Finding common ground

Deliberative democrats have abandoned the ideal of consensus in favour of a range of different, more realistic alternatives. But these alternatives provide little anchorage to guide or even evaluate deliberative practice – something acutely problematic given the contemporary context of accelerating polarization in many advanced liberal democracies. In this article, we turn to Stalnaker’s account of the ‘common ground’ – the shared pool of information that is agreed upon by the parties to a discourse – to reassert a distinct ideal standard for democratic deliberation which remains malleable enough to apply across messy contexts of real-world political contestation and debate. Our account offers an appropriate normative yardstick by which to assess deliberative practices across different discursive contexts, as well as impetus for further experimentation and innovation in efforts towards democratic renewal and reform.

seem to provide little by way of purchase for understanding and assessing what ends contemporary deliberative practices ought to be oriented to, nor for underpinning efforts at democratic renewal and reform in how they achieve those ends. The differentiated range of outcomes outlined by Mansbridge et al. (2010) can tell us very little about where and how to find common ground.
In this article, we seek to remedy this problem. We turn to linguistics -specifically the seminal work of Stalnaker (1975Stalnaker ( , 1984Stalnaker ( , 2002 -to clarify the concept of 'common ground', and to promote 'common ground' understood in this technical sense as an alternative ends of deliberative practice. Stalnaker's (2002) insight is that, in conversations, participants often accept certain propositions for the purpose of the conversation without necessarily believing them. For example, participants engaged in a discourse around climate change may accept that a particular projection scenario for climate change is the most likely scenario, without necessarily committing to believing that this is the case. The participants accept the scenario's validity, which means that the conversation proceeds as if this and no other scenario is the case. Meanwhile, participants may have doubts for various reasons that cause them not to believe that the projection is the most appropriate, but that allow them to accept that the projection is a viable model. 1 Propositions that are accepted in this way form the basis of the participants' 'common ground', such that any proposition inconsistent with what is accepted is considered impossible by participants. An important caveat on the common ground is that all participants of a conversation must believe that a certain proposition is accepted by all other participants. That is, they do not have to believe the propositions, but they do have to believe that every participant accepts the proposition for the purpose of the conversation. It is this property of the common ground -that participants must explicitly agree on what is accepted during the conversation -that leads Stalnaker (1975: 273) to conclude that the common ground forms a necessary 'resource for the communication of further information', against which the conversation's speech acts are understood. We argue that 'common ground' provides the anchorage missing from adapted conceptions of deliberative democracy which allows us to understand and assess deliberative and democratic practices. Normatively, it safeguards the ideal of inclusion in deliberative democratic theory because it is a standard that we can reasonably expect any discursive setting to attain. Unlike with consensus, or demanding alternatives proffered more recently in deliberative democratic theory, the pursuit of the common ground is not liable to exclude or marginalize vulnerable actors who lack resources to doggedly pursue their interests or develop an over-arching understanding of the dimensions of a given debate. Analytically, it offers a standard which is specific enough to apply to individual settings of deliberative practice, but also adaptable enough to apply to and across the broad range of interconnected settings envisaged in the contemporary deliberative system. It reasserts a distinctly deliberative conception for understanding and assessing the quality of individual settings and whole systems, while remaining attuned to the vagaries and pathologies seen to afflict the pursuit of consensus.
The article proceeds in three sections. The first outlines the problem with the goal of consensus in classic deliberative theory, and the subsequent problems associated with efforts to replace consensus with alternative goals. The second introduces the concept of 'common ground' from linguistic theory. The third works with the conception of 'common ground' to show how it represents a more appropriate replacement for consensus in understanding and evaluating quality within deliberative settings and across whole deliberative systems.

The problem with consensus and its alternatives
The appropriate ends of deliberation has become a core theme which theorists have revisited over the last two or three decades of deliberative theory. Initial accounts oriented deliberation to the goal of consensus, as found in Jurgen Habermas's (1984) ideal speech situation, or Joshua Cohen's (1989) 'forum'; the idea being that actors ought to open up their claims to a process of challenge and reflection in the exchange of reasons, with potential to switch sides in an argument or to reach a new agreement based on mutual accommodation. Our concern here is not whether Habermas, Cohen or anyone else genuinely believed in the prospect of consensus -our interest is in the strong and enduring association between deliberative democracy and the ideal of consensus on the basis of interpretation of their pioneering work, such that the pursuit of consensus continues to be perhaps the defining feature of the deliberative model in popular (mis)conception.
