Democratic Self-Determination and the Intentional Building of Consensus

This paper defends two fundamental but under-theorized insights coming from the theory of deliberative democracy. The first is that consensus is valuable as a precondition of democratic collective self-determination, since it ensures that democratic decisions display an adequate degree of integrity and consistency and therefore that the polity can act as a unified agent. The second is that consensus in this integrity-building role is essential if citizens need to act as decision-makers; it ensures that the decisions that issue from the exercise of their political rights are meaningful, and that they are so as the intended result of their joint agency. Aggregative approaches, which do not acknowledge this role of consensus, offer an atomistic account of voting and other political rights, and model the outcomes of democratic decision-making as unintended aggregative consequences of individual votes. In these models, democratic political agency and the decision-making power of citizens are curtailed, because citizens do not exert any intentional control on the final outcome of the decision-making process in which they participate. Although the insight on these shortcomings comes from the deliberative camp, I show that the most prominent accounts of how deliberation is supposed to further consensus in its integrity-building role can be subject to the same criticisms. In fact, in these models consensus is achieved as a by-product of people's engaging in deliberation. Although interactive, these approaches are still atomistic and unintentional. As an alternative, I propose a model of democratic decision-making that acknowledges the role played by the citizens' intentional consensus-building through the strategic use of their political rights. Author Biography Valeria Ottonelli is Associate Professor of Political Philosophy at the University of Genova, Italy. Her main research interests are in the normative theory of democracy and in the theory of justice in migration. Her work appeared in Political Studies, The Journal of Political Philosophy, Critical Review of Social and Political Philosophy, International Migration Review and Politics, Philosophy and Economics.


Democratic Self-Determination and the Intentional Building of Consensus
In the normative theory of democracy, consensus can play various roles and functions.In some theories, consensus works as an ideal point toward which democratic decision-making should tend for democracy to meet the presuppositions of citizens' communicative rationality (Habermas, 1996;Lafont, 2006) or their relations as free and equal (Cohen, 1989).Other theories stress the role of consensus as a guarantee against unfreedom and domination; the broader the consensus within a political community, the fewer who will be coerced into following rules they do not endorse. 1 A third role in which consensus features in democratic theory is as evidence-given the appropriate conditions-of the epistemic reliability of democratic decisions (Nino, 1996; for an updated discussion, see Landemore & Page, 2015).Finally, consensus has been recently theorized as motivated by aversion towards the destructive effects of untamed political conflict (Ani, 2014).
Besides these prominent functions of consensus in democratic theory, there is one that has been less explicitly theorized, but which nevertheless underlies many debates on the feasibility of the democratic ideal.This is the role of consensus as a guarantee of the unity and integrity 2 of the polity in its decisionmaking capacity.Integrity, in this context, is defined as the unity of agency of the polity through time, as instantiated by the coherence and stability of its lawmaking and decisions.Integrity is guaranteed when majorities converge on a common and stable voting pattern, so as to ensure that the collective decisions made though majority rule will be consistent and intelligible.I will hereafter refer to consensus in this specific role, which serves as a guarantee for integrity, as integrity consensus.
This role of consensus is crucial.In fact, integrity consensus is a necessary condition of the collective self-determination of the political community, which many see as a defining principle -along with political equality-of democracy itself.Democratic self-determination means that the members of the polity jointly and collectively determine the course of their political action and the laws that will rule their domestic and foreign affairs.Consensus is crucial to the fulfilment of the principle of collective self-determination because selfdetermination requires some degree of integrity and unity of agency of the polity, at least in the minimal sense that its acts are not incoherent or its decisions meaninglessly casual.A divided community, in which different and scattered majorities take over cyclically and randomly, cannot provide unity of agency.
Integrity consensus consists in a convergence on common policy alternatives that is broad enough to generate a stable and meaningful course of action at the collective level, thus providing the grounds for the required unity of agency of the democratic polity.
In fact, it is exactly consensus in this integrity-building role that has come under the attack of the long line of political theorists and scientists who have sought to discredit the notion of popular self-determination as meaningless and suspect, from Hans Kelsen (1955) to Joseph Schumpeter (1942) and to the well-known anti-populist manifesto by William Riker (1982).Summarising this line of thought, Riker famously argued that we cannot make sense of the ideal of popular self-determination because such an ideal presupposes that the polity displays at least a minimal degree of consistency or "integrity" through time, and such consistency requires a degree of consensus that cannot be expected in conditions of pluralism, that is, in a society where people have different value systems and worldviews.Indeed, in such a society people are likely to have different orderings of the alternatives on the agenda, and when this happens the aggregation of votes through democratic procedures is liable to generate cycling majorities that contradict each other.In other words, if we want selfdetermination, we need to have integrity consensus; but integrity consensus is impossible and undesirable in a pluralist society, in which people have different views of how the alternatives on the political agenda should be ranked, and vote accordingly; therefore, the ideal of self-determination is unachievable and should be purged from democratic theory.
