for constitutional reform as opposed to economic reform and criminal prosecution of those responsible for the banking crisis of 2008; the forums that were set up to address the constitutional issues suffered lower and lower turnouts and loss of support; the Council itself gathered powerful enemies as its legitimacy drained away; the outputs were merely recommendations; and with a change of government in 2013 the entire process was shelved. In short, the anticipatory effect of an actual, mass decision making moment was not present in the Icelandic case, which drained the energy out of the participatory, deliberative forums.

Conclusions

If we judge referendums against the classic criteria for deliberation, we usually find that there is little deliberative about them, but no institution can match all the criteria. Instead, this paper has asked a more systemic question: what roles can referendums play in deliberative democracies? Given a particular account of deliberative systems, I have argued that they can play two critical roles, one at each end of a deliberative sequence. The first is somewhat unexpected but highly qualified: referendums make salient not just proposals for action that emerge from minority-elite politics, but also the broad pool of narratives about collective life – indeed, they make the narratives particularly salient. In an individual referendum campaign it may be less important whether the proposals or narratives came first, so long as the proposals can be redeemed by reference to the narratives, although that is subject to an authenticity test that may be hard to meet in the real world of communicative power relations. It certainly matters in the aggregate if the proposals of certain sources are consistently privileged, and if wider narratives on particular issues are largely manufactured by those same interests.

The second role is that referendums provide an unmatched ability for the mass public to endorse or reject proposals or agreements reached elsewhere by some smaller group of representatives, elected or selected. I have suggested, along with Landemore and others, that ‘agreements reached elsewhere’ might be the agreements of a randomly-selected institution that lacks decision authority on its own, but possesses the deliberative virtues of reflective listening, weighing alternatives, and co-creation of paths forward. It might also be that referendums are needed at constitution-making moments, endorsing new rules of the democratic game rather than each move within it.

But the relationship between small, deliberative bodies and mass referendums runs in the other direction too. The small bodies – minipublics – act not so much as ‘trusted information proxies’ with ‘recommending force’ (MacKenzie & Warren, 2012), but rather as stages that allow clarified claims, with their supporting narratives, to be made more widely available to the mass public, who use such scripts in their own everyday deliberations before voting. The minipublics work this way because the referendum exists – it does not work (or, rather, does not work as well) without the focal, mass decision-making moment of a referendum vote.

What referendums cannot do on their own is produce deliberation, the noun. They can provide some of the pre-conditions for deliberation, but they cannot, by themselves, effectively connect claims with reasons, weigh alternatives, or produce creative pathways out of disagreement. We need other institutions to do that work. In some ways this is obvious: they are agenda-setting and preference aggregation tools, not creative problem-solving ones, and so it is unreasonable to expect them to feature all – or even very many – of the longer list of desirable features of a deliberative democracy.

To play effective roles in large-scale, deliberative democracies, referendums need to be the end point of longer, multi-site, multi-method processes that include at least one focal, well-publicised deliberative forum. At the same time, for such forums to serve their ends, they need to be plugged into both the informal public sphere and formal decision making. Referendums, in the limited ways suggested here, can help achieve both of those goals.

Acknowledgements

This paper relies on primary research conducted as part of the Sparking a National Conversation project, funded by the Australian Research Council (DP160102598). My thanks to the ARC for their support; the Centre for Deliberative Democracy and Global Governance at the University of Canberra, the project’s institutional home; and to Alice el-Wakil, Spencer McKay, and the journal’s anonymous reviewers for their constructive comments and criticisms.

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Notes on contributor

John Parkinson is Professor of Social and Political Philosophy at Maastricht University, and Adjunct Professor of Politics, University of Canberra. He works on the relationships between formal policy making and a wide variety of practices in the public sphere, crossing boundaries between normative political theory, public policy, political sociology, and cultural studies. His published works includes Deliberating in the Real World (Oxford, 2006), Deliberative Systems with Jane Mansbridge (Cambridge, 2012), Democracy and Public Space (Oxford, 2012), and Mapping and Measuring Deliberation with André Bächtiger (Oxford, 2019). His next book will be Deliberative Systems in Action with Sebastian Delaile and Núria Franco-Guillén, also with Oxford University Press. E-mail: [email protected]

Additional information

Funding

This paper relies on primary research funded by the Australian Research Council (DP160102598).

Notes

1 Compare Smith (2009) who argues that political institutions and practices should be evaluated in terms of the particular goods they help achieve or values they advance in a democratic system, and not against every possible democratic value simultaneously. See also Warren’s (2017) ‘problem-based’ approach, which recommends more holistic, context-sensitive evaluation and not an exclusive focus on institutional design.

2 The remaining 33 members of the Convention were representatives of political parties, mainly from the Oireachtas, but also a small number of members from the Northern Ireland Assembly.