Of course, the casual equation of deliberative democracy with the pursuit of consensus is a source of considerable irritation for most contemporary deliberative democrats. Mainstream deliberative democratic theorists have entirely abandoned consensus as the guiding goal of deliberative practice. Many -including, for example, influential thinkers like Dryzek (2000) and Warren (2002) -did so in response to a wave of theoretical critique from difference democrats. Critics like Sanders (1997) and Young (2000) saw the goal of consensus as an inappropriate standard given the pervasive pathologies and asymmetries of actually existing democratic practice. Upholding consensus as the gold standard, in this view, represents an extension of the excessive emphasis on rational argumentation in deliberative practice. It can thus equally serve to suppress radical critique and reinforce gender and power dynamics. Subsequent empirical observations have merely reinforced this direction of travel. First and most obviously, the vast bulk of empirical studies of deliberative practice shows that consensus rarely materialized despite the affordances of ideally structured forums and burgeoning expertise in 'best practice' facilitation and engagement (Niemeyer, 2011). Moreover, some of this research seems to bear out the difference democrats' concerns about the 'chilling effect' of an emphasis on consensus (Dahlgren, 2009). In other words, the pursuit of consensus might engender, in especially subtle and pernicious form, the very coercion against which the deliberative model of democracy is constituted. So consensus was not just an unrealistic standard, but one whose pursuit could be self-defeating.
It is thus not difficult to see why -despite the prevailing popular misconception -consensus has not played a prominent, positive role in any theoretical account of deliberative democracy for at least two decades. The root problem is not entirely resolved, however, because, we will argue, nothing has satisfactorily replaced the hole left by the abandonment of consensus. If deliberation is not oriented to consensus, what ought it be oriented towards? While there is no 'consensus' in response to this dilemma, two particularly prominent alternatives stand out.
One is in the work of Dryzek and Niemeyer (2006), and their notion of 'metaconsensus'. As the name suggests, the replacement standard here is not that actors agree in substance, but that they agree on the broader terms of debate. As such, the goal of meta-consensus requires that actors seek to clarify the frontiers of the discussion and the range of reasonable perspectives on the issue at hand. Dryzek and Niemeyer hope we can agree to disagree, but clarify precisely what it is we disagree about. Their account, importantly, is not merely conceptual but builds on a strong body of work mapping preference changes in deliberative settings. They show that effective deliberation can allow actors to become more fully aware of the frontiers of the debate, the internal coherence and consistency of different perspectives and the 'true' perspective that best represents their interests.
There is much to recommend this more subtle and nuanced account of what deliberation can and ought to achieve -and indeed we later seek to resurrect aspects of 'meta-consensus' in our own account of 'common ground'. But there is a significant problem. Dryzek and Niemeyer's account of meta-consensus is grounded empirically but also conceptually in a context of ideal deliberation. Yet both scholars (see Dryzek and Niemeyer, 2012) have been at the forefront of the shift away from conceiving of deliberation as an ideal activity occurring in a particular setting, and towards seeing it as something which occurs in imperfect and incomplete form across a range of differentiated but interconnected settings. They are both, in other words, proponents of the systemic account of deliberative democracy which is critical of a micro focus grounded in a context of ideal deliberation. It is not at all clear that their account of meta-consensus is or can be an appropriate standard to apply in systemic terms. Certainly, the notion of metaconsensus has no particular prominence in Dryzek's extensive writing on the deliberative system.
Moreover, in the context of a more organic and dynamic deliberative system, there is good reason to think that pursuit of a standard of meta-consensus has exclusionary potential. It seems highly unlikely that those who participate on the peripheries of the system -in activist enclaves, for example -can or should develop a sophisticated understanding of the frontiers of debate, the range of reasonable perspectives or the internal consistency of those perspectives. Take, for example, the engagement at 'knit ins' between ordinary passers-by and activists from the Knitting Nannas group, protesting coal seam gas extraction in rural communities in Australia (see Hendriks et al., 2020;Mendonc¸a et al., 2020). 'Knit-ins' -sitting and knitting colourful garments outside community centres and the like -offer a playful, non-confrontational 'conversation starter' with other members of the community as they walk by. The organic deliberation that this seeds -what Mansbridge (1999) calls 'everyday talk' -is extremely valuable in systemic terms, introducing new ideas to the shared pool of information, and connecting marginalized voices to the mainstream public sphere. But it does not push parties to understand the full dimensions of the debate, nor to question the coherence of different perspectives. It is, in other words, not aimed at meta-consensus. Nor should we expect or want it to be. As this example shows, meta-consensus is not a reasonable, inclusive standard to apply in the systemic account of deliberative democracy.