In this paper I have three aims.My first aim is to present an argument in defence of the importance of integrity consensus, drawing from some powerful insights-mainly developed within the deliberative camp-on the essential connection between integrity and self-determination at the collective level and the role of democratic citizens as decision-makers (first Section of the paper).My second aim is to show that, notwithstanding the fact that these insights were mainly developed within the deliberative camp, the most popular models of how integrity consensus can be achieved through democratic deliberation fail in two respects: a) they do not accommodate pluralism, which, in accord with a longstanding and established tradition of thought, I assume to be a permanent and defining feature of liberal democratic societies; 3 and b) they do not do justice to the notion of citizens as decision-makers (second and third Sections), if we believe that decision-making comprises an element of intentionality regarding the meaningfulness of one's decisions.My third aim is to sketch an alternative 3 The assumption that in a free democratic society people will have different systems of values and worldviews is shared not only by scholars like Kelsen (1955) or Riker (1982), but by most thinkers in the liberal democratic tradition, from Mill to Dewey and to Rawls and Habermas.On the relevance of this assumption for the theory of deliberative democracy, see for example Bohman (1996); Mansbridge et al. (2010); Chambers (2018, p. 63).
account or model of how integrity consensus can be produced, centred on a dimension of political agency I call 'political prudence.'This model, in which political prudence complements deliberative agency, provides a better account of the decision-making powers of the citizens of a democratic polity and is more respectful to pluralism than the standard models that suppose that integrity consensus can be produced through deliberative agency alone (fourth and fifth Sections of the paper).

The Importance of Integrity Consensus
Can and should we really do without integrity consensus as a defining element of democracy, as suggested by Riker and other critics?
As just recounted, this may appear to be a necessary conclusion if we want to preserve pluralism.In fact, social choice theory has shown that in conditions of pluralism any collective choice procedure that fulfils some minimal democratic desiderata is liable to lead to inconsistent sets of decisions, which makes the requirement of consistency and integrity impossible to fulfil.Conversely, if a democratic polity needs to escape inconsistency and voting paradoxes, it must achieve a degree of integrity consensus and homogeneity that in our pluralist societies is unlikely to occur and would be unreasonable to expect (Hardin, 1993;Knight & Johnson, 1994;van Mill, 1996).
According to this line of thought, however, not much is lost if we give up integrity consensus, since we can make perfect sense of the democratic ideal without it.In fact, it can be argued that the three distinctive principles of democracy-participation, freedom and equality-can be honoured by a 'liberal' model of democratic decision-making (Riker, 1982), in which what counts is that each and every member of the polity has an equal chance to affect the political process by participating through democratic rights.More recently, Fabienne Peter (2009) has responded to the worries about the voting paradoxes discovered by social choice theory and the newer literature on the paradoxes of judgment aggregation (Pettit 2001;List & Pettit, 2002;Dietrich, 2006) by pointing out that what really counts for democratic legitimacy is that decisions are arrived at through a fair and reasoned process.
In short, according to this line of thought integrity consensus, and the ideal of self-determination that it makes possible, should be dropped from democratic theory because they are incompatible with pluralism.Moreover, dropping integrity consensus and democratic self-determination does not cause any significant loss, because we can make sense of individual democratic rights without presupposing that they serve the purpose of building a unified, meaningful line of action at the collective level.
Both of these conclusions have been challenged by powerful arguments developed from within the deliberative conception of democracy.
Let's consider first, in this Section, the conclusion that not much is lost when we drop integrity consensus-and thus self-determination-as a democratic ideal.Contrary to this claim, some defenders of deliberative democracy have pointed out that if we give up the possibility of self-determination, we miss something crucial about the rights and prerogatives of the individual members of a democratic polity (see especially Bohman, 1996;Richardson, 1997Richardson, , 2002)).In fact, such rights lose much of their meaning and value if they are decoupled from the notion that they guarantee the individual participation in a collective process of decision-making.In a democracy, individuals are granted political rights that allow them to act as decision-makers over meaningful lines of action rather than participants in a random, meaningless and haphazard process that only guarantees a roughly equal chance to achieve one's ends or to protect one's negative liberty.
The poverty of accounts like Riker's (1982) in elucidating democracy as participation in a collective decision-making process has been exposed from various angles.Jürgen Habermas (1996), for example, has claimed that if we understand democracy simply as an institutional procedure for accommodating on a random basis the conflicting interests of different social groups, then we cannot make sense of the actual practice of democracy, in which citizens seem to act on the belief that they can affect the democratic process so as to produce rational and just decisions.If citizens did not believe this, parties' programs, public debates, and elections as we know them would lose much of their point and significance.David Estlund (2008) has pressed a similar point against purely proceduralist accounts of democracy.If democracy were simply a fair process for selecting the winning opinions or interests, there would be no point in adopting majority rule rather than a random decision-making device like tossing a coin.Habermas's (1996) argument is meant to vindicate the discursive side of democracy, while Estlund's (2008) is meant to show that democracy has an essential epistemic component.From these arguments, though, two morefundamental points can be drawn that are relevant for our present purposes.The first is that to make sense of democratic voting, we need to think of it as capable of producing meaningful decisions rather than random choices.Integrity consensus, as a minimal precondition of meaningfulness, cannot be given up so light-heartedly.Second, if we want to account for the role of the participants in the democratic process as decision makers, such meaningful decisions must come through institutional procedures that are responsive to their individual choices (votes).Citizens must be able to see that the decisions taken at the collective level (and their meaningfulness) are affected by their choices as voters.Majority rule is essential to democracy, while a random decision procedure such as coin-tossing is not, because majority rule, unlike selection by lot, responds to how each and every citizen decides to cast their vote.
We can further sharpen these insights.Consider a decision procedure by which each citizen has the right to express their requests or claims by a vote, and then an impartial judicial body adjudicates conflicts between them in a meaningful and principled way, thus producing a unified and consistent line of action. 4This procedure would be responsive to people's votes, and would issue meaningful decisions.However, in this process citizens would not act as decision-makers, but simply as sources of claims and bearers of interests that would constitute so to speak the raw material of meaningful decisions.Democracy, unlike the procedure just outlined, gives citizens the power and duty to directly participate in the production of meaningful collective decisions, by casting votes in a way that aims at a given result, in conjunction, of course, with the votes of all the other participants.
This means that if we want to account for the role of democratic citizens as decision-makers, rather than mere providers of inputs that feed a process they do not control, we need to conceive of democratic processes not simply as meaningful and responsive to citizens' vote, but as responsive to citizens' votes as intentionally aiming at producing meaningful results; in other words, the achievement of meaningful decisions must not be the product of chance, or of the intervention of some third party, but must be intended by citizens as they participate in decision making.
The comparisons with the drawing of lots and with the third-party hearing procedure, then, throws light on two major and interrelated shortcomings of conceptions of democracy such as Riker's (1982) and their dismissal of integrity consensus as a precondition for meaningful, unified lines of action at the collective level.The first is that they are atomistic conceptions of democratic decision-making.They conceive of individual participation in the political process as the separate expression of personal preferences or opinions rather than the active participation in a collective decision-making process.In these accounts of democracy citizens do not make decisions, because the final outcome of democratic procedures is determined, so to speak, 'behind their back': decisions are made as a result of their desiderata or preferences but are eventually determined by processes of aggregation and adjudication in which they do not participate as decision-makers.Atomism prevents them from acting as decision-makers because democratic decision-making is a collective process, which requires contributing with one's vote to a common effort to bring about a meaningful decision by a majority (or to oppose it by a minority).
The second and consequent shortcoming of these conceptions is that they conceive the outcomes of the democratic process as unintentional.Of course, in a process such as democracy in which many people participate with equal influence, for each of the participants the result cannot be 'intentional' in the sense that they have personal and absolute control over it.However, the results of the democratic process can be intentional in the sense that each of the participants aims at producing-by interacting with all the other participantsthe final content and meaning of the outcome of the collective process of decision-making.
In sum, by construing democratic decision-making as atomistic and unintentional, Rikerian models of democracy do not simply rule out democratic self-determination and integrity consensus.They also offer a very impoverished and shallow reconstruction of the value and functioning of democratic prerogatives, and of the individual political agency of democratic citizens.Integrity consensus is an essential condition for collective decisions to be meaningful and therefore for democratic participation to be conceived as a process of collective decision-making, rather than a random device fed by an atomistic pattern of individual votes.

Deliberation, Integrity Consensus and Pluralism
The insights about the essential connection between individuals' role as decision-makers and self-determination on which I built my argument in the last section mostly come from the deliberative camp.They aimed at vindicating not only the importance of self-determination for the democratic ideal but also a richer model of political agency than the one assumed by minimalist accounts of democracy like Riker's (1982).In the deliberative model, political agency does not simply consist in the expression of one's wishes and opinions, but in the participation in a reasoned collective process of decision-making, in which citizens exercise what is called discursive rationality (Bohman, 1996;Dryzek, 1992).
Another important, and related, suggestion coming from the deliberative camp responds to the second major claim of the Rikerian model of democracy, that is the claim that the kind of consensus required for collective self-determination and integrity is impossible to achieve, and in any case undesirable, because such integrity consensus is incompatible with pluralism.
In response to the claim that integrity consensus is impossible to achieve, some deliberativists have suggested that once we construe political agency as the participation in a reasoned collective process of decision-making, then we also find the way to overcome the paradoxical results exposed by social choice theory, thus proving that integrity consensus is not only needed, but also possible (for some classical statements see Cohen, 1989;Dryzek, 2002;Elster, 1986;Sunstein, 1988), and therefore that democratic decisions can be meaningful.
The fundamental idea is that democratic decision-making turns out to be a random and meaningless process only if we assume, like Riker's model does, that democracy consists in the aggregation of idiosyncratic preferences or opinions that people develop and entertain in private.If instead we conceive of democracy as a decision-making method in which voting only comes at the end of a process of collective reasoning, then we also produce the conditions for avoiding cycles and inconsistent sets of decisions, thus ensuring an adequate level of integrity and unity at the collective level.The factual premise that underlies this claim is that by deliberating-by exchanging reasons in publicthe democratic polity will naturally converge towards a reasoned consensus, and reasoned consensus guarantees consistency and integrity in the decisions made through majoritarian devices.