A second prominent alternative to consensus -one much more clearly linked and attuned to the systemic turn in deliberative democratic theory -is found in Mansbridge et al.'s (2010) provocative account. Mansbridge et al. take seriously the objections to what they term 'classic deliberation', including the pursuit of consensus, and instead seek to stake out an account of deliberative democracy which recognizes the ineliminability of conflict reflected and reinforced by pervasive asymmetries in democratic politics. As part of an effort to assert the place of self-interest in this context, they outline four different modes of what they call 'deliberative negotiation'. Each is oriented to different ends: convergence, where actors engage in deliberation (in a context of low conflict) and agree on a single outcome for the same reason; incompletely theorized agreements, where actors engage in deliberation and agree on a single outcome for different reasons, and their differences and the conflicts that lie within them are not fully fleshed out; integrative negotiation, where actors engage in deliberation to expand the borders of the problem and introduce new perspectives to 'dissolve' apparent conflict; and fully co-operative negotiation, where actors engage in deliberation to clarify competing interests and reach a distributive agreement that all consider fair.
Mansbridge et al.'s expansion of the legitimate ends of deliberation is very appealing in its adaptability to different political contexts. It also represents an important stepping stone on the way to the more fully fleshed out systemic account of deliberative democracy which now represents the orthodoxy. But their account is also emblematic of the concern surrounding this systemic turn (and indeed most systemic accounts in the social sciences) -that of a worrying functionalism. In recent prominent accounts, Owen and Smith (2015) and Goodin (2018) have questioned the validity of the recent 'systemic turn' in deliberative theory, and point to omissions, elisions and confusions in the rush to move from ideal settings of democratic deliberation to less-than-ideal deliberative systems. Goodin (2018: 888) puts it especially pithily: The long and short of it is just this. The original version of democratic deliberation was that of a cooperative quest for a rationally motivated consensus based on the respectful exchange of reasons among free and equal citizens. Subsequent stretching has extended the concept of deliberative democracy far beyond that -all the way to what looks more like a fractious struggle to strike a deal underwritten more by pragmatism than reason among people who are not particularly free or equal in their power and influence. . . . Should we really be willing to accept everything within that wide rubric as an equally acceptable form of deliberative democracy?
Mansbridge et al.'s differential account would appear to be a prominent example of this broader phenomenon. Put simply, their four proposed ends of deliberative negotiation could be read into, and thus used to justify, almost any setting. For better or worse, we know when a setting is aimed at consensus and we can assess how successful (or not) it is at achieving that standard. But how would we even know which of Mansbridge et al.'s four ends a setting ought to be or is aiming for, let alone whether it has been successful on these grounds? As Owen and Smith and Goodin forewarn, in foregrounding adaptability to real-world contexts, the risk is a loosening of normative standards and a lack of specificity as to reasonable expectations and demands of deliberative practice -a problem not just because it leaves normative theorists wrestling with ambiguity, but because it entails a lack of precision in how we empirically evaluate real-world practices and how we pragmatically seek to go about democratic renewal and reform.
The challenge that remains, then, is one of finding a suitable replacement for the notion of consensus. The forerunning discussion outlines three key characteristics essential to any such replacement.
First, as our discussion of the well-known limitations with the goal of consensus made clear, any such standard needs to enable sensitivity to the prospect of coercion. Aiming for consensus runs the risk of reinforcing the invisibility of underlying deliberative and democratic pathologies. The ends of deliberation must allow for the need to acknowledge and occasionally sharpen conflict in the context of pervasive distortions and pathologies in the real world of democracy.
Second, as our discussion of the limitations of the meta-consensus alternative made clear, any such standard requires adaptability. It needs to be loose enough to apply to and across the range of settings in a complex deliberative system. Defining the ends of deliberation too tightly risks failing to meaningfully account for, or speak to, the great diversity of communicative action and political participation that contemporary deliberative theorists have in mind.
Third, as our discussion of the limitations of the differentiated account made clear, any such standard also needs operationalizability. It needs to be tight enough to provide a meaningful normative standard in and across these settings. Defining the ends of deliberation too loosely runs the risk of falling into the functionalist trap, blunting the critical edge of deliberative democracy by permitting anything and everything as somehow constitutive of the system.
The key to developing an appropriate standard, and thus to finding common ground in everyday parlance, we hold, is in understanding 'common ground' in linguistic theory.

Understanding 'common ground'
Considering that it is first and foremost a theory of communication, deliberative democracy has had remarkably little overlap with linguistic theory or the philosophy of language. Indeed, in a colourful recent review, John Parkinson (2018) launches a critique of the account of communication at the heart of deliberative democracy scholarship on these grounds. His central concern is that the ideal discourse ethics taken up and applied by deliberative democrats in scholarship and practice are naive to the complexities of real-world communication in social interaction. He takes up the work of Saussure in particular to problematize core assumptions about how speech and language work in practice. Parkinson's efforts in de-constructing ideal linguistic assumptions in the context of real-world language use are a key first step. But they do not construct a viable alternative standard for the sort of normative account at the heart of deliberative democracy. We argue that an account of 'common ground', drawn from philosophy of language and applied linguistics, can bridge the gap between the ideal and the real.
In this endeavour, we see affinities with important recent work elsewhere in political theory that draws inspiration from the same philosophical wellspring (especially the work of Stalnaker). Stanley (2015), for example, sketches out standards for public reason in his recent work on propaganda in liberal democracies. In a more forensic examination still, Bosworth (2020) seeks to identify standards for salience in political argument. There is rich potential in exploring these affinities and better (re-)integrating developments in deliberative democratic theory with Rawlsian political philosophy -albeit potential we see as well beyond the scope of this article. In this section, therefore, we more straightforwardly focus on outlining how we understand 'common ground'.
Normative political approaches to the production of utterances in communicative situations, such as those we have mentioned above, rely implicitly on a notion of communication that places a primary emphasis on a speaker's intention in performing an utterance. In philosophy of language, this view is most influentially espoused by Grice (1957: 382) when he claims that a speaker means something by an utterance if (a) they have the intention of 'inducing a certain belief' in those listening, and (b) they intended for an audience to 'recognise the intention behind the utterance'. The act of interpretation is therefore the act of guessing the speaker's intention in producing a particular utterance at a particular time. This is not an unproductive view, and has, indeed, led to some important insights, particularly in linguistic pragmatics (see, for instance, the excellent Horn (1984)). However, the application of this principle of meaning has yielded some unsatisfactory results. Chief amongst these is the identification of a single, 'correct' interpretation of an utterance, following from the identification of a singular interpretation in producing an utterance. Under this view, the utterance acts primarily as a conduit between the speaker's intention and the production of a certain belief (in Grice's terms) in the hearer.
There is growing research that suggests that the conduit view of interpretation is unsatisfactory when investigating interpersonal aspects of communication. In particular, face theory (see Bargiela-Chiappini and Haugh, 2009;Grainger et al., 2010;Haugh, 2010;Spencer-Oatey, 2007), politeness theory (see Bargiela-Chiappini and Harris, 2006;Bargiela-Chiappini and Ka´da´r, 2010;Mitchell and Haugh, 2015) and studies of communication in languages other than English (see Haugh and Watanabe, 2009;Samra-Fredericks, 2010) have illustrated that the notion of a 'correct' interpretation may not be a particularly useful one. Instead, these studies suggest that an understanding of meaning in which an utterance's interpretation is subject to a process of negotiation is a more accurate picture of how interlocutors communicate. Theories that form part of what we might call an 'interactionalist' paradigm have been proposed by many authors, going as far back as Goffman (1967), Morrissey (2017). In such theories, no single interpretation is privileged at the level of the model. Instead, interactions between two interlocutors Hillary and Donald are constituted of overlapping 'cycles of negotiation', where each cycle takes the following general form.
• Hillary, based on an intention to communicate p, designs and produces an utterance. Note that designing an utterance for Donald to communicate p implies that Hillary tries to guess how Donald will interpret her utterance. • Donald, upon observing Hillary's utterance, tries to ascertain Hillary's intention in producing that utterance. Donald produces an utterance that provides Hillary with some evidence of how he has interpreted Hillary's utterance. In particular, Donald's utterance provides evidence of whether he has accepted p or rejected p.
What is important is that Donald's utterance determines how p is treated in subsequent turns of the conversation; put more radically, Donald's interpretation, evidenced by his utterance, determines the meaning of Hillary's utterance in subsequent turns of the conversation. This may seem an outlandish claim, but consider the phrase 'Take back control', which could be understood as a call to reinstitute national sovereignty, to save money to be spent on essential services or an injunction to restrict migration, and so on. While the speaker of this sentence certainly has a meaning in mind, the meaning that is taken forth into the conversation is determined by the actions of all players. This continues indefinitely and repeatedly: every utterance in a conversation forms the basis of further negotiations, and through repeated cycles of negotiation, a shared pool of information against which other utterances are interpreted is formed.
This shared pool of information is precisely the role that the common ground fulfils. Stalnaker (2002: 716) specifies that 'it is common ground that p in a group if all members accept (for the purpose of the conversation) that p, and all believe that all accept that p, and all believe that all believe that all accept that p, etc.'. The important distinction in Stalnaker's definition is between acceptance and belief. For Stalnaker (1975Stalnaker ( , 1984Stalnaker ( , 2002, these notions rely on a theory of possible worlds, which has subsequently come to be associated with the formalization of an intuitive means of talking about counterfactuals. When a politician implies that a hospital would have been built in the electorate that she contested at the last election, had she won that election, she is referencing a state of affairs that differs from the 'actual' state of affairs in some significant way. The term 'actual' is important, because in possible worlds theory the world that one takes to be the true world is just one amongst infinitely many candidates for the actual world.