This factual premise, in turn, is based on the assumption that public deliberation has a transformative effect on individual preferences (Christiano, 1993;Goodin, 1986;Pildes & Anderson, 1990;Sunstein, 1991).Deliberation makes people justify their preference orderings in public; this induces them to revise their preference orderings in such a way that they will be justifiable according to reasons and principles others can share.Selfish or idiosyncratic preference orderings will not withstand public scrutiny and will therefore be discarded or revised (Elster, 1998b;Goodin, 1986).The process will then naturally order individual preferences in common and publicly shared patterns.In other words, it will produce consensus (Cohen, 1989;Elster, 1986).
These claims have been challenged by a copious empirical and theoretical literature. 5But even if it be granted for the sake of the discussion that deliberation does produce consensus, we are left with the second main challenge to collective integrity, which is the claim that, even if integrity consensus is possible, it is undesirable because it suppresses pluralism, which is a defining and valuable feature of a free, democratic society (Knight & Johnson, 1994;Rescher, 1993).In fact, the suppression of difference is politically suspect (Sanders, 1997;Young, 1996) and the dissolution of disagreement may even undermine rational public discourse (Friberg-Fernros & Schaffer, 2014;Sunstein, 2003). 65 Serious doubts have been raised about the assumption that in a pluralistic society, in which people are free to reason in public, the natural outcome of these deliberative interactions would be consensus (Knight & Johnson, 1994;Ottonelli, 2010;Van Mill, 1996).For an assessment of empirical data, see Steiner, 2012, Ch. 6.For overviews of the empirical literature, see Thompson (2008) and Ryfe (2005). 6For a recent overview and discussion of the contrast between pluralism and consensus in deliberative democracy, see Martí (2017).
The preoccupation with difference and pluralism, as is well known, has made many leading deliberativists give up consensus as an ideal or a goal of deliberation (Mansbridge et al., 2010).However, some strands of the theory of deliberative democracy specifically preoccupied with integrity consensus -i.e., with the specific role that consensus plays in ensuring the conditions for integrity unity of agency at the collective level-have not abandoned this ideal; for it may be argued that, contrary to other functions of democratic consensus, integrity consensus does not require especially high levels of homogeneity in people's views. 7All that is needed is for the polity to display a sufficient degree of consensus to make its decisions meaningful and consistent.
In fact, a well-known discovery of social choice theory is that for majority voting to be immune from paradoxical and inconsistent results it is enough that individual orderings of the alternatives on the political agenda be 'single-peaked' (Black, 1958), meaning that they can be represented along a common spatial dimension.Borrowing this model from social choice theory, some mainstream deliberative theories of democracy have replaced the ideal of full consensus as a guarantee for integrity with less demanding ideals, such as 'metaconsensus', 'agreement at a metalevel' or consensus on the dimensions of political decisions (Dryzek & List, 2003;Dryzek & Niemeyer, 2006;List, 2002;List, 2018;List & Koenig-Archibugi, 2011;List, Luskin, Fishkin, & McLean, 2013).These lighter forms of consensus do not demand that citizens have identical preferences with regard to the available policy options; they only require that they share the same ways of conceptualizing such policy options and their implications along the same dimensions.It can be shown that if this requirement is fulfilled, then the individual orderings of the options on the political agenda will be 'single-peaked' (Dryzek & List, 2003;Dryzek & Niemeyer, 2006;Miller, 1992), and therefore majoritarian decision-making will produce decisions that display consistency and integrity.
The role of deliberation in achieving consensus has been revised accordingly: we cannot hope for deliberation to produce full convergence on how people rank different policy alternatives on the agenda, but we can still expect deliberation to produce consistent and rational collective decisions to the extent that it will generate a common dimension along which to conceptualize and evaluate such alternatives (List, 2002).This common conceptualization, by fitting the requirement of single-peakedness, will ensure that the outcomes of democratic procedures display integrity.So, for example, in a society that is deeply divided about the immigration policies to adopt, people will have very different views about how to rank the available options in terms of the right of immigrants to access to the national territory, their enjoyment of social and economic rights, and the acquisition of citizenship.But through deliberation people might come to characterise and weigh those disparate stances along one and the same evaluative dimension, for example exclusion/inclusion, ranging from those positions that most value excluding foreigners from all sort of access, to those that most value their full access and integration into society.If this shared conceptualization is made possible, then it can be shown that the choices made by majority voting will be consistent and display a unified line of action, despite the deep pluralism that characterizes the society in this policy area.

Why the Deliberative Approach to the Building of Integrity Consensus Does Not Fulfil Its Promises
The models that rely on deliberation as a device for producing integrity consensus (from now on, D-models), by stressing the importance of discursive rationality as an essential component of democratic agency, promise to rescue the notion of popular control and self-determination as essential to the democratic ideal.According to this line of reasoning, a democratic process guided by discursive rationality is conducive to consensus to a degree sufficient to guarantee integrity and consistency at the collective level, thus avoiding the paradoxical outcomes that would make the idea of self-determination nonsensical.Moreover, to the extent that D-models allow for lighter forms of consensus than full substantive unanimity, they promise to achieve such a goal without undermining pluralism.In this section, I would like to raise doubts about the validity of these claims.