Individuals estimate which of these candidates is the actual world, based on their beliefs about the possible worlds, which we discuss in greater detail below. According to possible worlds theory, then, the politician is speaking of a possible world, and claiming that in one of the plausible candidates for the actual world, the facts of the world were different enough to yield an advantageous outcome for the residents of the contested electorate.
Following Stalnaker (1984: 45), a possible world is a 'state of affairs', or a 'way things might have been'. Importantly, a given possible world is identified by the sum total of all of the facts that are true in that possible world. Possible worlds are fully specified, insofar as in any given world w, any fact that we are interested in is either true or false in a particular possible world, provided that we are able to examine the possible world sufficiently well. As such, a world is defined by specifying all of the statements that are true in that world. So, 'that Hillary is a politician' or 'that the sky is blue' are statements, and therefore constituents of a possible world. If a single statement's truth differs between two possible worlds, then they are not identical. Two worlds are therefore distinct not only by whether a particular politician represents a particular electorate, but also by the height of that politician, or whether her name differs between worlds. Suppose, then, that there are multiple possible worlds in which Hillary is precisely 180 centimetres tall. According to Stalnaker, the statement 'that Hillary is precisely 180 centimetres tall' denotes a set of possible worlds.
To be clear, the notion of possible worlds is used here as a useful conceit -Stalnaker (1984: 3) calls it a 'convenient fiction' -to delineate the true statements relative to their content and in particular situations. The utility of this 'fictitious' view derives from the intuitiveness of the metaphor: the notion that certain facts are contingent, and that there is conceivably a state of affairs in which those facts differ, does not appear to be problematic. Generalizing this intuition to notions of possible worlds is merely a formal device that allows us to examine general properties of statements and their truth-values without being burdened by minute differences between statements. But there is something else that renders this fiction useful. If a statement is 'that p', then it can also model belief 'that p'. To claim that Hillary believes the statement 'that p' to be true is equivalent to claiming that she believes the worlds that are consistent with the statement are plausible candidates for the actual world. An agent's beliefs are the sum total of what she considers consistent with the 'actual' world.

Finding common ground
Belief is a trivial condition in discourses that are uncontested. Few people believe that the sky is not blue, and so common belief of this statement will not be difficult to ascertain. In discourses where the facts themselves are contested, however, requiring belief for meaningful conversation becomes cumbersome. If each agent considers the others' proposals to be inconstant with her estimation of the 'actual' world, then she will reject the formation of common beliefs. By merely requiring assent to 'that p' -that is, assuming or granting that 'that p' is the case for the purposes of the conversation -acceptance emerges as a more plausible alternative to belief in contested discourses. We claim here that building a common ground, in the technical sense put forth by Stalnaker, should provide the standard for discursive settings in a deliberative system. The advantages of this approach are manifold. It provides a linguistically cogent model of discourse, for it permits a more meaningful model of utterance design and interpretation in discursive contexts.
In the following sections, we demonstrate how anchorage in this understanding of 'common ground' can fill the hole left by the abandonment of consensus as a normative ideal in deliberative theory. We address in turn how an emphasis on finding or building common ground meets each of the three criteria outlined in the previous section: the need for sensitivity to the prospect of coercion; the need for adaptability in the shape of a loose enough standard to flexibly apply to different discursive settings or across a deliberative system as a whole; and the need for operationalizability in the form of a tight enough standard to enable meaningful application in assessments of deliberative quality. In each case, we demonstrate the point through reference to existing scholarship on deliberative democracy. To be clear, none of the research that we point to expressly uses 'common ground' as a yardstick. We think that if the work can be fruitfully seen through this lens, and in the context of the article, it helps to demonstrate the utility of common ground as a normative standard in rather more concrete terms.

Sensitivity: From common belief to common acceptance
First, common ground can provide a normative standard that remains highly sensitive to the potential for coercion; the expectation is not that everyone commits to a particular proposal or outcome, nor even that everyone seeks commitment, but that they agree upon a pool of shared information that can be used to inform later proposals or outcomes.
The easiest way to introduce and explain this point is to distinguish common ground from the notion of common belief, which underpins the old consensus ideal in deliberative theory. Consider a group, each member of which has her own beliefs. Common belief 'that p' occurs when every member of the group commits to the truth of 'that p'. Imagine a set of the worlds that contains those worlds that are in every individual's set of plausible candidates for the actual world; this is called the group's common belief. If all individuals believe 'that p', then we would also expect all worlds consistent with the statement to be in the group's common belief. Common belief may appear to be better suited to acting as a shared resource for the interpretation of utterances. Indeed, a degree of shared belief constitutes the basic condition of intelligibility for communication between members of a group, and the establishment of background information that can act as a resource for understanding appears to be a primary concern of interlocutors early in conversations (see, for instance, Arundale, 2006;Asher et al., 2017;Goffman, 1967;McCready, 2015). The complication that arises is that common belief may be too strident a condition in many cases, particularly in discourses surrounding contested or contentious issues. As difference democrats were quick to assert over two decades ago, the consequences of an emphasis on creating common belief (akin to consensus) can be dangerous and coercive in practices, especially for historically marginalized groups.
In contrast, common ground lacks such strident conditions. As opposed to common belief or common knowledge, the common ground is deliberately constructed through mutual acceptance of particular sentences, which requires interlocutors to (a) understand that a common ground exists, and that the information in the common ground is necessary for the successful interpretation of utterances during the conversation; and (b) explicitly identify certain propositions as elements of the common ground. Compare this with earlier definitions provided by Grice (1989: 65, 275) as propositions that operate in the background of a conversation, and by Stalnaker (1975: 273) as the 'information [taken] for granted' throughout a conversation. Neither of these suggest that this information is agreed upon, nor that the information is intentionally identified as common ground information. Indeed, these definitions apply more appropriately to common belief and common knowledge, since these arise without any intention, and are formed by the group members' beliefs or knowledge.
A useful example of these dynamics in action can be found in Edwina Barvosa's (2018) recent history of public engagement around LGBT rights in the US. Through careful examination of everyday practices of deliberation, she shows how individuals with rich social connections -what she labels 'deliberative entrepreneurs' -acted to slowly transform the terms of debate. She details how these 'entrepreneurs' were able to use their 'social familiarity' to seed new ideas and evidence within a wide range of networks. To be clear, Barvosa's argument is not that these 'entrepreneurs' succeeded in overcoming engrained prejudices directly or immediately. Many individuals engaged in these social networks continued -and still continue -to cling to prior beliefs about LGBT equality. But there is broader acceptance of the validity of alternative ways of understanding the issue. In other words, Barvosa's 'entrepreneurs' have disrupted a coercive and damaging 'consensus' about LGBT rights in many enclaves of social and political life. Their interventions, as credible conduits for new ideas and claims, have extended and deepened the shared information pool in these settings, gradually extending and transforming the common ground.
Crucially, as well as offering the anchorage to understand and evaluate the expansion of inclusion in deliberative systems (as in the example above), a focus on the common ground can also help to shed light on ongoing dimensions of exclusion and domination, including elements that an ideal of consensus or meta-consensus might obscure. Common ground is a yardstick against which to evaluate the extent to which parties to a discursive setting acknowledge each other's claims and ideas, and to determine whether exclusions of particular claims and ideas are reasonable.
To flesh out this point, we return here to the example of deliberation on climate change in Australia with which we introduced the concept of common ground. In the study underpinning this example, Hobson and Niemeyer (2013) asked participants to use a series of escalating climate scenarios as the basis for their deliberations. The involvement of climate change sceptics, however, presented a quandary for a project grounded in the deliberative ideal of inclusion. In the end, all bar two participants were willing to acknowledge the scenarios as the basis for discussion, despite many privately professing views that they were unrealistic (many perceiving them as too extreme, or based on invalid data or assumptions). The two who would not were the two most strident climate sceptics. They rejected the scenarios outright, and refused offers from the event organizers even to counter the scenarios with other sources of information in the deliberations. They simply would not engage. Their (self-)exclusion from the remainder of the process should be seen as reasonable not because they failed to reach agreement (consensus) nor because they failed even to agree to disagree (meta-consensus). Instead, it is better understood as reasonable because they refused to even acknowledge the shared pool of information on which further deliberation could proceed (common ground). 2

Adaptability: From amicable conversations to polarized discourses
The second point is that, unlike Dryzek and Niemeyer's ideal of metaconsensus, common ground is loose enough to fit across the variety of discursive contexts that a deliberative system entails. Common ground is adaptable in two primary ways. One, it applies equally to amicable and deeply contested scenarios; two, it scales from the level of one-on-one conversation to that of a broad public discourse at the systemic level. We outline both, and the relationship between them, below.
On the first point, the less strident condition of acceptance that governs inclusion of sentences in the common ground permits it to accommodate a wide range of discursive scenarios. Building a shared resource of accepted propositions is simple to imagine in amicable conversation, but it is also possible to construct such a resource in a more polarized or adversarial context.