Let us consider first the claim that D-models do not undermine pluralism.We have seen that this claim relies on the notion that to guarantee integrity and consistency in democratic decisions we do not need to achieve a high degree of uniformity in how people rank the options on the political agenda and vote on them.Integrity consensus, that is the kind of consensus that is needed in order to ensure collective integrity, can simply amount to a sufficient degree of uniformity, or to the 'metaconsensus' over shared dimensions of policy evaluation that ensures single-peakedness.However, the apparent attenuation of these uniformity requirements does not really reconcile D-models with pluralism.In fact, the problem with D-models does not lie as much in the degree of uniformity that they require in order to ensure integrity as in the way in which such uniformity, however mitigated, is supposed to be promoted by deliberation.
As we have seen, in fact, D-models explain the integrity-promoting role of deliberation by claiming that deliberation creates common patterns of evaluation, opinion and principles that guarantee a sufficient degree of convergence in how people cast their votes.This represents a serious threat to pluralism, because pluralism concerns above all exactly such a variety of patterns of evaluation, opinions and principles.
Most importantly, this also holds for the varieties of the D-model that rely on 'metaconsensus'-the representation of political issues along the same conceptual dimensions, which guarantees single-peakedness.As we mentioned, metaconsensus is compatible with different preference orderings of the alternatives on the political agenda and different ways in which people cast their vote; however, it requires that the members of a polity conceptualize the issues on the political agenda in exactly the same way.This substantially curbs pluralism by making everyone look at the world through the same conceptual lenses (Ottonelli & Porello, 2013).Again, it can be argued that pluralism does not concern so much how people rank policy alternatives but most importantly consists in people's having different worldviews-different ways to conceptualize their social world and represent different policies, their impact and their relations to underlying values and principles.For instance, in our example of a community divided on immigration policies, the reduction of all their different stances to the dimension exclusion/inclusion significantly shrinks the variety of ways in which the difficult political choices involved can be conceptualised, ruling out other possible dimensions of the debate, such as the relation to local and global economic justice, or the degree in which personal freedom and free movement are achieved.
In sum, with regard to pluralism, the problem with the D-models of how integrity consensus can be built-including the apparently less demanding variety based on 'metaconsensus'is not so much that they produce uniformity in voting patterns, but that they promise to achieve such an outcome by narrowing the cognitive and evaluative landscape of the members of the polity.
Let us now turn to a second problem with the way D-models promise to rescue the integrity-building role of consensus.In D-models, integrity consensus, and the consistency of collective decisions that such consensus makes possible, are generated as side effects of deliberation, and therefore cannot be seen as the intentional product of participants' political agency.In this respect, deliberative accounts of consensus do not rescue democratic participation from the main shortcomings of Rikerian models: atomism, by which democratic decisionmaking is conceived as the product of separate decisions at the individual level, and unintentionality, by which the outcomes of the democratic process of decision-making are produced as unintended results of the participants' involvement.
These shortcomings appear evident if we look at how deliberation is assumed to produce consensus in D-models.As previously mentioned, according to Dmodels the convergence of preference orderings, or their alignment along single-peaked patterns, is achieved because deliberation acts as a 'filter', by discarding those preference orderings that cannot be publicly justified to others (Dryzek, 1990;Dryzek & List, 2003, quoting Elster, 1998a;Goodin, 1992).
However, this filtering is produced as a side effect of the participants' engagement in discursive processes in which they aim at providing reasons and making claims that others can also share.The final vote they take on the issues on the political agenda is fully determined by this process, rather than by their conscious attempt to steer collective decision-making in a way that instantiates integrity and consistency.In deliberative theories of the democratic process, consensus-whenever it emerges-is a by-product of deliberation (Fuerstein, 2014).
It might be objected that this is not true of all accounts of deliberation.In Habermasian accounts, for example, participants do not simply aim at exchanging reasons that can be shared by others, but actively aim at reaching (rational) consensus because such a consensus is an essential presupposition of their discursive practices.However, this is not enough to make the process intentional in the sense relevant for the present discussion.In fact, if we want to obviate the faults of Rikerian models of the democratic process and fully account for citizens as engaged in a process of collective decision-making, we need to conceive of citizens as pursuing consensus in view of its importance for achieving meaningful decisions.That is, consensus must be pursued in its integrity-building role, as a guarantee of collective integrity and democratic self-determination.In D-models, instead, participants are at best interested in consensus simply as an ideal endpoint of their rational deliberation.When such consensus is reached, and therefore the choices produced through the democratic process display the degree of consistency and integrity necessary for them to be meaningful, this further outcome happens without being actively sought by the participants in deliberation.In this sense, the reaching of consensus-in its integrity-building role-can be described as unintentional.