The key here is not just that seeking 'common ground' lacks classical deliberation's requirement that interlocutors reach common belief (indeed, as just discussed, deliberation may sharpen conflict), but also that it lacks the requirement in Dryzek and Niemeyer's adapted meta-consensus ideal that all interlocutors develop an abstract understanding of the range of perspectives on the issue. To reach common ground, those engaged in discussion in, for example, a face-to-face debate need not come to terms with the perspectives that adversaries put forward, nor even reflect on the coherence and consistency of their own account they subscribe to. Both require a level of reflexivity often missing from the immediate cut and thrust of real arenas of political contestation and discussion. Instead, the much more realistic expectation is that actors simply need to agree on, and seek to further build, the shared pool of information on which all interlocutors in the given setting base their interpretations.
Examples from micro-deliberative practice prove the value of this reduced, adaptable expectation. Consider, for example, David Ryfe's (2006) account of anecdotal storytelling in mini-public discourse. Ryfe's study is centred around trying to understand how individuals engage in small group deliberations, focused on close scrutiny of five example interactions within the National Issues Forum programme in the US. One of his key insights is to show how storytelling -itself a key mode of real-world communication, especially in the context of conflictallows marginalized actors an acceptable 'way in' to political discussions on controversial topics. In Ryfe's examples, the anecdotes that actors tell seldom generate common belief among interlocutors (as in the classical account). They seldom go so far as to prompt a form of reflection on the very contours of debate or on the variety of perspectives brought to bear on contentious issues (as in Dryzek and Niemeyer's account). What they do instead is introduce new information which is accepted as consistent with the possible worlds that all interlocutors imagine. In other words, the intuitive deliberative democratic value of this form of storytelling is that it strengthens the common ground.
Second, common ground is scalable from conversations between two interlocutors to discourses involving manifold parties across multiple settings. We argue that explicit negotiation of and agreement upon a set of propositions is necessary for interpretation in any given discourse, for it is precisely by identifying the topic of a conversation, and by specifying what is assumed, that interpretations are able to be selected felicitously. However, it appears that the kinds of information that are accepted are substantially different. In broader discourses, common ground propositions must generally be to some extent justified. The reason for this is simple: each participant to a discourse has a set of beliefs about the world. As the number of participants increases, the intersections and disjunctions of these beliefs become more numerous and complicated. In formal terms, the intersection of the extensions of participants' beliefs decreases in size as more beliefs are added. Morrissey (2017) provides a mathematical proof of how the common ground's extension decreases as more propositions are accepted into it, and a similar process is undertaken with the intersection of beliefs: the greater the number of propositions in an individual's belief, the fewer worlds that are consistent with that belief. We can expect that the number of worlds consistent with multiple sets of beliefs is smaller again.
So, justification, in the form of evidence or argumentation, is aimed at making propositions plausible. A plausible proposition in world w is one that is at least not inconsistent with w. This includes propositions that are believed: a proposition 'that p' that is believed in w is naturally plausible in w. However, an agent also considers a proposition plausible if it does not violate her beliefs. While possible worlds are fully specified, agents' knowledge of possible worlds is limited, and they expect that there are things about a given possible world that they are not aware of. 3 Making propositions plausible is an aim in discourses that are adversarial or cooperative, and of whatever size. However, plausibility is easier to establish with a smaller number of participants, and therefore persuasion will play a more important role in larger discourses, and we anticipate that this will be particularly true in discourses that are polarized or contested. This means in practice that we ought to demand greater justification from settings of high-profile, elite debate (such as legislatures) in the cause of building common ground for the deliberative system as whole.
In this sense, recent work comparing the role of legislatures to other settings in deliberative systems points to the scalability of common ground as a normative standard. In conceptual terms, there has been increasing emphasis on legislatures performing a symbolic role in democratic life (see Parkinson, 2012), especially as outlets for the expression of ineliminable conflicts in the reaching of messy, compromised democratic decisions (see Rummens, 2012); in other words, these are democratic venues where the common ground is most publicly staked out. In empirical terms, too, scholars have noted that political elites in legislatures tend to offer higher justificatory rationality, and lower mutual respect, than everyday citizens in more informal settings (see Pedrini, 2014); in other words, building and maintaining common ground in different settings entails different norms and practices. Either way, the point is that common ground provides a useful yardstick which we can apprehend and apply in contextually appropriate ways across deliberative systems.

Operationalizability: From slippery functionalism to specifiable standard
The final point is that common ground offers a tight enough standard by which to assess different discursive scenarios. Recall the problem, discussed in an earlier section, that the alternative ends of deliberation adopted and developed by theorists in place of consensus are slippery in nature, and their relevance or application to any given scenario uncertain and seemingly ad hoc. Common ground, in contrast, ensures a specific and stable yardstick against which real-world practice can be evaluated.
Conceptually, we can know when a discursive setting is oriented towards finding common ground. There are a specified set of ends that we can expect, such that we can assess any individual discursive scenario or deliberative system on the basis of whether it promotes or precludes the finding of common ground. Morrissey (2017) identifies that common ground is constructed through an iterative process of cycles of negotiation, in which proposals are made, which are subsequently (partially or fully) accepted or rejected. Propositions are proposed when new information is introduced by an interlocutor into the conversation, and the proposition may be subsequently accepted and rejected either by explicit disputation of or agreement with the information, or else by tacitly accepting the information. In the former, explicit case, interlocutors may engage in what Krifka (2007: 17) identifies as common ground management: 'information about the manifest communicative interests and goals of the participants'. Krifka notes that questions, for instance, do not generally introduce new information into the common ground, but they do communicate an interlocutor's requirement for particular information. Repp (2012) argues that illocutionary negation and acceptance -that is, negation and acceptance of the intention in producing an utterance, rather than what the utterance proposes -targets the status of particular propositions in the common ground. Thus by accepting or rejecting a speaker's intention in producing an utterance, the common ground is explicitly managed, and the constitution of the common ground is deliberately negotiated.
Empirically, too, common ground is readily operationalizable, using a set of methodological tools drawn originally from linguistics but already well established in disciplines as diverse as communication, sociology, public administration and political science. We see this in, for example, studies of policy narratives, which are, according to Roe (1994: 3), 'the assumptions for policymaking in the face of [a certain] issue's uncertainty'. The policy narrative, according to this view, forms a basis upon which discussions of the issue and of possible policy measures may be discussed. And, indeed, studies of policy narratives suggest that polarization occurs when participants in a policy discourse do not share a sufficient degree of common ground.
Take, for example, Mark Roberts's (2018) recent analysis of the 'communication breakdown' surrounding the 'Citizen Power' initiative in Peterborough, England. Citizen Power was a wider programme in British civil society, and in the case of Peterborough it took the form of an initiative to empower citizens in four specific policy areas -community arts, education, drugs and alcohol misuse and environmental concerns. Roberts deploys narrative policy analysis to track the trajectory of contestation about the initiative -from an initial period of optimistic consensus through to considerable antagonism and recrimination, largely because proponents of competing narratives about 'Citizen Power' simply 'could not understand one another'. His account provides penetrating insight into how an apparent consensus can break down in the absence of sufficient common ground. Roberts reveals in vivid detail how embracing a messier account of deliberative practice, and (in his case implicitly) accepting the ineliminability of political conflicts, does not have to blunt the analyst's critical edge. Indeed, his account provides meaningful, tangible purchase on the pathologies that can afflict deliberative practices at the broad scale.

Conclusion
We have argued that common ground -by which we mean a shared pool of information that is agreed upon by the parties to a discourse -provides means of overcoming the problems that plague other theories of deliberative democracy. Similarly to other theories, common ground provides a standard of assessment of discourses, but by targeting common ground rather than common belief, it allows for a more plausible model of deliberative discourse, particularly when the discourse is polarized. It can do so because, unlike common belief, it does not require agents to commit to the truth of a particular statement, but requires merely that they accept the statement. The aim of persuasion in deliberative discourses is thus shifted from a form that aims to disabuse one's opponent of what they believe, to a form that aims to convince one's opponent to accept some statement for the purpose of a conversation. It is clear that this provides a more realistic goal for highly polarized debates, and is thus a better fit with the 'realist' impulse underpinning the systemic turn in deliberative democratic theory (see Mansbridge et al., 2012;Boswell 2016). But, unlike the other alternatives to the old consensus model, common ground is also a recognizably and measurably deliberative standard which can be applied to and across multiple settings of political debate.
The next task is to think through how a commitment to finding common ground can improve the quality of deliberation. The old consensus model of classic deliberation provided an important inspiration for innovative institutional design and experimentation. Indeed, many of the most prominent and powerful 'deliberative designs' are modelled on the pursuit of consensus (Hendriks and Carson, 2008). In contrast, the existing alternatives -though implicitly and at times explicitly critical of the implications of the pursuit of consensus in practical experimentation -have enjoyed rather less purchase in deliberative practice (albeit see Asenbaum, 2016). Because these alternative prescriptions lack a clear orienting standard, they do not provide very useful guidance on what reformers should actually do to enhance deliberative quality. An emphasis on finding common ground can help to reimagine and reinvigorate innovation, stretching the realms of experimentation well beyond the confined forum and across sites of set-piece political theatre, closeddoors negotiation and 'wild talk' online. All these settings -and the myriad spaces in between -can be tweaked and geared towards the pursuit of common ground. The upshot can be a reduction and amelioration, overall, in the accelerating fragmentation and polarization which threatens contemporary democratic discourse.