It might be objected that this discussion assumes too narrow a view of what it takes for integrity consensus to be intentional.After all, in D-models participants in deliberation know that integrity consensus follows, as a natural by-product, from their deliberative activities, even if they do not directly aim at it.Why should not this count this as an intentional result?In reply to this possible objection, compare this to the imagined procedure that we discussed in the first Section of the paper, in which people submit their votes to a commission that subsequently mingles them into a series of policies that display integrity and consistency, and therefore a meaningful and unified line of action.Also, in this case the participants know that a meaningful result will ensue from their participation and take an active part in producing such result by submitting their vote; however, this is not enough to make them the intentional authors of the meaningfulness and unity of such line of action.In the same way, participants in a deliberative process that does not aim-among other thingsat producing integrity consensus cannot be said to be the intentional authors of the unity of such line of action.
This feature of D-models pairs with their atomism.In D-models a participant's stance on the decision to be made after deliberation and the way she votes is independent of how other people vote.It might be objected that this charge of atomism assumes that D-models display only a form of 'weak dialogicality' (McMahon, 2000).In 'weakly dialogical' models, people exchange reasons in public but then issue separate individual judgments about the decision to be made.But this is not the only way we can conceptualize deliberation in Dmodels.In 'strongly dialogical' models not only the deliberative stage in which individual judgments are formed and transformed, but also the very moment at which people vote or otherwise express their conclusive views about what should be done are collective and reciprocal-each participant expresses her judgment while knowing that others will do the same.In William Rehg's (1991) words, I can be rationally convinced of the worthiness of a norm only if I suppose that others are rationally convinced, which in turn depends on their supposing that I am rationally convinced.If this is not to be a vicious circle, then rational conviction must be something that we arrive at together (p.44). 8 Appealing to strong dialogicality, however, is not enough to dispel the charge of atomism in the relevant sense for our discussion.If citizens must act as decision-makers, and therefore must directly aim to achieve the consensus needed to produce meaningful decisions, then the decision-making process must be non-atomistic in the sense that it must allow them to mindfully coordinate their actions.An account of decision-making is genuinely nonatomistic in the sense specified, then, when in making such an effort of coordination each voter takes other people's votes as an independent reason for deciding how to vote.In D-models, instead, voters do not take into account the way other people vote or express their political preferences as an independent reason for action, but only as evidence of other people's judgments and only to the extent that they are supposed to head in the right direction.This is required by the deliberative nature of the process.When I am deliberating with others, to take their vote as a basis for my voting, I need to assume it reflects good or valid reasons; this means that in casting my vote what I actually take into account is not my fellow citizens' votes, but the reasons behind those votes.In this sense, although in deliberation people influence each other by arguing and exchanging reasons, the deliberative model of decision-making is no less atomistic than Riker's model.
In sum, D-models of democratic decision-making, notwithstanding that they offer a richer account of political agency than Rikerian and minimalist models, still represent democratic decision-making as atomistic in the sense that votes 8 Quoted by McMahon (2000, p. 521).
are fully independent of each other, and are unintentional in the sense that integrity consensus is not aimed at by participants but is instead a by-product of their deliberative exchanges.Therefore, D-models cannot rescue the fundamental notion that, in a democracy, citizens must act as decision-makers.

The Missing Element: Political Prudence
If we want to fully rescue the integrity consensus that underlies selfdetermination, we need a different account of consensus-building from the one provided by D-models.Such an account must explain consensus-building as the intentional product of the agency of participants in democratic decision-making.
In this section, I suggest that what is needed is to add a further dimension of political agency to the discursive rationality instantiated in D-models.I propose to call this missing element political prudence (PP).
By political prudence I mean the general capability of devising the opportunity and rationale for engaging in the different political activities allowed by democratic political rights.Examples of the exercise of political prudence are the decision whether to participate in a general strike against the government; whether to desert the ballot box as a form of protest; whether to engage in deliberation and pursue one's political vision by trying to convince the general public of the rightness of one's views, or whether to resort to compromise and accept a middle ground with one's political opponents; and more generally all the instances in which it behoves citizens to judge how best to exercise their political rights.
Many deliberative theories of democracy, and notably all those that allow for compromises and deviations from pure deliberation, seem to rely on the capacity for political prudence: they assume that citizens may judge when the moment has come to relinquish the purely deliberative mode and engage in negotiations and compromises (Bohman, 1996;Habermas, 1996;Mansbridge, 1999;Mansbridge & Martin, 2013), or, conversely, when they should try to push for a given political issue to be debated in the deliberative mode.However, in deliberative accounts, such a capacity is seldom theorized as a distinct component of democratic agency.
Among the tasks of political prudence, I want to suggest, is also the assessment of the reasons, circumstances and means for achieving integrity consensus.In a model of how integrity consensus is achieved through the resort to political prudence (from now on, PP-model), citizens engage in deliberation and other political activities enabled by their democratic rights while at the same time keeping an eye on the overall effects of their activities on collective decisionmaking.This may guide their choices and actions in at least two ways.
First, it can encourage citizens to think about the right timing for pressing an issue for public deliberation, in consideration, for example, of the destabilizing effects that it may have on the political system and existing majorities, or of the proximity of electoral junctures that would make such effects dangerous for political stability.Deliberation does not flow by itself, and the order and mode in which specific issues come to the fore in the public sphere is decided by political parties, movements, associations and individual citizens.These decisions have an enormous impact on the capacity of the polity to preserve integrity and unity of agency.In fact, in his attack on the populist ideal, William Riker (1982) provided various examples of how throwing a divisive issue on the agenda, or framing the underlying principles in an inflammatory way, can create cycles, paradoxes and inconsistency.9There is no reason to believe that this knowledge can only be used (as Riker suggests) by malevolent political actors who aim at destabilizing existing majorities.It can be used, and is often used, by the citizens of a democratic polity to maintain consensus and integrity when they are needed.
Second, and very importantly, citizens can also decide to vote 'insincerely' or 'strategically'-that is, in a way that does not reflect the way they would order the policy options if they had merely to decide on the basis of their own conscience or preferences.Strategic voting is often associated with manipulation, by which voters falsely represent their preferences to produce an outcome that favours their preferred options.However, strategic voting can also be used for other purposes than self-advantage.It can be instrumental to the pursuit of the consensus that is necessary to make meaningful collective decisions, when citizens realize that 'sincere' voting would create instability or inconsistent results at the collective level, thus compromising collective integrity and the meaningfulness of democratic choices.
Upon reflection, we can see that the PP-model offers a familiar description of powers and activities already instantiated in our democratic polities and protected by democratic rights.The introduction of the notion of political prudence is not a call for dramatic changes in our institutional practices; rather, it is meant to account for the importance of the ample leeway democratic rights offer citizens in terms of intentionally steering collective decision-making through their political activity and engagement.
If we conceive the building of integrity consensus as the product of intentional actions and decisions by the citizens of a democratic polity, then we also recognize that it is up to them whether and to what extent they want to pursue such an aim.The premise of our argument so far is that the members of a democratic polity have an interest in acting as decision-makers, rather than mere gears in a mechanical device that produces random selections among policy alternatives.This interest must translate into the possibility of intentionally taking action to ensure the consistency of democratic decisions and the integrity of the democratic polity.However, this is an interest that citizens always need to balance against other considerations, including the dictates of their conscience or the need to keep dissensus alive.In other terms, conceiving of consensus-building as the product of the intentional actions of citizens, as the PP-model does, also leaves citizens the room to decide how to balance and negotiate the unity of agency of the polity with other interests that pull in the opposite direction.So, for example, the PP-model recognizes that citizens can decide when restoring collective integrity would require too high a price to pay in terms of justice, so that they should pursue their view of justice by arguing for it and voting accordingly.Moreover, this model recognises that citizens have the discretion to decide when it would be a good thing, after all, if the existing consensus and integrity were broken. 10 The PP-model of how consensus is built in democratic decision-making is obviously-almost trivially-immune from the flaws that affect the models we have previously discussed.First, in the PP-model, democratic decision-making is not atomistic, since citizens, in casting their votes, take into account the consequences their votes will produce once aggregated with the votes of others.This means that the way other people will vote counts as an independent reason for deciding how to vote, rather than mere evidence for what counts as a good decision, as is the case with D-models.Convergence and consensus are created as the result of an intentional pursuit of coordination by each and every citizen.
Second, in the PP-model the seeking and achievement of integrity consensus can be intentional.Citizens cast their vote and participate in public deliberation having in view the overall results of their actions in terms of the integrity and consistency of their political community as a self-directing polity.Integritybuilding consensus, then, will not be achieved mechanically, or as a by-product of democratic activities that are directed to other goals (be that deliberating, compromising, 11 or other forms of political interaction), but it will be the product of the exercise of intentional actions by citizens.
10 This is a main difference from those models of democratic decision-making that are based on a pre-commitment to abide by the decisions of the majority (Gilbert, 2006), as well as from those models that make the acceptance of the position of the majority conditional on the quality of the deliberative process (Moore & O'Doherty, 2014). 11Henry Richardson (2002) has addressed the problem of agreement within a democratic polity by appealing to 'deep compromises.'These compromises are motivated by the joint intention to achieve convergence in the light of the realization that each party has its own separate and legitimate aims and goals.Richardson's compromises are not atomistic, because they require coordination.However, they are not a satisfying solution to the problem we are considering here, because like other kinds of compromise they cannot guarantee integrity at the collective level (Besson, 2005); and, whenever they happen to produce such integrity, integrity is not the object of the intentional action of the participants in the process.
Third, the PP-model can find room not only for pluralism, but also for open dissent.This is true in two senses.First, in the PP model the substantive consensus that guarantees collective consistency and integrity does not need to be reached through a transformation and reductio ad unum of people's underlying views, values and principles.Second, as we mentioned political prudence allows for a careful and calculated balancing of the need to reach collective integrity and meaningful decisions with the contrasting interest in expressing one's differences with, or open dissent from, the views of the majority.The control that political prudence leaves to the democratic public allows minorities to consider acceding to the consensus only to the extent strictly required to ensure integrity at the collective level.

Coordination, Motivations, and the Risk of Manipulation
The PP-model of democratic decision-making, whatever its merits, may not look feasible.The exercise of political prudence, even if limited to the achievement of collective integrity, requires a high degree of coordination and the ability to forecast the overall results of one's choices once they interact with other people's actions through democratic procedures.In response to this concern, two important points need to be stressed.
First, the PP model is just a normative and regulative model.It does not ensure citizens will always be able to meet its standards.What such a model does is to make sense of our intuition that in democratic government an essential connection exists between individual agency and collective self-determination and that this connection requires that citizens may intentionally contribute to steering the course of collective decision-making.
Second, the objection overstates the coordination problems involved in the PPmodel, and overlooks some of the powerful instruments citizens can count on in devising their deliberative and voting strategies.A complete account of how coordination problems relate to how collective integrity can be overcome exceeds the limits of present discussion.However, I would like to mention two obvious tools we are currently using, in our democracies, to coordinate our actions with those of our fellow citizens and forecast their deliberative and voting behaviours.The first tool is parties and other political organisations.Parties themselves are powerful means for coordination (Budge, 2006;Dewan & Myatt, 2007;Sartori, 2005), but their internal life and relation to the rest of the polity are also a fundamental source of information about the intentions and aims of their affiliates.A second tool is opinion polls.These are very controversial elements of our current practices, especially when they are uncritically taken as plain expressions of 'public opinion' (Bourdieu, 1973), or when they are subject to manipulation by party leaders to fake consensus (Herbst, 1995;Jacobs & Shapiro, 1995).However, they play an essential informative function for the democratic public.Polls change, and we may conjecture that this happens because citizens reorient their intentions in light of data acquired though previous polls (Fey, 1997).
Even if concerns about the actual capacity for coordination are dispelled, doubts can be raised about whether the democratic public can ever develop the right motivations for engaging in such complex practices.This motivational challenge can be addressed by recalling the response that some deliberativists have offered to a parallel objection that was directed against the idea that citizens would be motivated to engage in deliberation.They have rightly pointed out that, in spite of being presented as 'realistic', the alternative models that depict democracy as a game for the competition of interests are unable to explain the actual practice of democracy and fail to offer a credible account of why citizens should be motivated to participate in such a dismal game.In the same vein, it may be argued that political prudence, far from being unrealistic, is what makes sense of the actual practice of democracy.Citizens would not be as motivated to participate if they knew they could not exercise any direct control on the meaningfulness of democratic decisions, or if they knew their participation would be likely to issue in inconsistent and paradoxical decisions.So, they have an inherent interest in steering the decision-making process in such a way that its final results make sense and display integrity, and that they do so as a result of citizens' actions and choices.
A final, important concern may be raised about the fairness and nonmanipulability of citizens' voting as guided by political prudence.The PPmodel may seem to hand too much power and discretion to participants, and we may worry about possible abuses and path-dependence.Minorities can be induced to consensus by the threat of breaking up collective integrity. 12Even worse, when a sufficient degree of consensus is lacking and needs to be built through the conversion of some participants to voting patterns that will guarantee consistency and stability, which groups get saddled with that task may depend on arbitrary circumstances and may unfairly burden disempowered minorities.These effects are supposedly not possible in D-models, in which the whole process of consensus-building is guided by public argument and therefore is not subject to arbitrariness.However, three important facts need to be recalled about the PP-model.First, the danger of manipulation can be tamed by a high degree of publicity and transparency in the institutional tools used for coordinating votes.For example, measures can be taken to ensure the independence, openness and non-manipulability of the data provided by opinion polls, and the internal life of parties can be made more transparent to the general public.Second, in the PP model the exercise of political prudence is not supposed to replace deliberation, but to complement it.Moreover, the PP-model extends deliberation not only to the merits of different proposals but to the very process of elaborating collective strategies for arriving at integrity-building consensus.Arguments for voting in a way that does not reflect one's judgment on what the correct decision would be, arguments for withdrawing one's consensus and openly expressing dissent, and arguments for doing so even when this threatens collective integrity, should be made public and be part of the deliberative process. 13So, if deliberation has any power to resist manipulation and arbitrary path-dependence, it could be counted upon also in the PP-model.

Conclusion
Integrity consensus is an essential element of the democratic ideal because it guarantees that citizens can act as decision-makers rather than mere givers of inputs into random and potentially inconsistent decision procedures.However, if our valuing integrity consensus is grounded in our interest in citizens' selfdetermination and agency, then such consensus must be reached in ways that allow for self-determination and agency.I have argued that purely deliberative models of how integrity consensus is to be achieved cannot fulfil this requirement.If we want to account for the integrity-building role of consensus as a guarantee of democratic self-direction, along with the exercise of deliberative rationality we need to explicitly theorise a dimension of political agency-political prudence-that consists in exercising intentional control on the overall effects at the collective level of one's actions in the course of collective decision-making.